Files
bid-doctor/server/tests/test_check_flow.py
YANG JIANKUAN df3ebbb6a1 feat(check): 偏离栏直读、响应错位判断;复查复用和硬预算(C2)
- deviation.py:模型判已响应的条目,证据所在表格行的偏离格写着「偏离 / 部分满足」等
  就改判负偏离;照抄要求格、响应格写的是别的内容的改判「响应可能对不上」
- reuse.py:按「提示词 + 要求 + 候选句原文」算缓存键,存模型原始结论;复查时内容没变、
  证据找得回的条目直接沿用,不调模型
- 每次检查送给模型的字数有上限,超出的条目列为请人工核对,写进覆盖范围
- 开发集 17 份:召回 61% → 73%,负偏离未说明 64% → 90%,规矩投标文件问题数不变

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-04 17:47:12 +08:00

363 lines
20 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

"""查标第 1 步(m3-plan §5):上传投标文件 → 检查 → 处理问题 → 复查后状态沿用;以及检查的计费(红线 9)。
假模型按字面重合判断有没有响应,结论没有参考价值;这里验证的是链路、证据、状态沿用和计费。
"""
from __future__ import annotations
from typing import Any
from uuid import UUID
import httpx
from sqlalchemy import text
from tests.conftest import FIXTURES, User, balance, project_with_tender, register, run_parse, upload
async def upload_bid(u: User, project_id: str, fixture: str, replaces: str | None = None) -> dict[str, Any]:
done = await upload(u, fixture, (FIXTURES / fixture).read_bytes())
assert done.status_code == 200, done.text
body: dict[str, Any] = {"upload_id": done.json()["id"], "role": "bid_volume", "volume_kind": "technical"}
if replaces:
body["replaces_document_id"] = replaces
r = await u.post(f"/projects/{project_id}/documents", json=body)
assert r.status_code == 201, r.text
return r.json()
async def run_check(u: User, project_id: str, doc_ids: list[str], expected: int = 0) -> dict[str, Any]:
"""发起检查,并在测试进程里直接跑流水线(不经过 worker)。"""
from bid_doctor.check import runner
r = await u.post(f"/projects/{project_id}/check-runs", json={"document_ids": doc_ids, "expected_charge": expected})
assert r.status_code == 201, r.text
await runner.run(UUID(u.tenant_id), UUID(r.json()["id"]))
return (await u.get(f"/check-runs/{r.json()['id']}")).json()
async def checked_project(u: User) -> tuple[str, dict[str, Any], dict[str, Any]]:
project_id, tender_id = await project_with_tender(u)
assert (await run_parse(u, tender_id))["status"] == "succeeded"
bid = await upload_bid(u, project_id, "bid-mini-v1.pdf")
run = await run_check(u, project_id, [bid["id"]])
assert run["status"] == "succeeded", run
return project_id, bid, run
async def findings(u: User, run_id: str) -> list[dict[str, Any]]:
return (await u.get(f"/check-runs/{run_id}/findings")).json()["items"]
async def use_pro_plan(u: User) -> None:
from bid_doctor.core.db import TenantContext, tenant_tx
ctx = TenantContext(tenant_id=UUID(u.tenant_id), user_id=UUID(str(u.me["user"]["id"])))
async with tenant_tx(ctx) as s:
await s.execute(text("UPDATE subscription SET plan_code = 'pro' WHERE tenant_id = :t"), {"t": u.tenant_id})
async def test_check_flow(user: User) -> None:
project_id, tender_id = await project_with_tender(user)
await run_parse(user, tender_id)
preview = (await user.get(f"/projects/{project_id}/check-preview")).json()
assert preview["blockers"] == ["no_bid_documents"] and not preview["can_start"]
bid = await upload_bid(user, project_id, "bid-mini-v1.pdf")
assert bid["label"] == "技术分册" and bid["volume_kind"] == "technical" and bid["is_latest_version"]
preview = (await user.get(f"/projects/{project_id}/check-preview")).json()
assert preview["can_start"] and preview["billing"] == "free_plan" and preview["charge"] == 0
assert preview["checklist_confirmed"] is False and [d["id"] for d in preview["bid_documents"]] == [bid["id"]]
run = await run_check(user, project_id, [bid["id"]])
assert run["status"] == "succeeded" and run["is_latest"] and run["checklist_confirmed"] is False
