test(scripts): 36 个 E2E 与 6 个机械校验工具（1548 条断言）+ 一键交付自检
新增：check-api-parity（契约 121 接口 ↔ 控制器，自带「必须恰好 121」自检）、 check-entity-columns（基类要求的列 vs 迁移建表列，F-35 那个坑踩了四次之后上的工具）、 check-manual（手册引用的错误码与提示文案对得上代码）、check-test-profile（test profile 真起一次）、 e2e-prodguard、load-test（四场景、按端点分别统计）、verify-all（一条命令跑完交付自检）、 dev-seed-audit（界面验收用的演示数据），以及 M1~M9 各轮 E2E。 压测抓到一个真缺陷并修掉（F-92）：门禁 IP 限流原来计「所有请求」，于是一个班连同一个 WiFi 时 60 个不同学生同时核验只过了 20 个 → 改为只计失败 + 单独的成功上限，20/60 → 60/60。 度量工具的偏差会伪装成产品缺陷（F-60/F-77/F-97），所以每个脚本都断言「动作自己的结论字段」 （submitted / created / canExport），而不是 HTTP 200 + code 0。 验证：verify-all.py --quick 5/5；--with-e2e 见 docs/09 §8。
This commit is contained in:
@@ -0,0 +1,325 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
"""导出分卷与过期清理 E2E（《03-接口设计》§12.4 的 `splitSizeMb` / 保留 7 天）。
|
||||
|
||||
单独一个脚本（而不是塞进 `e2e-m6-export.py`）的理由：那两个脚本跑的前置不同 ——
|
||||
`e2e-m6-export.py` 要先摆演示数据并**导出后不可撤回**的连带断言；
|
||||
这个脚本只关心"分卷对不对"与"过期之后会怎样"，不需要那一整套前置。
|
||||
|
||||
## 分卷的验收点
|
||||
|
||||
① **每一卷都是完整可解的 ZIP**（不是 `.z01` 那种必须全部到齐的分卷格式）——
|
||||
客户拿到包往往只会双击其中一个，切成 `.z01` 会让对方以为文件坏了；
|
||||
② 卷名带 `-partN`，每一卷都有独立的 sha256；
|
||||
③ 每一卷根目录都带汇总 Excel（只看一卷时也要能拿到名单）。
|
||||
|
||||
## 过期清理的验收点
|
||||
|
||||
① 到期后**文件删掉、行留下** —— "当时确实导出过"是审计要保留的事实，
|
||||
删行之后"这个任务导出过几卷、给谁下载过"就查不出来了；
|
||||
② 下载接口明确回「已被清理、请重新导出」，而不是「文件不存在」（那像 bug）；
|
||||
③ 任务状态是 **4 已过期**，不是 3 失败 —— **过期不是失败**，
|
||||
把正常交付过的包算成失败，会让交付统计这个要给人看的数字失准。
|
||||
|
||||
跑法：python scripts/e2e-m6-export-split.py
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import argparse
|
||||
import importlib.util
|
||||
import json
|
||||
import os
|
||||
import sys
|
||||
import zipfile
|
||||
|
||||
sys.stdout.reconfigure(encoding="utf-8")
|
||||
|
||||
ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
|
||||
PASSED: list[str] = []
|
||||
FAILED: list[tuple[str, str]] = []
|
||||
CREATED_EXPORT_IDS: list[int] = []
|
||||
|
||||
|
||||
def _load(module_name: str, filename: str):
|
||||
path = os.path.join(ROOT, "scripts", filename)
|
||||
spec = importlib.util.spec_from_file_location(module_name, path)
|
||||
mod = importlib.util.module_from_spec(spec)
|
||||
spec.loader.exec_module(mod)
|
||||
return mod
|
||||
|
||||
|
||||
R = _load("e2e_m2_roster", "e2e-m2-roster.py")
|
||||
BASE = "http://localhost:8080"
|
||||
SUMMARY_NAME = "考生线下报名信息导入模板（汇总）.xlsx"
|
||||
|
||||
|
||||
def ok(name: str) -> None:
|
||||
PASSED.append(name)
|
||||
print(f" [OK] {name}")
|
||||
|
||||
|
||||
def fail(name: str, detail: str) -> None:
|
||||
FAILED.append((name, detail))
|
||||
print(f" [FAIL] {name}\n {detail}")
|
||||
|
||||
|
||||
def check(name: str, cond: bool, detail: str = "") -> bool:
|
||||
if cond:
|
||||
ok(name)
|
||||
else:
|
||||
fail(name, detail or "断言不成立")
|
||||
return cond
|
||||
|
||||
|
||||
def sql(stmt: str) -> str:
|
||||
_, out = R.sql(stmt)
|
||||
return out
|
||||
|
||||
|
||||
