UHD_Media_Tool/apps/server/test/sources-identity/failure-sample.test.ts

import type { SiteError } from "@umt/adapters";
import { afterAll, beforeAll, beforeEach, describe, expect, it } from "vitest";
import { SAMPLES_PER_SITE } from "../../src/modules/sources-identity/service/failure-sample.ts";
import { snapshotSite } from "../../src/modules/sources-identity/service/snapshot.ts";
import { createHarness, type Harness, pgAvailable, READ_KEY } from "./harness.ts";

// 站点请求失败留样:列表 / 搜索报「页面结构无法识别」时存脱敏页面样本(site_failure_sample),每站只留最近 5 份;
// 站点页技术信息经 GET /sites/{id}/failure-samples(列表不带正文)与 /failure-samples/{sampleId}(带正文)取。
// 站点前面是可控的 fetch:返回什么页面由测试决定(不访问真实站点)。

const up = await pgAvailable();
const SECRET = "0123456789abcdef0123456789abcdef";
/** 改版页:认不出种子表,还带着要脱敏的东西 */
const BROKEN = `<html><head><title>Audiences :: 种子</title></head><body>
<table id="info_block"><tr><td>欢迎 <a href="userdetails.php?id=29191"><b>alice_cn</b></a> UID 29191</td></tr></table>
<p>邮箱 [email protected]</p><a href="download.php?id=1&passkey=${SECRET}">dl</a>
<script>var authkey = "${SECRET}"; document.cookie = "c_secure_pass=topsecretvalue";</script>
<div class="new-grid"><div class="card">Dune</div></div></body></html>`;

// biome-ignore lint/suspicious/noExplicitAny: 响应体断言
const json = async (r: Response): Promise<any> => {
  const body = await r.json();
  if (!r.ok) throw new Error(`${r.status} ${JSON.stringify(body)}`);
  return body.data;
};

/** 轮询到条件成立(留样是异步写入的) */
async function until<T>(f: () => Promise<T>, ok: (v: T) => boolean, ms = 5_000): Promise<T> {
  const deadline = Date.now() + ms;
  for (;;) {
    const v = await f();
    if (ok(v) || Date.now() > deadline) return v;
    await new Promise((r) => setTimeout(r, 50));
  }
}

describe.skipIf(!up)("站点失败留样", () => {
  let h: Harness;
  let siteId: string;
  let page = BROKEN;
  const fetchImpl = (async () =>
    new Response(page, {
      status: 200,
      headers: { "content-type": "text/html; charset=utf-8", server: "nginx" },
    })) as typeof fetch;
  const site = () => h.db.selectFrom("site").selectAll().where("id", "=", siteId).executeTakeFirstOrThrow();
  const rows = () =>
    h.db.selectFrom("site_failure_sample").selectAll().where("site_id", "=", siteId).orderBy("created_at").execute();

  beforeAll(async () => {
    h = await createHarness("failsample", { fetch: fetchImpl });
    const s = await json(
      await h.req("/sites", {
        body: {
          key: "audiences",
          name: "观众",
          adapter: "nexusphp",
          baseUrl: "https://aud.test/",
          rateLimit: { minIntervalMs: 0, concurrency: 10 },
        },
      }),
    );
    siteId = s.id;
    await json(
      await h.req(`/sites/${siteId}/credentials`, {
        method: "PUT",
        body: { kind: "cookie", cookie: "c_secure_uid=1; c_secure_pass=abc" },
      }),
    );
  });
  afterAll(async () => {
    await h?.close();
  });
  beforeEach(async () => {
    page = BROKEN;
    await h.db.deleteFrom("site_failure_sample").where("site_id", "=", siteId).execute();
  });

  it("列表快照报 parse_error → 留脱敏样本(URL、状态码、关键响应头、请求类型);接口可看", async () => {
    const r = await snapshotSite(h.rt, { siteId, kind: "manual" });
    expect(r.status).toBe("failed");
    expect(r.error).toContain("parse_error");
    const got = await until(rows, (x) => x.length > 0);
    expect(got).toHaveLength(1);
    const s = got[0]!;
    expect(s).toMatchObject({
      kind: "parse_error",
      request_kind: "list",
      http_status: 200,
      headers: { server: "nginx" },
    });
    expect(s.message).toContain("页面结构无法识别");
    for (const leak of [SECRET, "[email protected]", "alice_cn", "29191", "topsecretvalue"]) {
      expect(s.body).not.toContain(leak);
      expect(s.url).not.toContain(leak);
    }
    expect(s.body).toContain("new-grid");

    const list = await json(await h.req(`/sites/${siteId}/failure-samples`, { key: READ_KEY }));
    expect(list.items).toHaveLength(1);
    expect(list.items[0]).toMatchObject({ id: s.id, kind: "parse_error", requestKind: "list", httpStatus: 200 });
    expect(list.items[0].body).toBeUndefined();
    expect(list.items[0].bodyBytes).toBeGreaterThan(0);
    const one = await json(await h.req(`/sites/${siteId}/failure-samples/${s.id}`, { key: READ_KEY }));
    expect(one.body).toBe(s.body);
    const miss = await h.req(`/sites/${siteId}/failure-samples/00000000-0000-4000-8000-000000000000`);
    expect(miss.status).toBe(404);
  });

  it(`每站只留最近 ${SAMPLES_PER_SITE} 份`, async () => {
    const adapter = h.rt.siteAdapter(await site());
    for (let i = 1; i <= 7; i++) {
      const e = await adapter.search(`k${i}`).catch((x) => x);
      expect((e as SiteError).kind).toBe("parse_error");
    }
    const kw = (x: { url: string }[]) => x.map((r) => /search=(k\d)/.exec(r.url)?.[1]).join(",");
    const got = await until(rows, (x) => kw(x) === "k3,k4,k5,k6,k7");
    expect(kw(got)).toBe("k3,k4,k5,k6,k7");
    expect(got.every((r) => r.request_kind === "search")).toBe(true);
  });

  it("正常页面不留样", async () => {
    page = `<html><head><title>Audiences :: 种子</title></head><body><table class="torrents"><tr><td>没有种子</td></tr></table></body></html>`;
    const adapter = h.rt.siteAdapter(await site());
    await adapter.search("x").catch(() => null);
    await new Promise((r) => setTimeout(r, 200));
    expect(await rows()).toEqual([]);
  });
});