UHD_Media_Tool/apps/server/test/sources-identity/organize-episode-samples.test.ts

import { randomUUID } from "node:crypto";
import { EpisodeMappingInspectOutput, EpisodeMappingSimulateOutput } from "@umt/contracts";
import { parseRelease } from "@umt/domain";
import { afterAll, beforeAll, describe, expect, it } from "vitest";
import { resolveEpisodeDecision } from "../../src/modules/sources-identity/service/episode-mapping-admin.ts";
import {
  applyAgentEpisodeRule,
  inspectEpisodeMapping,
  simulateForAgent,
} from "../../src/modules/sources-identity/service/episode-mapping-agent.ts";
import {
  clearReleaseEpisodes,
  type EpisodeFileSample,
  episodeFileSamples,
  reresolveEntityEpisodes,
  resolveReleaseEpisodes,
  settleEpisodeDecision,
  settleOrganizeEpisodeDecisions,
} from "../../src/modules/sources-identity/service/episode-resolution.ts";
import { isEntityWatched } from "../../src/modules/sources-identity/service/watched.ts";
import { setWordsText } from "../support/identify-words.ts";
import { createHarness, type Harness, pgAvailable, seedSite } from "./harness.ts";

const pg = await pgAvailable();
const ACTOR = { type: "user" as const, id: "admin" };

describe.skipIf(!pg)("整理真实文件的季集映射闭环", () => {
  let h: Harness;
  let siteId: string;
  beforeAll(async () => {
    h = await createHarness("organize_episode_samples");
    siteId = (await seedSite(h, { baseUrl: "https://example.test" })).site.id;
  }, 120_000);
  afterAll(async () => {
    await h?.close();
  });

  const series = async () => {
    const entity = await h.db
      .insertInto("media_entity")
      .values({ title: "整理分季测试", kind: "series", metadata: JSON.stringify({ airStatus: "Ended" }) })
      .returning("id")
      .executeTakeFirstOrThrow();
    await h.db
      .insertInto("content_unit")
      .values([
        {
          media_entity_id: entity.id,
          kind: "season",
          season_number: 1,
          episode_total: 12,
          source: "tmdb",
          status: "confirmed",
        },
        {
          media_entity_id: entity.id,
          kind: "season",
          season_number: 2,
          episode_total: 13,
          source: "tmdb",
          status: "confirmed",
        },
      ])
      .execute();
    return entity.id;
  };
  const file = async (
    entityId: string,
    episode: number,
    group = "ANi",
    publishedAt: string | null = null,
  ): Promise<Omit<EpisodeFileSample, "recordId">> => {
    const path = `/downloads/[${group}] Test S01E${episode}.mkv`;
    const parsed = parseRelease(path.split("/").at(-1)!);
    const mapped = await resolveReleaseEpisodes(h.db, { entityId, kind: "series", parsed, publishedAt });
    return { path, parsed, resolution: mapped.resolution, publishedAt };
  };
  const pending = (entityId: string) =>
    h.db
      .selectFrom("decision")
      .selectAll()
      .where("kind", "=", "episode_mapping")
      .where("subject_id", "=", entityId)
      .where("status", "in", ["open", "investigating", "awaiting_user"])
      .execute();
  const release = async (entityId: string, episode: number, group = "ANi") => {
    const title = `[${group}] Test S01E${episode} [1080p]`;
    const parsed = parseRelease(title);
    const mapped = await resolveReleaseEpisodes(h.db, { entityId, kind: "series", parsed });
    const row = await h.db
      .insertInto("source_release")
      .values({
        site_id: siteId,
        site_release_id: randomUUID(),
        title,
        media_entity_id: entityId,
        resolution_status: "resolved",
        episode_resolution: JSON.stringify(mapped.resolution),
        parsed: JSON.stringify(parsed),
      })
      .returningAll()
      .executeTakeFirstOrThrow();
    return { row, parsed, resolution: mapped.resolution };
  };

  it("无订阅、无资源站资源也按真实文件归集，重复记录样本不累加，inspect保留未知发布时间", async () => {
    const entityId = await series();
    expect(await isEntityWatched(h.db, entityId)).toBe(false);
    const recordId = randomUUID();
    const files = [await file(entityId, 40), await file(entityId, 41), await file(entityId, 50, "Other")];
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, { recordId, entityId, files });
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, { recordId, entityId, files });
    const decisions = await pending(entityId);
    expect(decisions).toHaveLength(2);
    const ani = decisions.find((d) => d.dedupe_key.endsWith(":ani"))!;
    expect(episodeFileSamples(ani.evidence)).toHaveLength(2);
    expect(ani.evidence).toMatchObject({ sourceReleaseIds: [], suggestion: null, count: 2 });
    const inspect = EpisodeMappingInspectOutput.parse(await inspectEpisodeMapping(h.db, { decisionId: ani.id }));
    expect(inspect.releases).toEqual([]);
    expect(inspect.files).toHaveLength(2);
    expect(inspect.files[0]).toMatchObject({
      organizationRecordId: recordId,
      sourcePath: files[0]!.path,
      publishedAt: null,
      source: { season: 1, episodes: { start: 40, end: 40 } },
    });
    expect(inspect.suggestion).toBeNull();
  });