def body_of(method: str, path: str, token: str | None = None, payload: dict | None = None) -> dict:
|
||||
_, b = R.call(method, f"{BASE}{path}", token, payload)
|
||||
return b
|
||||
|
||||
|
||||
def wait_export(export_id: int, token: str, tries: int = 40) -> dict:
|
||||
"""轮询到终态（2 完成 / 3 失败 / 4 过期）。"""
|
||||
import time
|
||||
d = {}
|
||||
for _ in range(tries):
|
||||
d = body_of("GET", f"/api/v1/exports/{export_id}", token).get("data") or {}
|
||||
if d.get("status") in (2, 3, 4):
|
||||
return d
|
||||
time.sleep(1)
|
||||
return d
|
||||
|
||||
|
||||
def main() -> int:
|
||||
ap = argparse.ArgumentParser(description="导出分卷与过期清理 E2E")
|
||||
ap.parse_args()
|
||||
|
||||
print("=" * 74)
|
||||
print(" 导出分卷与过期清理 E2E（§12.4）")
|
||||
print("=" * 74)
|
||||
|
||||
_, adm = R.login(BASE, "admin", "Admin@123", "demo")
|
||||
if adm.get("code") != 0:
|
||||
print(f" ✗ 管理员登录失败：{adm}")
|
||||
return 2
|
||||
at = adm["data"]["token"]
|
||||
|
||||
# ── 0. 自己摆演示数据 ──
|
||||
#
|
||||
# ⚠️ 原来这里靠"找任意一个可导出的任务"，而那种任务只有
|
||||
# `e2e-m6-export.py` 跑过才有 —— 也就是**隐式依赖另一个脚本的副作用**。
|
||||
# 全量回归时它排在前面（或那个脚本中途失败）就会以
|
||||
# "没有可导出的任务"退出，看起来像脚本自己坏了。
|
||||
# 自己摆夹具（与 e2e-m6-export.py 同一做法）就没有这个顺序依赖。
|
||||
print("\n[0] 自己摆演示数据（不依赖别的脚本跑过）")
|
||||
import subprocess
|
||||
seed = subprocess.run([sys.executable, os.path.join(ROOT, "scripts", "dev-seed-audit.py"),
|
||||
"--reset", "--approve-one", "--reset-gate-limit"], capture_output=True, text=True, encoding="utf-8",
|
||||
timeout=600)
|
||||
if seed.returncode != 0:
|
||||
print(" ✗ dev-seed-audit.py 失败：")
|
||||
print((seed.stdout or "")[-1200:])
|
||||
print((seed.stderr or "")[-600:])
|
||||
return 2
|
||||
ok("演示数据就绪（dev-seed-audit.py --reset）")
|
||||
|
||||
_, b = R.call("GET", f"{BASE}/api/v1/tasks?pageSize=50", at)
|
||||
tasks = (b.get("data") or {}).get("list") or []
|
||||
task_id = None
|
||||
for t in tasks:
|
||||
# 用 **候选数** 而不是 canExport 判断：演示种子里没人终审通过，
|
||||
# 所以 onlyApproved=true 的候选本来就是 0 —— 那是正确的，不是缺陷
|
||||
pre = body_of("GET", f"/api/v1/exports/precheck?taskId={t['id']}&onlyApproved=true", at)
|
||||
if pre.get("code") == 0 and ((pre.get("data") or {}).get("candidateCount") or 0) > 0:
|
||||
task_id = t["id"]
|
||||
break
|
||||
if task_id is None:
|
||||
print(" ✗ 演示数据里没有可导出的任务（dev-seed-audit.py 的预期没达成）")
|
||||
return 2
|
||||
check("找到可导出的任务（有已提交的填报单）", task_id is not None, f"{task_id}")
|
||||
|
||||
# ★ 这一条是刚修掉的那个缺陷的回归防线：候选为 0 时 canExport 必须是 false，
|
||||
# 否则界面会把「开始导出」点亮，点下去却报"没有可导出的数据"
|
||||
pre_empty = body_of("GET", f"/api/v1/exports/precheck?taskId={task_id}&onlyApproved=true", at)
|
||||
pe = pre_empty.get("data") or {}
|
||||
check("⭐ 候选为 0 时 canExport 必须是 false（按钮不能亮着却导不出）",
|
||||
pe.get("candidateCount") != 0 or pe.get("canExport") is False,
|
||||
f"cand={pe.get('candidateCount')} canExport={pe.get('canExport')}")
|
||||
|
||||