assert run["counts"] == {"fatal": 1, "important": 0, "hint": 0, "open_fatal": 1, "resolved_this_run": 0, "new": 1}
assert "清单未确认,检查结果以当前清单为准" in run["coverage"]["not_checked"]
assert run["coverage"]["items"]["checked"] == 3
(f,) = await findings(user, run["id"])
assert f["type"] == "unresponded" and f["severity"] == "fatal" and f["presence"] == "new"
assert "BFD" in f["title"] and f["detector"] == "ai" and f["checklist_item"]["importance"] == "substantive"
assert f["tender_citations"][0]["quote"].startswith("★交换机") and f["bid_citations"] == []
assert f["search_scope"]["documents"] == ["技术分册"]
assert "建议" not in f["explanation"] # 只写事实和出处,不写整改建议
# 「已响应」也留结论和出处(C9)
results = (await user.get(f"/check-runs/{run['id']}/item-results")).json()["items"]
responded = [r for r in results if r["verdict"] == "responded"]
assert len(results) == 3 and len(responded) == 2
for r in responded:
(c,) = r["bid_citations"]
assert r["evidence_status"] == "verified" and c["document_label"] == "技术分册" and c["quads"]
project = (await user.get(f"/projects/{project_id}")).json()
assert project["status"] == "checked" and project["readiness"] == {
"status": "checked",
"open_fatal_count": 1,
"ready": False,
}
notes = (await user.get("/notifications")).json()["items"]
assert notes[0]["kind"] == "check_succeeded"
async def test_resolve_and_reopen(user: User) -> None:
project_id, _, run = await checked_project(user)
(f,) = await findings(user, run["id"])
r = await user.post(f"/findings/{f['id']}/resolve", json={"status": "confirmed_ok"})
assert r.status_code == 422 and r.json()["code"] == "note_required" # 致命问题确认无误也要写说明(U4)
r = await user.post(f"/findings/{f['id']}/resolve", json={"status": "ignored", "note": " "})
assert r.status_code == 422 and r.json()["code"] == "note_required"
r = await user.post(f"/findings/{f['id']}/resolve", json={"status": "confirmed_ok", "note": "附件 5 第 3 页有说明"})
assert r.status_code == 200, r.text
body = r.json()
assert body["finding"]["status"] == "confirmed_ok" and body["finding"]["resolved_by"]["name"] == "张工"
assert body["project"] == {"status": "checked", "open_fatal_count": 0, "ready": True}
from bid_doctor.core.db import TenantContext, tenant_tx
ctx = TenantContext(tenant_id=UUID(user.tenant_id), user_id=UUID(str(user.me["user"]["id"])))
async with tenant_tx(ctx) as s:
fb = (await s.execute(text("SELECT object_type, kind, source, note FROM feedback"))).all()
assert [tuple(x) for x in fb] == [("finding", "false_positive", "fatal_confirm", "附件 5 第 3 页有说明")]
r = await user.post(f"/findings/{f['id']}/reopen")
assert r.json()["finding"]["status"] == "pending" and r.json()["project"]["open_fatal_count"] == 1
r = await user.post(
f"/check-runs/{run['id']}/findings/bulk-resolve",
json={"finding_ids": [f["id"]], "status": "ignored", "note": "x"},
)
assert r.status_code == 422 and r.json()["code"] == "bulk_only_hint" and r.json()["finding_ids"] == [f["id"]]
assert (await user.get(f"/projects/{project_id}")).json()["readiness"]["open_fatal_count"] == 1
async def test_recheck_carries_status_and_resolves(user: User) -> None:
project_id, bid, first = await checked_project(user)
(f,) = await findings(user, first["id"])
await user.post(f"/findings/{f['id']}/resolve", json={"status": "ignored", "note": "BFD 写在附件里"})
# 同一份投标文件再查:问题还在,沿用上次的处理状态
second = await run_check(user, project_id, [bid["id"]])
(g,) = await findings(user, second["id"])
assert g["presence"] == "carried" and g["status"] == "ignored" and g["carried_status"] == "ignored"
assert g["resolution_note"] == "BFD 写在附件里" and second["counts"]["open_fatal"] == 0