  it("模拟同时验证每个文件，无法只靠一个正确资源标题放行整包", async () => {
    const entityId = await series();
    await h.db
      .insertInto("subscription")
      .values({ media_entity_id: entityId, media_kind: "series", status: "active" })
      .execute();
    const r = await release(entityId, 40);
    await settleEpisodeDecision(h.ctx, h.db, {
      sourceReleaseId: r.row.id,
      entityId,
      parsed: r.parsed,
      resolution: r.resolution,
    });
    const files = [await file(entityId, 41), await file(entityId, 60)];
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, { recordId: randomUUID(), entityId, files });
    const d = (await pending(entityId))[0]!;
    const simulation = EpisodeMappingSimulateOutput.parse(
      await simulateForAgent(h.db, { decisionId: d.id, line: "S01 @ANi >> S02 EP-30" }, null),
    );
    expect(simulation.summary).toMatchObject({ samples: 3, resolved: 2, ok: false });
    expect(simulation.failures.join(";")).toContain(files[1]!.path);
    expect(simulation.items.filter((i) => i.sourceReleaseId === null)).toHaveLength(2);
    await expect(
      applyAgentEpisodeRule(h.ctx, { decisionId: d.id, line: simulation.line, reasons: ["核对真实文件"] }, ACTOR, null),
    ).rejects.toThrow("模拟没通过");
    expect(
      await h.db.selectFrom("identify_words").select("id").where("media_entity_id", "=", entityId).execute(),
    ).toEqual([]);
  });

  it("纯文件规则模拟通过后自动重算并关闭，后续同范围文件沿用作品规则", async () => {
    const entityId = await series();
    const recordId = randomUUID();
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId,
      entityId,
      files: [await file(entityId, 40), await file(entityId, 41)],
    });
    const d = (await pending(entityId))[0]!;
    const sim = await simulateForAgent(
      h.db,
      { decisionId: d.id, line: "S01 @ANi >> S02 EP-30", reasons: ["资源站与文件对应 S02E10-11"] },
      { runId: null },
    );
    expect(sim.summary).toMatchObject({ samples: 2, resolved: 2, passed: 2, ok: true });
    expect(sim.suggestionRecorded).toBe(true);
    const out = await applyAgentEpisodeRule(
      h.ctx,
      { decisionId: d.id, line: sim.line, reasons: ["文件编号已核对"] },
      ACTOR,
      null,
    );
    expect(out.decisionStatus).toBe("resolved");
    expect(await pending(entityId)).toEqual([]);
    const after = await h.db.selectFrom("decision").select("evidence").where("id", "=", d.id).executeTakeFirstOrThrow();
    expect(episodeFileSamples(after.evidence).map((f) => f.resolution.canonical)).toEqual([
      { season: 2, episodes: { start: 10, end: 10 } },
      { season: 2, episodes: { start: 11, end: 11 } },
    ]);
    expect((await file(entityId, 42)).resolution.canonical).toEqual({ season: 2, episodes: { start: 12, end: 12 } });
  });

  it("普通保存作品识别词也重算纯文件事项，无资源不能提前返回", async () => {
    const entityId = await series();
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 40)],
    });
    await setWordsText(h.db, entityId, "S01 @ANi >> S02 EP-30");
    expect((await reresolveEntityEpisodes(h.ctx, h.db, entityId)).releases).toBe(0);
    expect(await pending(entityId)).toEqual([]);
  });

  it("摘除资源站资源不会关闭仍有未解决文件的同范围事项", async () => {
    const entityId = await series();
    await h.db
      .insertInto("subscription")
      .values({ media_entity_id: entityId, media_kind: "series", status: "active" })
      .execute();
    const r = await release(entityId, 40);
    await settleEpisodeDecision(h.ctx, h.db, {
      sourceReleaseId: r.row.id,
      entityId,
      resolution: r.resolution,
      parsed: r.parsed,
    });
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 41)],
    });
    await clearReleaseEpisodes(h.ctx, h.db, r.row.id);
    const d = (await pending(entityId))[0]!;
    expect(d).toBeDefined();
    expect(d.evidence).toMatchObject({ sourceReleaseIds: [], count: 1 });
    expect(episodeFileSamples(d.evidence)).toHaveLength(1);
  });

