UHD_Media_Tool/packages/adapters/src/sites/spider/spider.test.ts

import { createMockNexusPhpSite, type MockNexusPhpSite } from "@umt/testkit";
import { afterEach, describe, expect, it } from "vitest";
import {
  audiencesDefinition,
  createSiteRateLimiter,
  createSpiderSiteAdapter,
  type IndexerDefinition,
  type SiteAdapter,
  SiteError,
  SiteSpider,
  staticSecretResolver,
} from "../index.ts";

let site: MockNexusPhpSite | undefined;
afterEach(async () => {
  await site?.close();
  site = undefined;
});

function adapterFor(s: MockNexusPhpSite, def = audiencesDefinition, cookie: string | null = s.cookie): SiteAdapter {
  return createSpiderSiteAdapter(def, {
    baseUrl: `${s.baseUrl}/`,
    secrets: staticSecretResolver(cookie ? { [def.id]: { cookie } } : {}),
    limiter: createSiteRateLimiter({ minIntervalMs: 0 }),
  });
}

async function expectSiteError(p: Promise<unknown>, kind: string): Promise<SiteError> {
  try {
    await p;
  } catch (e) {
    expect(e).toBeInstanceOf(SiteError);
    expect((e as SiteError).kind).toBe(kind);
    return e as SiteError;
  }
  throw new Error(`expected SiteError ${kind}`);
}

describe("SiteSpider(MP spider 迁移)", () => {
  const def: IndexerDefinition = {
    ...audiencesDefinition,
    domain: "https://example.com/",
    category: {
      movie: [{ id: "401" }],
      tv: [{ id: "402" }, { id: "403" }],
    },
  };

  it("__get_search_url:默认参数、search_area 仅 IMDb 保留、mtype 选分类组、cat 白名单", () => {
    const u1 = new URL(new SiteSpider(def, { keyword: "Dune", page: 2 }).getSearchUrl());
    expect(u1.pathname).toBe("/torrents.php");
    expect(Object.fromEntries(u1.searchParams)).toEqual({
      search_mode: "0",
      search_area: "0",
      page: "2",
      notnewword: "1",
      search: "Dune",
      cat401: "1",
      cat402: "1",
      cat403: "1",
    });
    const u2 = new URL(new SiteSpider(def, { keyword: "tt15239678" }).getSearchUrl());
    expect(u2.searchParams.get("search_area")).toBe("4");
    const u3 = new URL(new SiteSpider(def, { keyword: "x", mtype: "tv", cat: "403" }).getSearchUrl());
    expect([...u3.searchParams.keys()].filter((k) => k.startsWith("cat"))).toEqual(["cat403"]);
    const u4 = new URL(new SiteSpider(def, { keyword: ["A B", "C"] }).getSearchUrl());
    expect(u4.searchParams.get("search")).toBe("A_B C");
    expect(u4.searchParams.get("search_mode")).toBe("1");
    // 浏览:browse.path + start
    expect(new SiteSpider(def, { page: 3 }).getSearchUrl()).toBe("https://example.com/torrents.php?page=3");
  });

  it("分类参数 param(cat[] 重复)与 field 拼接两种写法", () => {
    const withParam: IndexerDefinition = { ...def, category: { param: "cat[]", tv: [{ id: "402" }, { id: "403" }] } };
    const u = new URL(new SiteSpider(withParam, { keyword: "x" }).getSearchUrl());
    expect(u.searchParams.getAll("cat[]")).toEqual(["402", "403"]);
    const withField: IndexerDefinition = { ...def, category: { field: "cat", delimiter: ",", tv: [{ id: "402" }] } };
    expect(new URL(new SiteSpider(withField, { keyword: "x" }).getSearchUrl()).searchParams.get("cat")).toBe(",402");
  });

  it("get_info 字段 DSL:模板标题、remove、dateparse、case 系数、querystring、hr 配置后两态", () => {
    const html = `<html><head><title>t</title></head><body><table class="torrents"><tr><td>h</td></tr>
      <tr><td><a href="?cat=402"><img></a></td><td><table class="torrentname"><tr><td class="embedded">
        <a href="details.php?id=100" title="Full.Title.2025.2160p">Full.Title…</a> <img class="free">
        <font class="subtitle">副标题 <span>[置顶]</span><a>x</a></font><span class="label">中字</span><span class="label">DIY</span>
      </td></tr></table></td><td>0</td><td><span title="2025-05-01 12:13:14">1天</span></td><td>1.5 GB</td><td>1,234</td><td>5/7</td><td>9</td></tr>
      <tr><td><a href="?cat=401"></a></td><td><table class="torrentname"><tr><td class="embedded">
        <a href="details.php?id=101">No.Title.Attr</a> <img class="hitandrun">
      </td></tr></table></td><td>0</td><td><span>3 小时</span></td><td>--</td><td>x</td><td></td><td></td></tr></table></body></html>`;
    const d: IndexerDefinition = {
      ...def,
      torrents: {
        list: { selector: 'table.torrents > tr:has("table.torrentname")' },
        fields: {
          ...def.torrents.fields,
          description: { selector: "font.subtitle", remove: "span,a" },
          labels: { selector: "span.label" },
          hr: { selector: "img.hitandrun" },
          downloadvolumefactor: { case: { "img.free": 0, "*": 1 } },
        },
      },
    };
    const { torrents } = new SiteSpider(d).parse(html);
    expect(torrents[0]).toMatchObject({
      title: "Full.Title.2025.2160p",
      description: "副标题",
      page_url: "https://example.com/details.php?id=100",
      size: 1610612736,
      seeders: 1234,
      peers: 5,
      grabs: 9,
      downloadvolumefactor: 0,
      uploadvolumefactor: 1,
      pubdate: "2025-05-01 12:13:14",
      labels: ["中字", "DIY"],
      hit_and_run: false,
      category: "电视剧",
      category_id: "402",
    });
    expect(torrents[1]).toMatchObject({
      title: "No.Title.Attr",
      hit_and_run: true,
      size: 0,
      seeders: 0,
      category: "电影",
    });
    expect(torrents[1]!.pubdate).toBeUndefined(); // 相对时间不写入 pubdate(MP __is_invalid_pubdate_text)
  });