# 历史批次只读
r = await user.post(f"/findings/{f['id']}/reopen")
assert r.status_code == 409 and r.json()["code"] == "run_superseded"
# 上次「已修改」又查出来:回到待处理
await user.post(f"/findings/{g['id']}/resolve", json={"status": "modified"})
third = await run_check(user, project_id, [bid["id"]])
(h,) = await findings(user, third["id"])
assert h["presence"] == "carried" and h["status"] == "pending" and h["carried_status"] == "modified"
# 换上补了 BFD 的新版本:问题本次已解决
v2 = await upload_bid(user, project_id, "bid-mini-v2.pdf", replaces=bid["id"])
assert v2["label"] == "技术分册 v2" and v2["lineage_id"] == bid["lineage_id"]
assert not (await user.get(f"/documents/{bid['id']}")).json()["is_latest_version"]
preview = (await user.get(f"/projects/{project_id}/check-preview")).json()
assert [d["id"] for d in preview["bid_documents"]] == [v2["id"]]
fourth = await run_check(user, project_id, [v2["id"]])
(k,) = await findings(user, fourth["id"])
assert k["presence"] == "resolved" and fourth["counts"]["resolved_this_run"] == 1
assert fourth["counts"]["fatal"] == 0 and fourth["counts"]["open_fatal"] == 0
r = await user.post(f"/findings/{k['id']}/resolve", json={"status": "ignored", "note": "x"})
assert r.status_code == 409 and r.json()["code"] == "finding_resolved_this_run"
project = (await user.get(f"/projects/{project_id}")).json()
assert project["readiness"]["ready"] is True and project["latest_check_run_id"] == fourth["id"]
async def test_pro_plan_charges_then_waives_recheck(app_client: httpx.AsyncClient) -> None:
u = await register(app_client)
await use_pro_plan(u)
project_id, tender_id = await project_with_tender(u)
await run_parse(u, tender_id)
bid = await upload_bid(u, project_id, "bid-mini-v1.pdf")
assert await balance(u) == 2
preview = (await u.get(f"/projects/{project_id}/check-preview")).json()
assert preview["billing"] == "charged" and preview["charge"] == 1 and preview["waive_until"] is None
r = await u.post(f"/projects/{project_id}/check-runs", json={"document_ids": [bid["id"]], "expected_charge": 0})
assert r.status_code == 409 and r.json()["code"] == "billing_changed" and r.json()["preview"]["charge"] == 1
assert await balance(u) == 2 # 扣点说明和实际不一致:拒绝,不扣点(API4)
first = await run_check(u, project_id, [bid["id"]], expected=1)
assert first["billing"] == "charged" and await balance(u) == 1
preview = (await u.get(f"/projects/{project_id}/check-preview")).json()
assert preview["billing"] == "waived_recheck" and preview["charge"] == 0 and preview["waive_until"]
second = await run_check(u, project_id, [bid["id"]], expected=0)
assert second["billing"] == "waived_recheck" and await balance(u) == 1
from bid_doctor.core.db import TenantContext, tenant_tx
ctx = TenantContext(tenant_id=UUID(u.tenant_id), user_id=UUID(str(u.me["user"]["id"])))
async with tenant_tx(ctx) as s:
rows = (
await s.execute(
text("SELECT kind, amount, reason FROM quota_entry WHERE ref_type = 'check_run' ORDER BY created_at")
)
).all()
assert [tuple(x) for x in rows] == [("hold", -1, None), ("settle", 0, None), ("waive", 0, "recheck_24h")]
async def test_unreadable_bid_fails_and_refunds(app_client: httpx.AsyncClient) -> None:
"""投标文件读不出文字(扫描件):按输入问题失败,退回额度,不生成问题。"""
u = await register(app_client)
await use_pro_plan(u)
project_id, tender_id = await project_with_tender(u)
await run_parse(u, tender_id)
bid = await upload_bid(u, project_id, "blank.pdf")
run = await run_check(u, project_id, [bid["id"]], expected=1)
assert run["status"] == "failed" and run["error"]["code"] == "bid_text_missing" and run["error"]["refunded"]
