UHD_Media_Tool/apps/server/test/sources-identity/identity-change-rescrape.test.ts

import { readFile } from "node:fs/promises";
import { join } from "node:path";
import { createMockTmdb, createTempTree, type MockTmdb, type TempTree, TMDB_FIXTURES } from "@umt/testkit";
import { afterAll, beforeAll, describe, expect, it } from "vitest";
import type { AoPorts } from "../../src/modules/acquisition-organize/ports.ts";
import type { AoDeps } from "../../src/modules/acquisition-organize/services/common.ts";
import {
  RESCRAPE_DELAY_MS,
  runScrapeRecord,
  SCRAPE_TASK,
  scrapeEntity,
} from "../../src/modules/acquisition-organize/services/scrape.ts";
import { createEntityFromTmdbCandidate } from "../../src/modules/sources-identity/service/entity-from-candidate.ts";
import { setEntityIdentity } from "../../src/modules/sources-identity/service/identity-admin.ts";
import { refreshMetadata, setFieldOverride } from "../../src/modules/sources-identity/service/metadata.ts";
import { createHarness, type Harness, pgAvailable } from "./harness.ts";

// 改作品的 TMDB 映射后:清掉旧作品的 tmdb 来源字段(人工锁定的保留)→ full 刷新 → 刮削过的作品延后排一次「更新」模式重刮;
// 重刮后的 NFO 标题 / 简介 / IMDb / 海报都是新作品的,不残留旧作品;刷新期间再刷新合进同一条排队记录(防抖)。

const up = await pgAvailable();
const enc = (s: string) => new TextEncoder().encode(s);
const VIDEO = "Movies/奥本海默 (2023)/Oppenheimer.2023.1080p.mkv";
const DIR = "Movies/奥本海默 (2023)";
const NFO = `${DIR}/Oppenheimer.2023.1080p.nfo`;
type Final = { value: unknown; source: string; locked: boolean };

describe.skipIf(!up)("改映射后不再新旧混写:清字段 → 刷新 → 更新模式重刮", () => {
  let h: Harness;
  let tmdb: MockTmdb;
  let lib: TempTree;
  let d: AoDeps;
  let id = "";
  const actor = { type: "user" as const, id: "tester" };
  const enqueueOpts: { task: string; availableAt?: Date | undefined }[] = [];

  beforeAll(async () => {
    // 沙丘 2 补一张海报,验证改映射后海报换成新作品的
    const fixtures = TMDB_FIXTURES.map((f) => (f.id === 693134 ? { ...f, poster_path: "/dune.jpg" } : f));
    tmdb = await createMockTmdb({ fixtures });
    h = await createHarness("idrescrape", { tmdbUrl: tmdb.url });
    lib = await createTempTree({ [VIDEO]: 1000 }, { sentinel: true, prefix: "umt-idscr-" });
    const orig = h.engine.enqueue.bind(h.engine);
    h.engine.enqueue = async (task, input, o) => {
      enqueueOpts.push({ task, availableAt: o.availableAt });
      return orig(task, input, o);
    };
    d = {
      ctx: h.ctx,
      ports: {
        now: () => new Date(),
        scrapeImage: async (p: string) => ({ content: enc(p), mime: "image/jpeg" }),
        scrapeTmdbDetail: async () => null,
        scrapeTmdbSeason: async () => null,
      } as unknown as AoPorts,
    };
  }, 60_000);
  afterAll(async () => {
    await h?.close();
    await tmdb?.close();
    await lib?.cleanup();
  });

  const queuedUpdates = () =>
    h.db
      .selectFrom("scrape_record")
      .select(["id", "mode", "trigger", "status"])
      .where("media_entity_id", "=", id)
      .where("status", "=", "queued")
      .execute();

  it("先按奥本海默建档、刷新、刮削;没刮削过时刷新不排重刮", async () => {
    const r = await createEntityFromTmdbCandidate(
      h.db,
      () => h.rt.tmdb(),
      { externalId: "872585", kind: "movie" },
      actor,
    );
    id = r.mediaEntityId;
    expect((await refreshMetadata(h.rt, id, { full: true })).status).toBe("refreshed");
    expect(h.engine.enqueued.some((x) => x.task === SCRAPE_TASK)).toBe(false);
    await setFieldOverride(h.rt, id, "tagline", { value: "我的标语", lock: true }, actor);

    const root = await h.db
      .insertInto("library_root")
      .values({ name: "lib-idscr", media_kind: "mixed", local_path: lib.root, management_mode: "managed" })
      .returning("id")
      .executeTakeFirstOrThrow();
    const asset = await h.db
      .insertInto("media_asset")
      .values({ media_entity_id: id, kind: "video" })
      .returning("id")
      .executeTakeFirstOrThrow();
    await h.db
      .insertInto("asset_location")
      .values({ media_asset_id: asset.id, library_root_id: root.id, relative_path: VIDEO })
      .execute();