  it("登录/权限页判定(MP __is_login_or_permission_page)", () => {
    const r = new SiteSpider(def).parse("<html><head><title>观众 :: 登录</title></head><body></body></html>");
    expect(r).toMatchObject({ isError: true, loginPage: true, torrents: [] });
    const r2 = new SiteSpider(def).parse("<html><body><p>该页面必须在登录后才能访问</p></body></html>");
    expect(r2.loginPage).toBe(true);
  });
});

describe("观众适配器(spider + mock NexusPHP)", () => {
  it("列表:page=1 发送 page=0;免费+截止、时间、原始分类、副标题;H&R 未配置为 unknown", async () => {
    site = await createMockNexusPhpSite();
    const page = await adapterFor(site).list(1);
    const req = site.requests[0]!;
    expect(req.path).toBe("/torrents.php?page=0");
    expect(req.headers.cookie).toBe(site.cookie);
    expect(page.items).toHaveLength(5);
    expect(page.totalPages).toBe(1);
    const dune = page.items[0]!;
    expect(dune).toMatchObject({
      siteId: "audiences",
      siteReleaseId: "812345",
      title: "Dune.Part.Two.2024.2160p.UHD.BluRay.REMUX.DV.HDR.HEVC.TrueHD.Atmos.7.1-FraMeSToR",
      subtitle: "沙丘2 / 沙丘：第二部 | 杜比视界 国英双语 中字",
      category: { id: "401", name: "电影" },
      sizeBytes: 84202833838,
      seeders: 1204,
      leechers: 17,
      snatched: 3388,
      promotion: { type: "free", downloadFactor: 0, uploadFactor: 1, expiresAt: "2026-10-01T04:00:00.000Z" },
      hr: "unknown",
      uploadedAt: "2026-09-26T13:14:03.000Z",
    });
    expect(dune.raw).toMatchObject({ category: "电影", category_id: "401", pubdate: "2026-09-26 21:14:03" });
    expect(dune.detailUrl).toBe(`${site.baseUrl}/details.php?id=812345&hit=1`);
    expect(JSON.stringify(page)).not.toContain("mock-session-secret");
  });

  it("列表:综艺保留 403、50%+2x、2x、截断标题、TB/KB", async () => {
    site = await createMockNexusPhpSite();
    const { items } = await adapterFor(site).list(1);
    const byId = Object.fromEntries(items.map((i) => [i.siteReleaseId, i]));
    expect(byId["812346"]!.category).toEqual({ id: "403", name: "综艺" });
    expect(byId["812346"]!.raw.category).toBe("电视剧");
    expect(byId["812346"]!.promotion.type).toBe("none");
    expect(byId["812347"]!.promotion).toMatchObject({ type: "2x50%", downloadFactor: 0.5, uploadFactor: 2 });
    expect(byId["812347"]!.promotion.expiresAt).toBe("2026-09-29T00:30:00.000Z");
    expect(byId["812347"]!.sizeBytes).toBe(1294529987);
    expect(byId["812348"]!.title).toBe("The.Last.of.Us.S02.2160p.MAX.WEB-DL.DDP5.1.Atmos.D…");
    expect(byId["812348"]!.sizeBytes).toBe(2308974418330);
    expect(byId["812349"]!.promotion).toMatchObject({
      type: "2x",
      downloadFactor: 1,
      uploadFactor: 2,
      expiresAt: null,
    });
    expect(byId["812349"]!.category).toEqual({ id: "406", name: "纪录片" });
  });

  it("配置 hr 字段后 H&R 为 true/false", async () => {
    site = await createMockNexusPhpSite();
    const def: IndexerDefinition = {
      ...audiencesDefinition,
      torrents: {
        ...audiencesDefinition.torrents,
        fields: { ...audiencesDefinition.torrents.fields, hr: { selector: "img.hitandrun" } },
      },
    };
    const { items } = await adapterFor(site, def).list(1);
    expect(items.find((i) => i.siteReleaseId === "812346")!.hr).toBe(true);
    expect(items.find((i) => i.siteReleaseId === "812345")!.hr).toBe(false);
  });