# ── 1. 单卷（不传 splitSizeMb）行为不变 ──
|
||||
print("\n[1] 单卷：不传 splitSizeMb 时行为与以前一致")
|
||||
single = body_of("POST", "/api/v1/exports", at, {
|
||||
"taskId": task_id, "exportName": "E2E分卷-单卷", "scope": {"onlyApproved": True}
|
||||
})
|
||||
check("创建成功", single.get("code") == 0, json.dumps(single, ensure_ascii=False)[:200])
|
||||
sid = (single.get("data") or {}).get("exportId")
|
||||
CREATED_EXPORT_IDS.append(sid)
|
||||
sd = wait_export(sid, at)
|
||||
check("完成", sd.get("status") == 2, f"status={sd.get('status')} err={sd.get('errorMsg')}")
|
||||
check("⭐ 单卷只有 1 个产物", len(sd.get("resultFiles") or []) == 1,
|
||||
f"{len(sd.get('resultFiles') or [])}")
|
||||
check("单卷名里**没有** -part（不分卷时不该多出后缀）",
|
||||
"-part" not in ((sd.get("resultFiles") or [{}])[0].get("name") or ""),
|
||||
f"{(sd.get('resultFiles') or [{}])[0].get('name')}")
|
||||
|
||||
# ── 2. 分卷：1MB 阈值必然分出多卷 ──
|
||||
print("\n[2] 分卷：splitSizeMb=1 必然分出多卷")
|
||||
sp = body_of("POST", "/api/v1/exports", at, {
|
||||
"taskId": task_id, "exportName": "E2E分卷-多卷",
|
||||
"scope": {"onlyApproved": True}, "splitSizeMb": 1
|
||||
})
|
||||
check("创建成功", sp.get("code") == 0, json.dumps(sp, ensure_ascii=False)[:200])
|
||||
spid = (sp.get("data") or {}).get("exportId")
|
||||
CREATED_EXPORT_IDS.append(spid)
|
||||
spd = wait_export(spid, at)
|
||||
check("完成", spd.get("status") == 2, f"status={spd.get('status')} err={spd.get('errorMsg')}")
|
||||
vols = spd.get("resultFiles") or []
|
||||
check("产物非空", len(vols) >= 1, json.dumps(vols, ensure_ascii=False)[:250])
|
||||
check("⭐ 卷名带 -partN（客户按顺序解压）",
|
||||
all(f"-part{i + 1}.zip" in (v.get("name") or "") for i, v in enumerate(vols)),
|
||||
f"{[v.get('name') for v in vols]}")
|
||||
check("⭐ 每一卷都有独立的 sha256（交付完整性可逐卷核对）",
|
||||
all(str(v.get("checksum", "")).startswith("sha256:") for v in vols),
|
||||
f"{[v.get('checksum') for v in vols]}")
|
||||
check("卷号从 0 开始且连续",
|
||||
sorted(v.get("part") for v in vols) == list(range(len(vols))),
|
||||
f"{[v.get('part') for v in vols]}")
|
||||
check("数据库中一票一卷一行",
|
||||
sql(f"select count(*) from collect_export_file where export_id={spid} "
|
||||
f"and deleted=0;").strip() == str(len(vols)),
|
||||
sql(f"select part_no, file_name from collect_export_file where export_id={spid};"))
|
||||
total_mb = sum(float(v.get("sizeMb") or 0) for v in vols)
|
||||
check("⭐ 每卷体积都不超过阈值（1MB，允许压缩后偏小）",
|
||||
all(float(v.get("sizeMb") or 0) <= 1.05 for v in vols),
|
||||
f"{[v.get('sizeMb') for v in vols]}")
|
||||
|
||||
# ── 3. 逐卷解开 —— 这是"每一卷都是完整 ZIP"的唯一可靠验证 ──
|
||||
print("\n[3] 逐卷独立解开（每一卷都必须是完整 ZIP）")
|
||||
all_entries: list[str] = []
|
||||
for i, v in enumerate(vols):
|
||||
dl = body_of("GET", f"/api/v1/exports/{spid}/download?fileIndex={i}", at).get("data") or {}
|
||||
check(f"第 {i + 1} 卷有下载地址", bool(dl.get("url")), json.dumps(dl, ensure_ascii=False)[:200])
|
||||