assert await balance(u) == 2 and await findings(u, run["id"]) == []
project = (await u.get(f"/projects/{project_id}")).json()
assert project["status"] == "drafting" and project["latest_check_run_id"] is None
async def test_duplicate_check_is_rejected(user: User) -> None:
project_id, tender_id = await project_with_tender(user)
await run_parse(user, tender_id)
bid = await upload_bid(user, project_id, "bid-mini-v1.pdf")
body = {"document_ids": [bid["id"]], "expected_charge": 0}
assert (await user.post(f"/projects/{project_id}/check-runs", json=body)).status_code == 201
r = await user.post(f"/projects/{project_id}/check-runs", json=body)
assert r.status_code == 409 and r.json()["code"] == "job_already_running" and r.json()["job_id"]
preview = (await user.get(f"/projects/{project_id}/check-preview")).json()
assert "job_already_running" in preview["blockers"]
async def test_platform_encrypted_bid_is_rejected(user: User) -> None:
done = await upload(user, "投标文件.kedt", b"PK\x03\x04not-really-a-zip")
assert done.status_code == 422 and done.json()["code"] == "bid_file_platform_encrypted"
assert "苏采云" in done.json()["detail"]
async def test_other_tenant_cannot_see_check(app_client: httpx.AsyncClient) -> None:
a = await register(app_client, "甲公司")
b = await register(app_client, "乙公司")
project_id, _, run = await checked_project(a)
(f,) = await findings(a, run["id"])
for path in (
f"/projects/{project_id}/check-preview",
f"/projects/{project_id}/check-runs",
f"/check-runs/{run['id']}",
f"/check-runs/{run['id']}/item-results",
):
r = await b.get(path)
assert r.status_code == 404, (path, r.status_code, r.text)
assert (await b.get(f"/check-runs/{run['id']}/findings")).status_code == 404
assert (await b.post(f"/findings/{f['id']}/resolve", json={"status": "modified"})).status_code == 404
assert (await b.post(f"/findings/{f['id']}/reopen")).status_code == 404
assert (await b.post(f"/check-runs/{run['id']}/cancel")).status_code == 404
async def test_response_modes_are_stored_editable_and_used(user: User) -> None:
"""响应方式(C12):解析时存进清单;用户改成「默认接受」后,查标对没复述的一般要求不再报问题。"""
project_id, tender_id = await project_with_tender(user)
await run_parse(user, tender_id)
modes = {m["category"]: m for m in (await user.get(f"/projects/{project_id}/checklist")).json()["response_modes"]}
assert set(modes) == {"technical", "commercial"}
assert modes["technical"]["mode"] == "explicit" and modes["technical"]["status"] == "unknown"
r = await user.put(f"/projects/{project_id}/checklist/response-modes/technical", json={"mode": "implicit"})
assert r.status_code == 200, r.text
tech = next(m for m in r.json()["response_modes"] if m["category"] == "technical")
assert tech["origin"] == "human" and tech["mode"] == "implicit" and tech["set_by"]["name"] == "张工"
bid = await upload_bid(user, project_id, "bid-mini-v1.pdf")
run = await run_check(user, project_id, [bid["id"]])
# 漏掉的 BFD 是 ★ 条款:默认接受下不报致命,改为提示级「请人工核对」
(f,) = await findings(user, run["id"])
assert f["type"] == "manual_review" and f["severity"] == "hint"
assert run["counts"]["fatal"] == 0
async def test_rule_findings_are_written_with_citations_and_carried(user: User) -> None:
"""规则检查(报价形式、形式要件):问题挂规则编号和投标原文,复查时按 match_key 沿用处理状态。"""
project_id, tender_id = await project_with_tender(user)
await run_parse(user, tender_id)
bid = await upload_bid(user, project_id, "bid-mini-forms.pdf")
run = await run_check(user, project_id, [bid["id"]])
assert run["status"] == "succeeded", run
rule = {f["rule"]["code"]: f for f in await findings(user, run["id"]) if f["detector"] == "rule"}