  it("确认智能体建议前重新核对当前文件，成功时保存文件重算结果", async () => {
    const entityId = await series();
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 40)],
    });
    const d = (await pending(entityId))[0]!;
    await simulateForAgent(
      h.db,
      { decisionId: d.id, line: "S01 @ANi >> S02 EP-30", reasons: ["文件证据"] },
      { runId: null },
    );
    const cur = await h.db.selectFrom("decision").selectAll().where("id", "=", d.id).executeTakeFirstOrThrow();
    const result = await resolveEpisodeDecision(h.ctx, cur, "confirm_agent_suggestion", null, ACTOR);
    expect(result.status).toBe("resolved");
    expect(episodeFileSamples(result.evidence)[0]!.resolution.canonical).toEqual({
      season: 2,
      episodes: { start: 10, end: 10 },
    });
  });

  it("已有同发布组资源被规则弄成待确认时拒绝写入", async () => {
    const entityId = await series();
    await release(entityId, 1);
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 40)],
    });
    const d = (await pending(entityId))[0]!;
    const simulation = await simulateForAgent(h.db, { decisionId: d.id, line: "S01 @ANi >> S02 EP-30" }, null);
    expect(simulation.summary).toMatchObject({ samples: 1, resolved: 1, regressions: 1, ok: false });
    await expect(
      applyAgentEpisodeRule(h.ctx, { decisionId: d.id, line: simulation.line, reasons: ["只核对了文件"] }, ACTOR, null),
    ).rejects.toThrow("已对上的资源");
  });

  it("首次混合整理保留已对上的文件，纯文件偏移也要通过回归保护", async () => {
    const entityId = await series();
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 1), await file(entityId, 40)],
    });
    const d = (await pending(entityId))[0]!;
    expect(episodeFileSamples(d.evidence)).toHaveLength(2);
    const simulation = await simulateForAgent(h.db, { decisionId: d.id, line: "S01 @ANi >> S02 EP-30" }, null);
    expect(simulation.summary).toMatchObject({ samples: 2, resolved: 1, regressions: 1, ok: false });
    await expect(
      applyAgentEpisodeRule(h.ctx, { decisionId: d.id, line: simulation.line, reasons: ["同批文件"] }, ACTOR, null),
    ).rejects.toThrow("已对上的文件");
    expect(
      await h.db.selectFrom("identify_words").select("id").where("media_entity_id", "=", entityId).execute(),
    ).toHaveLength(0);
  });

  it("保存部分有效的映射保留未决文件与已解决样本，不误关闭事项", async () => {
    const entityId = await series();
    const files = [await file(entityId, 40), await file(entityId, 60)];
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, { recordId: randomUUID(), entityId, files });
    await setWordsText(h.db, entityId, "S01 @ANi >> S02 EP-30");
    await reresolveEntityEpisodes(h.ctx, h.db, entityId);
    const d = (await pending(entityId))[0]!;
    expect(episodeFileSamples(d.evidence).map((f) => f.resolution.method)).toEqual(["rule", "unresolved"]);
    expect(d.evidence).toMatchObject({ count: 1 });
    expect((await inspectEpisodeMapping(h.db, { decisionId: d.id })).files).toHaveLength(2);
  });

  it("旧建议确认前必须按后来进入的文件再次模拟，失败不关闭也不写规则", async () => {
    const entityId = await series();
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 40)],
    });
    const d = (await pending(entityId))[0]!;
    await simulateForAgent(
      h.db,
      { decisionId: d.id, line: "S01 @ANi >> S02 EP-30", reasons: ["旧样本"] },
      { runId: null },
    );
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: randomUUID(),
      entityId,
      files: [await file(entityId, 60)],
    });
    const cur = (await pending(entityId))[0]!;
    expect(cur.evidence).toHaveProperty("agentSuggestion");
    await expect(resolveEpisodeDecision(h.ctx, cur, "confirm_agent_suggestion", null, ACTOR)).rejects.toThrow("文件");
    expect((await pending(entityId))[0]!.status).toBe("open");
    expect(
      await h.db.selectFrom("identify_words").select("id").where("media_entity_id", "=", entityId).execute(),
    ).toEqual([]);
  });

  it("覆盖同一记录样本时移除旧范围，保留其他整理记录的未决文件", async () => {
    const entityId = await series();
    const recordA = randomUUID();
    const recordB = randomUUID();
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: recordA,
      entityId,
      files: [await file(entityId, 40)],
    });
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, {
      recordId: recordB,
      entityId,
      files: [await file(entityId, 41)],
    });
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, { recordId: recordA, entityId, files: [] });
    const d = (await pending(entityId))[0]!;
    expect(episodeFileSamples(d.evidence).map((f) => f.recordId)).toEqual([recordB]);
    await settleOrganizeEpisodeDecisions(h.ctx, h.db, { recordId: recordB, entityId, files: [] });
    expect(await pending(entityId)).toEqual([]);
  });
});