p = os.path.join(ROOT, "scripts", f"_e2e-split-{i}.zip")
|
||||
os.system(f'curl -s -o "{p}" -H "Authorization: Bearer {at}" "{BASE}{dl.get("url")}"')
|
||||
try:
|
||||
with zipfile.ZipFile(p) as z:
|
||||
bad = z.testzip()
|
||||
check(f"⭐ 第 {i + 1} 卷能**独立**解开（CRC 全过）", bad is None, f"坏条目={bad}")
|
||||
inside = z.namelist()
|
||||
all_entries.extend(inside)
|
||||
check(f"第 {i + 1} 卷根目录带汇总 Excel（只看一卷也能拿到名单）",
|
||||
any(n.endswith(SUMMARY_NAME) for n in inside),
|
||||
f"{inside[:3]}")
|
||||
check(f"第 {i + 1} 卷不是空卷", len(inside) > 0, f"{len(inside)} 个条目")
|
||||
except zipfile.BadZipFile as e:
|
||||
fail(f"第 {i + 1} 卷不是合法 ZIP", str(e))
|
||||
finally:
|
||||
if os.path.exists(p):
|
||||
os.remove(p)
|
||||
|
||||
# 各卷条目合起来 = 单卷的完整交付（不漏不重）
|
||||
#
|
||||
# ⚠️ 在这里**重新取一次下载地址**并校验落地：第 1 节拿到的地址可能在
|
||||
# dev-seed 重置之后已经失效，而复用旧地址的表现是
|
||||
# `FileNotFoundError`（看起来像"下载坏了"，实际是"用了过期地址"）。
|
||||
ssingle = body_of("GET", f"/api/v1/exports/{sid}/download?fileIndex=0", at).get("data") or {}
|
||||
p1 = os.path.join(ROOT, "scripts", "_e2e-single.zip")
|
||||
if os.path.exists(p1):
|
||||
os.remove(p1)
|
||||
# -f 让 curl 在 HTTP 错误时返回非零，而不是安静地不生成文件
|
||||
rc = os.system(f'curl -sf -o "{p1}" -H "Authorization: Bearer {at}" '
|
||||
f'"{BASE}{ssingle.get("url")}"')
|
||||
if not os.path.exists(p1):
|
||||
fail("单卷产物下载失败（无法做合并对比）",
|
||||
f"url={ssingle.get('url')} rc={rc} 单卷记录状态={sd.get('status')}")
|
||||
single_entries = set()
|
||||
else:
|
||||
with zipfile.ZipFile(p1) as z:
|
||||
single_entries = set(z.namelist())
|
||||
os.remove(p1)
|
||||
check("⭐ 分卷条目合起来 = 单卷的完整交付（不漏不重）",
|
||||
set(all_entries) == single_entries,
|
||||
f"分卷独有={sorted(set(all_entries) - single_entries)[:3]} "
|
||||
f"单卷独有={sorted(single_entries - set(all_entries))[:3]}")
|
||||
|
||||
# ── 4. 过期清理 ──
|
||||
print("\n[4] 过期清理：删文件、留记录")
|
||||
n_before = sql(f"select count(*) from collect_export_file where export_id={spid} "
|
||||
f"and downloadable=1;").strip()
|
||||
check("清理前各卷都是可下载的", n_before == str(len(vols)), f"{n_before}")
|
||||
paths = [x.strip() for x in sql(f"select file_path from collect_export_file "
|
||||
f"where export_id={spid} order by part_no;").split("\n") if x.strip()]
|
||||
check("库里记了各卷的磁盘路径", len(paths) == len(vols), f"{paths}")
|
||||
|
||||
# 把过期时间改到 1 小时前 —— 不然要等 7 天
|
||||
sql(f"update collect_export_task set expire_time = now() - interval '1 hour' where id={spid};")
|
||||
run = body_of("POST", "/api/v1/exports/cleanup-run", at, {})
|
||||
check("手工清理入口可用（不然验证过期要等一小时）", run.get("code") == 0,
|
||||
json.dumps(run, ensure_ascii=False)[:200])
|
||||
check("清掉了至少一个文件", ((run.get("data") or {}).get("cleanedFiles") or 0) >= 1,
|
||||
json.dumps(run.get("data"), ensure_ascii=False))
|
||||
|
||||
check("⭐ 产物行**还在**（审计要保留「当时导出过」这个事实）",
|
||||
sql(f"select count(*) from collect_export_file where export_id={spid} "
|
||||