assert set(rule) == {
"PRICE.CAPITAL_MISMATCH",
"PRICE.MULTIPLE_QUOTES",
"FORM.BLANK_FIELD",
"FORM.CONDITION_ADDED",
}
cap = rule["PRICE.CAPITAL_MISMATCH"]
assert cap["type"] == "price_form" and cap["severity"] == "important" and cap["checklist_item"] is None
assert cap["rule"]["description"] == "大写金额和小写金额核对"
(c,) = cap["bid_citations"] # 同一页的几句合成一条出处,各句分别画框
assert "壹拾贰万捌仟元整" in c["quote"] and "1,280,000.00" in c["quote"] and len(c["quads"]) >= 2
assert c["document_label"] == "技术分册"
# 「让利后总价」和投标总价不同:多半另报了一个价,但不是「备选方案」那种明确的两个报价,按重要
assert rule["PRICE.MULTIPLE_QUOTES"]["severity"] == "important" and rule["FORM.BLANK_FIELD"]["severity"] == "hint"
assert any("报价形式" in x for x in run["coverage"]["checked"])
assert run["coverage"]["rule_set_version"] == "2026.11-r2"
await user.post(f"/findings/{cap['id']}/resolve", json={"status": "ignored", "note": "大写栏已手工更正"})
again = await run_check(user, project_id, [bid["id"]])
cap2 = next(
f for f in await findings(user, again["id"]) if (f["rule"] or {}).get("code") == "PRICE.CAPITAL_MISMATCH"
)
assert cap2["presence"] == "carried" and cap2["status"] == "ignored"
async def test_manual_checks_are_kept_per_project(user: User) -> None:
"""需人工核对:来自清单(按主题合并),勾选按项目存、写勾选人和时间,复查后还在(决策 P2)。"""
project_id, _, _ = await checked_project(user)
for title in ("须用CA数字证书加密投标文件", "开标后30分钟内在线解密投标文件", "本次采购不收取投标保证金"):
r = await user.post(
f"/projects/{project_id}/checklist/items", json={"categories": ["schedule"], "title": title}
)
assert r.status_code == 201, r.text
items = (await user.get(f"/projects/{project_id}/manual-checks")).json()["items"]
assert [(i["check_key"], i["checked"]) for i in items] == [
("rule:MANUAL.DECRYPT", False),
("rule:MANUAL.SIGN_ENCRYPT", False),
]
assert items[1]["requirement_text"] == "须用CA数字证书加密投标文件"
r = await user.put(
f"/projects/{project_id}/manual-checks",
json={"check_key": "rule:MANUAL.DECRYPT", "checked": True, "note": "已演练"},
)
assert r.status_code == 200, r.text
m = r.json()
assert m["checked"] and m["note"] == "已演练" and m["checked_by"]["name"] and m["checked_at"]
bad = await user.put(
f"/projects/{project_id}/manual-checks", json={"check_key": "rule:MANUAL.NOPE", "checked": True}
)
assert bad.status_code == 422
bid_id = (await user.get(f"/projects/{project_id}/check-preview")).json()["bid_documents"][0]["id"]
again = await run_check(user, project_id, [bid_id])
assert any("列入「需人工核对」2 条" in x for x in again["coverage"]["not_checked"])
kept = (await user.get(f"/projects/{project_id}/manual-checks")).json()["items"]
assert [i["checked"] for i in kept] == [True, False]
async def test_recheck_reuses_unchanged_model_results(user: User) -> None:
"""复查复用(C2):同一份投标文件再查,送给模型的内容没变,模型那一层沿用上次的结论,不再调模型。"""
from bid_doctor.ai.models import AiCall
from bid_doctor.core.db import TenantContext, tenant_tx
project_id, bid, first = await checked_project(user)
second = await run_check(user, project_id, [bid["id"]])
assert second["counts"]["fatal"] == first["counts"]["fatal"]
reused = [x for x in second["coverage"]["checked"] if "沿用上次的结论" in x]
assert reused, second["coverage"]["checked"]
ctx = TenantContext(tenant_id=UUID(user.tenant_id), user_id=UUID(str(user.me["user"]["id"])))
async with tenant_tx(ctx) as s:
from sqlalchemy import func, select
n = (await s.execute(select(func.count()).where(AiCall.job_id == UUID(second["id"])))).scalar_one()
assert n == 0
results = (await user.get(f"/check-runs/{second['id']}/item-results")).json()["items"]
assert all(r["bid_citations"] for r in results if r["verdict"] == "responded") # 沿用的结论证据也找回来了