  it("搜索:关键词清洗与分类下推;无结果返回空页", async () => {
    site = await createMockNexusPhpSite();
    const a = adapterFor(site);
    const r = await a.search("奔跑吧!", { categoryIds: ["403"] });
    const u = new URL(site.requests[0]!.path, site.baseUrl);
    expect(u.searchParams.get("search")).toBe("奔跑吧");
    expect(u.searchParams.get("cat403")).toBe("1");
    expect(u.searchParams.get("cat401")).toBeNull();
    expect(r.items.map((i) => i.siteReleaseId)).toEqual(["812346"]);
    expect((await a.search("不存在的片名")).items).toEqual([]);
  });

  it("分页:pageSize=2 时 page=2 发送 page=1,总页数由分页链接估计", async () => {
    site = await createMockNexusPhpSite({ pageSize: 2 });
    const p2 = await adapterFor(site).list(2);
    expect(site.requests[0]!.path).toBe("/torrents.php?page=1");
    expect(p2).toMatchObject({ totalPages: 3, totalPagesEstimated: true });
    expect(p2.items.map((i) => i.siteReleaseId)).toEqual(["812347", "812348"]);
  });

  it("详情:标题/副标题/体积/同伴/促销/IMDb/豆瓣;不存在 → not_found", async () => {
    site = await createMockNexusPhpSite();
    const a = adapterFor(site);
    const d = await a.detail("812345");
    expect(d).toMatchObject({
      title: "Dune.Part.Two.2024.2160p.UHD.BluRay.REMUX.DV.HDR.HEVC.TrueHD.Atmos.7.1-FraMeSToR",
      subtitle: "沙丘2 / 沙丘：第二部 | 杜比视界 国英双语 中字",
      sizeBytes: 84202833838,
      seeders: 1204,
      leechers: 17,
      imdbId: "tt15239678",
      doubanId: "35575567",
      uploadedAt: "2026-09-26T13:14:03.000Z",
      category: { id: "401", name: "电影" },
      promotion: { type: "free", expiresAt: "2026-10-01T04:00:00.000Z" },
    });
    await expectSiteError(a.detail("999999"), "not_found");
  });

  it("下载目标:带 Cookie 取 .torrent 并校验 infohash", async () => {
    site = await createMockNexusPhpSite();
    const t = await adapterFor(site).downloadTarget("812345");
    expect(t.kind).toBe("torrent");
    if (t.kind !== "torrent") return;
    expect(t.infoHash).toBe(site.infoHashOf("812345"));
    expect(t.name).toContain("Dune.Part.Two");
    expect(t.fileName).toContain(".torrent");
  });

  it("登录失效识别:列表/详情/下载抛 auth_expired,健康检查报告", async () => {
    site = await createMockNexusPhpSite();
    const a = adapterFor(site);
    expect((await a.healthCheck()).status).toBe("ok");
    site.setLoggedIn(false);
    const err = await expectSiteError(a.list(1), "auth_expired");
    expect(err.isAuthError).toBe(true);
    await expectSiteError(a.detail("812345"), "auth_expired");
    await expectSiteError(a.downloadTarget("812345"), "auth_expired");
    expect(await a.healthCheck()).toMatchObject({ ok: false, status: "auth_expired" });
  });

  it("登录失效(302 login.php)与缺少 Cookie", async () => {
    site = await createMockNexusPhpSite({ loginMode: "redirect" });
    await expectSiteError(
      adapterFor(site, audiencesDefinition, "c_secure_uid=1; c_secure_pass=wrong").list(1),
      "auth_expired",
    );
    const before = site.requests.length;
    await expectSiteError(adapterFor(site, audiencesDefinition, null).list(1), "auth_missing");
    expect(site.requests.length).toBe(before);
  });

  it("页面改版 → parse_error;429 + Retry-After → rate_limited", async () => {
    site = await createMockNexusPhpSite();
    const a = adapterFor(site);
    site.setBrokenLayout(true);
    await expectSiteError(a.list(1), "parse_error");
    site.setBrokenLayout(false);
    site.failNext(1, 429, "7");
    expect((await expectSiteError(a.list(1), "rate_limited")).retryAfterMs).toBe(7000);
  });

  it("站点级限流:列表/搜索/详情/下载共享同一令牌桶", async () => {
    site = await createMockNexusPhpSite();
    let clock = 0;
    const sleeps: number[] = [];
    const limiter = createSiteRateLimiter({
      minIntervalMs: 3000,
      now: () => clock,
      sleep: async (ms) => {
        sleeps.push(ms);
        clock += ms;
      },
    });
    const a = createSpiderSiteAdapter(audiencesDefinition, {
      baseUrl: `${site.baseUrl}/`,
      secrets: staticSecretResolver({ audiences: { cookie: site.cookie } }),
      limiter,
    });
    await a.list(1);
    await a.search("Dune");
    await a.detail("812345");
    await a.downloadTarget("812345");
    expect(sleeps).toEqual([3000, 3000, 3000]);
    expect(limiter.stats.granted).toBe(4);
  });
});