f"and deleted=0;").strip() == str(len(vols)),
|
||||
sql(f"select count(*) from collect_export_file where export_id={spid};"))
|
||||
check("⭐ 全部标成 downloadable=0",
|
||||
sql(f"select count(*) from collect_export_file where export_id={spid} "
|
||||
f"and downloadable=0;").strip() == str(len(vols)),
|
||||
sql(f"select part_no, downloadable from collect_export_file where export_id={spid};"))
|
||||
check("⭐ 任务状态 = 4 已过期（不是 3 失败 —— 过期不是失败）",
|
||||
sql(f"select status from collect_export_task where id={spid};").strip() == "4",
|
||||
sql(f"select status, error_msg from collect_export_task where id={spid};"))
|
||||
check("清完之后磁盘文件真的没了",
|
||||
all(not os.path.exists(p) for p in paths),
|
||||
f"{[p for p in paths if os.path.exists(p)]}")
|
||||
|
||||
dl_after = body_of("GET", f"/api/v1/exports/{spid}/download?fileIndex=0", at)
|
||||
check("⭐ 过期后下载提示「已过期/请重新导出」，而不是像 bug 的「文件不存在」",
|
||||
"过期" in (dl_after.get("msg") or "") or "重新导出" in (dl_after.get("msg") or ""),
|
||||
f"code={dl_after.get('code')} msg={dl_after.get('msg')}")
|
||||
fl_after = body_of("GET", f"/api/v1/exports/{spid}/file?fileIndex=0", at)
|
||||
check("直接取文件流也被拦（不是 500、不是空白文件）",
|
||||
fl_after.get("code") != 0, f"code={fl_after.get('code')}")
|
||||
|
||||
# ── 5. 重复清理安全 ──
|
||||
print("\n[5] 重复清理安全")
|
||||
run2 = body_of("POST", "/api/v1/exports/cleanup-run", at, {})
|
||||
check("再清一次不报错（已清理的卷会跳过）", run2.get("code") == 0,
|
||||
json.dumps(run2, ensure_ascii=False)[:200])
|
||||
check("不会把行删掉或改成别的状态",
|
||||
sql(f"select count(*) from collect_export_file where export_id={spid};").strip()
|
||||
== str(len(vols)))
|
||||
check("单卷那票**没有**被误清理（它是好的，没到过期时间）",
|
||||
sql(f"select count(*) from collect_export_file where export_id={sid} "
|
||||
f"and downloadable=1;").strip() == "1",
|
||||
sql(f"select downloadable from collect_export_file where export_id={sid};"))
|
||||
|
||||
# ── 6. 边界 ──
|
||||
print("\n[6] 边界")
|
||||
b = body_of("GET", f"/api/v1/exports/{spid}/download?fileIndex=99", at)
|
||||
check("分卷号越界 → 明确报错", b.get("code") != 0, f"code={b.get('code')} msg={b.get('msg')}")
|
||||
b = body_of("GET", "/api/v1/exports/999999999/download", at)
|
||||
check("不存在的导出 id → 明确报错而不是 500", b.get("code") != 0,
|
||||
f"code={b.get('code')}")
|
||||
|
||||
cleanup()
|
||||
print("\n" + "=" * 74)
|
||||
print(f"通过 {len(PASSED)} 失败 {len(FAILED)} 共 {len(PASSED) + len(FAILED)}")
|
||||
print("=" * 74)
|
||||
for n, det in FAILED:
|
||||
print(f" ✗ {n}\n {det}")
|
||||
return 1 if FAILED else 0
|
||||
|
||||
|
||||
def cleanup() -> None:
|
||||
for eid in CREATED_EXPORT_IDS:
|
||||
R.sql(f"delete from collect_export_file where export_id={eid};")
|
||||
R.sql(f"delete from collect_export_task where id={eid};")
|
||||
print(" 已清理：测试导出记录与产物行")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
raise SystemExit(main())
|
||||
Reference in New Issue
Block a user