UHD_Media_Tool/apps/server/test/sources-identity/migration-identify-words.test.ts

import { compileEntityWords } from "@umt/contracts";
import { createMigrator, type DB, migrateToLatest, sql } from "@umt/db";
import { Kysely, PostgresDialect } from "kysely";
import pg from "pg";
import { afterAll, describe, expect, it } from "vitest";
import { up } from "../../../../packages/db/migrations/0094_identify_words.ts";
import { PG_URL, pgAvailable, withSchemaLock } from "../auth-settings/harness.ts";

// 0094:episode_mapping_rule 结构化规则逐作品转成作品识别词文本行后删表。部署库只有一部作品 9 条规则
// (药屋少女的呢喃:S2 任意组 EP+24 → S1;S3 任意组 EP+48 → S1;S3 七个发布组 EP → S1),按它造数据。

const hasPg = await pgAvailable();

describe.skipIf(!hasPg)("0094 映射规则迁成作品识别词", () => {
  const schema = `iw_mig_${process.pid.toString(36)}${Math.random().toString(36).slice(2, 6)}`;
  const admin = async (q: string) => {
    const c = new pg.Client({ connectionString: PG_URL });
    await c.connect();
    await c.query(q).finally(() => c.end());
  };
  let db: Kysely<DB>;

  afterAll(async () => {
    await db?.destroy();
    await withSchemaLock(() => admin(`drop schema if exists ${schema} cascade`));
  });

  // biome-ignore lint/suspicious/noExplicitAny: 迁移前的旧表不在类型里
  const raw = () => db as unknown as Kysely<any>;
  const rule = (entity: string, v: Record<string, unknown>) =>
    raw()
      .insertInto("episode_mapping_rule")
      .values({ media_entity_id: entity, kind: "expression", created_by_type: "user", ...v })
      .execute();
  const groups = ["ADWeb", "MWeb", "CHDWEB", "黒ネズミたち", "Judas", "绿茶字幕组", "綠茶字幕組"];

  it("药屋 9 条规则合成 3 行;停用的变注释;单集对照一行一条;智能体的带注释;重跑只补缺的行", async () => {
    await admin(`create schema ${schema}`);
    db = new Kysely<DB>({
      dialect: new PostgresDialect({
        pool: new pg.Pool({ connectionString: PG_URL, max: 2, options: `-c search_path=${schema},public` }),
      }),
    });
    await withSchemaLock(async () => {
      const r = await createMigrator(db, { migrationTableSchema: schema }).migrateTo("0093_pi_azure_provider");
      if (r.error) throw r.error;
    });
    const entity = async (title: string) =>
      (
        await sql<{
          id: string;
        }>`insert into media_entity (kind, title) values ('series', ${title}) returning id`.execute(db)
      ).rows[0]!.id;
    const kusuriya = await entity("药屋少女的呢喃");
    const other = await entity("别的剧");
    await rule(kusuriya, { season: 2, expression: "EP+24", target_season: 1 });
    await rule(kusuriya, { season: 3, expression: "EP+48", target_season: 1 });
    for (const g of groups) await rule(kusuriya, { season: 3, release_group: g, expression: "EP", target_season: 1 });
    // 别的剧:停用(自动停用留下的)、单集对照、特别篇、智能体写的
    await rule(other, {
      season: 1,
      release_group: "ANi",
      expression: "EP-12",
      target_season: 2,
      status: "disabled",
      disabled_reason: JSON.stringify({ source: { season: 1 } }),
    });
    await rule(other, {
      season: 1,
      release_group: "Baha",
      kind: "table",
      expression: null,
      table_entries: JSON.stringify([
        { from: { start: 13, end: 13 }, to: { season: 0, episode: 1 } },
        { from: { start: 27, end: 28 }, to: { season: 2, episode: 2 } },
      ]),
    });
    await rule(other, { season: null, release_group: "SweetSub", kind: "special", expression: null, special: "extra" });
    await rule(other, {
      season: 2,
      release_group: "Nekomoe",
      expression: "EP-30",
      created_by_type: "agent",
      evidence: JSON.stringify({ reasons: ["发布时间对上播出日期"] }),
    });

    await withSchemaLock(() => migrateToLatest(db, { migrationTableSchema: schema }));

    const text = async (id: string) =>
      (await db.selectFrom("identify_words").select("text").where("media_entity_id", "=", id).executeTakeFirstOrThrow())
        .text;
    expect(await text(kusuriya)).toBe(
      ["S02 >> S01 EP+24", "S03 >> S01 EP+48", `S03 ${groups.map((g) => `@${g}`).join(" ")} >> S01 EP`, ""].join("\n"),
    );
    expect(await text(other)).toBe(
      [
        "S01E13 @Baha >> S00E01",
        "S01E27-E28 @Baha >> S02E02",
        "# S01 @ANi >> S02 EP-12",
        "# 智能体：发布时间对上播出日期",
        "S02 @Nekomoe >> EP-30",
        "* @SweetSub >> 花絮",
        "",
      ].join("\n"),
    );
    // 迁出来的文本能原样编译,规则与原来等价(药屋 2 + 7 条)
    const c = compileEntityWords(await text(kusuriya));
    expect(c.errors).toEqual([]);
    expect(c.rules.map((r) => [r.season, r.releaseGroup, r.expression, r.targetSeason])).toEqual([
      [2, null, "EP+24", 1],
      [3, null, "EP+48", 1],
      ...groups.map((g) => [3, g, "EP", 1]),
    ]);
    expect(compileEntityWords(await text(other)).errors).toEqual([]);
    const tables = await sql<{
      t: string | null;
    }>`select to_regclass(${`${schema}.episode_mapping_rule`})::text as t`.execute(db);
    expect(tables.rows[0]!.t).toBeNull();
    const version = await db
      .selectFrom("identify_words")
      .select(["version", "updated_by_type"])
      .where("media_entity_id", "=", kusuriya)
      .executeTakeFirstOrThrow();
    expect(version).toEqual({ version: 1, updated_by_type: "system" });

    // 重跑(表还在的情况):已有的行不重复,只追加新的
    await sql
      .raw(`create table episode_mapping_rule (
        id uuid primary key default gen_random_uuid(), media_entity_id uuid not null, season integer, release_group text,
        kind text not null, expression text, target_season integer, table_entries jsonb, special text,
        status text not null default 'active', evidence jsonb not null default '{}'::jsonb, created_by_type text not null,
        created_at timestamptz not null default now())`)
      .execute(db);
    await rule(kusuriya, { season: 2, expression: "EP+24", target_season: 1 });
    await rule(kusuriya, { season: 4, expression: "EP+72", target_season: 1 });
    await up(raw());
    expect(await text(kusuriya)).toBe(
      [
        "S02 >> S01 EP+24",
        "S03 >> S01 EP+48",
        `S03 ${groups.map((g) => `@${g}`).join(" ")} >> S01 EP`,
        "S04 >> S01 EP+72",
        "",
      ].join("\n"),
    );
  });
});