    const q = await scrapeEntity(d, id, { trigger: "manual" });
    expect(q).toHaveLength(1);
    const rec = await runScrapeRecord(d, q[0]!.recordId);
    expect(rec).toMatchObject({ status: "succeeded", mode: "update" });
    expect(rec.files.every((f) => f.action !== "written" || /^[0-9a-f]{64}$/.test(f.hash ?? ""))).toBe(true);
    const nfo = await readFile(join(lib.root, NFO), "utf8");
    expect(nfo).toContain("<title>奥本海默</title>");
    expect(nfo).toContain("tt15398776");
    expect(nfo).toContain("<tagline>我的标语</tagline>");
    expect(await readFile(join(lib.root, DIR, "poster.jpg"), "utf8")).toBe("/oppenheimer.jpg");
  }, 60_000);

  it("改映射到沙丘 2:tmdb 来源字段清空,锁定字段保留,排 full 刷新", async () => {
    const before = h.engine.enqueued.length;
    const r = await setEntityIdentity(
      h.rt,
      id,
      { provider: "tmdb", entityType: "movie", externalId: "693134", lock: true },
      actor,
    );
    expect(r.changed).toBe(true);
    const e = await h.db.selectFrom("media_entity").select(["metadata"]).where("id", "=", id).executeTakeFirstOrThrow();
    const meta = e.metadata as Record<string, Final>;
    expect(meta.overview).toBeUndefined();
    expect(meta.posterPath).toBeUndefined();
    expect(meta.externalIds).toBeUndefined();
    expect(meta.tagline).toMatchObject({ value: "我的标语", locked: true });
    const tmdbRows = await h.db
      .selectFrom("metadata_field")
      .select("field")
      .where("media_entity_id", "=", id)
      .where("source", "=", "tmdb")
      .execute();
    expect(tmdbRows).toEqual([]);
    expect(h.engine.enqueued.slice(before)).toEqual([
      expect.objectContaining({ task: "refreshMetadata", input: { mediaEntityId: id, full: true } }),
    ]);
  });

  it("刷新后延后排一次更新模式重刮;期间再刷新合进同一条", async () => {
    const t0 = Date.now();
    expect((await refreshMetadata(h.rt, id, { full: true })).status).toBe("refreshed");
    const queued = await queuedUpdates();
    expect(queued).toEqual([expect.objectContaining({ mode: "update", trigger: "auto" })]);
    const opt = enqueueOpts.filter((x) => x.task === SCRAPE_TASK).at(-1);
    expect(opt?.availableAt?.getTime()).toBeGreaterThanOrEqual(t0 + RESCRAPE_DELAY_MS - 1000);

    // 防抖:还没开始时再来一次刷新(哪怕字段又变了)不另建记录、不另排任务
    await h.db
      .updateTable("metadata_field")
      .set({ value: JSON.stringify("旧值") })
      .where("media_entity_id", "=", id)
      .where("field", "=", "overview")
      .execute();
    const scrapes = h.engine.enqueued.filter((x) => x.task === SCRAPE_TASK).length;
    await refreshMetadata(h.rt, id, { full: true });
    expect(await queuedUpdates()).toHaveLength(1);
    expect(h.engine.enqueued.filter((x) => x.task === SCRAPE_TASK).length).toBe(scrapes);
  }, 60_000);

  it("重刮后 NFO 的标题 / 简介 / IMDb / 海报都是沙丘 2,不残留奥本海默;锁定字段照写", async () => {
    const [q] = await queuedUpdates();
    const rec = await runScrapeRecord(d, q!.id);
    expect(rec.status).toBe("succeeded");
    expect(rec.files.find((f) => f.role === "movie_nfo")).toMatchObject({ action: "written" });
    const nfo = await readFile(join(lib.root, NFO), "utf8");
    expect(nfo).toContain("<title>沙丘2</title>");
    expect(nfo).toContain("<tmdbid>693134</tmdbid>");
    expect(nfo).toContain("tt15239678");
    expect(nfo).not.toContain("tt15398776");
    expect(nfo).not.toContain("奥本海默");
    expect(nfo).not.toContain("原子弹之父");
    expect(nfo).toContain("<tagline>我的标语</tagline>");
    expect(await readFile(join(lib.root, DIR, "poster.jpg"), "utf8")).toBe("/dune.jpg");

    // 字段没变的刷新不再排重刮
    const scrapes = h.engine.enqueued.filter((x) => x.task === SCRAPE_TASK).length;
    await refreshMetadata(h.rt, id, { full: true });
    expect(h.engine.enqueued.filter((x) => x.task === SCRAPE_TASK).length).toBe(scrapes);
  }, 60_000);
});