feat(search): 按作品搜资源时判定每条结果是否属于这部作品
对照 MP TorrentHelper.match_torrent：已识别到同一本地作品（或同一 TMDB 身份）的属于、识别到别的作品的不属于； 未识别的先排除音乐 / 游戏 / 软件 / 书分类，再核对类型、年份（剧集按各季首播年份 ±1）、季集范围（TMDB 季列表， 单季集号可按全部季总集数的绝对集号），最后按解析出的中 / 英文名、标题非英文词、副标题非英文词比对作品名、原名、别名。 不属于的不参与筛选与分面，响应里单独给出（unmatched / unmatchedTotal，只在第一页，最多 200 条）， 页面在列表最后折叠为「另有 N 条可能不是这部作品」，展开后仍可下载。关键词搜索不做这一步。 用「药屋少女的呢喃」109 条真实结果做样本：只有《迷宫饭》片尾曲专辑和标题写成 S01-S24 的一条被判为不属于。 Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01Cxwgnn2pbzQ7rXv3rq7fei
This commit is contained in:
rootandClaude Opus 5.5 committed
1 parent
f6d976a963
commit
9d7e20067b
9 files changed
+1560
-10
No files matched your search
@@ -100,7 +100,14 @@ describe.skipIf(!up)("两阶段搜索(SEARCH-2)", () => {
|
||||
snatched: "10",
|
||||
uploadedAt: "2026-09-20 10:00:00",
|
||||
};
|
||||
a = await createMockNexusPhpSite({ releases: [...AUDIENCES_FIXTURE_RELEASES, release] });
|
||||
// 按作品搜时的归属判定:同名蓝光(属于)、剧集(作品是电影)、1984 年的前作、名字对不上的(识别后才算属于)
|
||||
const nearby = [
|
||||
{ ...release, id: "700002", title: "Dune.Part.Two.2024.1080p.BluRay.x264-GRP2", subtitle: "沙丘2 | 简繁字幕" },
|
||||
{ ...release, id: "700003", title: "Dune Prophecy S01 2024 1080p WEB-DL H264-TV", subtitle: "沙丘2 衍生剧" },
|
||||
{ ...release, id: "700004", title: "Dune 1984 1080p BluRay x264-OLD", subtitle: "沙丘2 前作" },
|
||||
{ ...release, id: "700005", title: "Arrakis Chronicles 2024 1080p WEB-DL H264-ZZ", subtitle: "沙丘2特别版" },
|
||||
];
|
||||
a = await createMockNexusPhpSite({ releases: [...AUDIENCES_FIXTURE_RELEASES, release, ...nearby] });
|
||||
b = await createMockNexusPhpSite({ releases: [{ ...release, id: "800001" }] });
|
||||
tmdb = await createMockTmdb();
|
||||
|
||||
@@ -316,6 +323,59 @@ describe.skipIf(!up)("两阶段搜索(SEARCH-2)", () => {
|
||||
expect((await req(`/search/releases?tmdb=movie:999999999`)).status).toBe(404);
|
||||
});
|
||||
|
||||
it("按作品搜:不属于这部作品的不参与筛选与分面,第一页单独给出;已识别的按识别结果;关键词搜索不判定", async () => {
|
||||
const ids = (xs: Loose[]) => xs.map((i) => i.siteReleaseId).sort();
|
||||
const page = async (extra = "") =>
|
||||
(await req(`/search/releases?entityId=${dune}&sites=${sites.a}${extra}`)).body.data;
|
||||
let d = await page();
|
||||
expect(ids(d.items)).toEqual(["700001", "700002", "812345"]);
|
||||
expect(ids(d.unmatched)).toEqual(["700003", "700004", "700005"]);
|
||||
expect(d).toMatchObject({ total: 3, unfilteredTotal: 3, unmatchedTotal: 3 });
|
||||
// 分面只算属于的:没有 1984 年、没有季集
|
||||
expect(d.facets.year.map((f: Loose) => f.value)).toEqual(["2024"]);
|
||||
expect(d.facets.seasonEpisode).toEqual([]);
|
||||
expect(d.facets.site).toEqual([{ value: sites.a, label: "观众", count: 3 }]);
|
||||
// 筛选只作用于属于的;「可能不是」那组不受影响
|
||||
d = await page(`&qf=${encodeURIComponent("resolution:2160p")}`);
|
||||
expect(ids(d.items)).toEqual(["700001", "812345"]);
|
||||
expect(d.unmatchedTotal).toBe(3);
|
||||
// 只在第一页给出
|
||||
const p1 = await page("&limit=1");
|
||||
expect(p1.unmatched).toHaveLength(3);
|
||||
const p2 = await page(`&limit=1&cursor=${p1.nextCursor}`);
|
||||
expect(p2.unmatched).toEqual([]);
|
||||
expect(p2.unmatchedTotal).toBe(3);
|
||||
|
||||
// 识别结果优先:名字对不上的识别到这部作品后属于;名字对得上的识别到别的作品后不属于
|
||||
const other = (
|
||||
await db
|
||||
.insertInto("media_entity")
|
||||
.values({ kind: "movie", title: "沙丘", original_title: "Dune", year: 1984, status: "active" })
|
||||
.returning("id")
|
||||
.executeTakeFirstOrThrow()
|
||||
).id;
|
||||
const resolve = (id: string, entity: string) =>
|
||||
db
|
||||
.updateTable("source_release")
|
||||
.set({ media_entity_id: entity })
|
||||
.where("site_id", "=", sites.a)
|
||||
.where("site_release_id", "=", id)
|
||||
.execute();
|
||||
await resolve("700005", dune);
|
||||
await resolve("700002", other);
|
||||
d = await page();
|
||||
expect(ids(d.items)).toEqual(["700001", "700005", "812345"]);
|
||||
expect(ids(d.unmatched)).toEqual(["700002", "700003", "700004"]);
|
||||
// 按 TMDB 条目搜同一作品:同样判定
|
||||
const byTmdb = (await req(`/search/releases?tmdb=movie:693134&sites=${sites.a}`)).body.data;
|
||||
expect(ids(byTmdb.items)).toEqual(["700001", "700005", "812345"]);
|
||||
|
||||
// 关键词搜索不判定归属
|
||||
const kw = (await req(`/search/releases?q=${encodeURIComponent("沙丘2")}&sites=${sites.a}`)).body.data;
|
||||
expect(kw.unmatched).toBeUndefined();
|
||||
expect(kw.total).toBe(6);
|
||||
});
|
||||
|
||||
it("GET /media/tmdb/{type}/{id}:TMDB 信息 + 季列表 + 库存对比 + 是否已建档", async () => {
|
||||
const before = counts();
|
||||
const movie = await req(`/media/tmdb/movie/872585`);
|
||||
|
||||
@@ -0,0 +1,190 @@
|
||||
import type { ReleaseSearchQuery } from "@umt/contracts";
|
||||
import { normalizeTitle, parseRelease } from "@umt/domain";
|
||||
import { describe, expect, it } from "vitest";
|
||||
import { type BaseRelease, quickFilter } from "../../src/modules/search/releases.ts";
|
||||
import { belongsToWork, type ReleaseLink, type WorkTarget } from "../../src/modules/search/work-match.ts";
|
||||
import samples from "../fixtures/search-apothecary.json" with { type: "json" };
|
||||
|
||||
// 按作品搜资源的归属判定(对照 MP TorrentHelper.match_torrent)。样本是站点搜索「药屋少女的呢喃」的真实结果(109 条),
|
||||
// 按「刚搜到、还没识别」处理:只靠名字 / 类型 / 年份 / 季集判定。
|
||||
|
||||
const base = (o: { id: string; title: string; subtitle: string | null; category: string | null }): BaseRelease => {
|
||||
const p = parseRelease(o.title, o.subtitle ? { subtitle: o.subtitle } : {});
|
||||
return {
|
||||
key: `s:${o.id}`,
|
||||
sourceReleaseId: null,
|
||||
siteId: "00000000-0000-4000-8000-000000000001",
|
||||
siteName: "站",
|
||||
siteReleaseId: o.id,
|
||||
title: o.title,
|
||||
subtitle: o.subtitle,
|
||||
category: { id: null, name: o.category },
|
||||
sizeBytes: null,
|
||||
seeders: 1,
|
||||
leechers: 0,
|
||||
snatched: 0,
|
||||
promotion: { type: "none", downloadFactor: 1, uploadFactor: 1, expiresAt: null },
|
||||
free: false,
|
||||
hr: "unknown",
|
||||
uploadedAt: null,
|
||||
detailUrl: "http://x/",
|
||||
observedAt: "2026-10-06T00:00:00.000Z",
|
||||
parsed: {
|
||||
titleCn: p.titleCn ?? null,
|
||||
titleEn: p.titleEn ?? null,
|
||||
year: p.year ?? null,
|
||||
season: p.season ?? null,
|
||||
seasonEnd: p.seasonEnd ?? null,
|
||||
episodes: p.episodes ? { start: p.episodes.start, end: p.episodes.end } : null,
|
||||
resolution: p.resolution ?? null,
|
||||
source: p.source ?? null,
|
||||
videoCodec: p.videoCodec ?? null,
|
||||
hdr: [...p.hdr],
|
||||
releaseGroup: p.releaseGroup?.name ?? null,
|
||||
audio: [...p.audio],
|
||||
},
|
||||
};
|
||||
};
|
||||
|
||||
const set = (xs: string[]) => new Set(xs.map(normalizeTitle));
|
||||
// 本地作品与别名表(取自真实库里这部作品的 TMDB 别名)
|
||||
const apothecary: WorkTarget = {
|
||||
entityId: "e-apothecary",
|
||||
tmdb: "tv:220542",
|
||||
movie: false,
|
||||
titles: set(["药屋少女的呢喃", "薬屋のひとりごと"]),
|
||||
names: set([
|
||||
"The Apothecary Diaries",
|
||||
"药师少女的独语",
|
||||
"Kusuriya no Hitorigoto",
|
||||
"Kusuriya no Hitorigoto 2nd Season",
|
||||
"薬屋のひとりごと(第3期)",
|
||||
"Apothecary's Soliloquy",
|
||||
"Les Carnets de l'apothicaire",
|
||||
]),
|
||||
year: 2023,
|
||||
seasons: [
|
||||
{ season: 1, year: 2023, episodes: 24 },
|
||||
{ season: 2, year: 2025, episodes: 24 },
|
||||
{ season: 3, year: 2026, episodes: null },
|
||||
],
|
||||
};
|
||||
|
||||
const rows = samples.map(base);
|
||||
const byId = (id: string) => rows.find((r) => r.siteReleaseId === id)!;
|
||||
const ALBUM = "926704"; // Channel U(《迷宫饭》片尾曲专辑,简介里提到这部作品的片头曲)
|
||||
const S01_S24 = "1012256"; // 标题写成 S01-S24(实为 24 集),超出作品的季范围
|
||||
const out = (t: WorkTarget, links: Map<string, ReleaseLink> = new Map()) =>
|
||||
rows.filter((r) => !belongsToWork(t, r, links.get(r.key) ?? null)).map((r) => r.siteReleaseId);
|
||||
|
||||
describe("按作品判定归属(未识别,真实样本)", () => {
|
||||
it("109 条里只有专辑与季范围不可能的那条不属于;其余(S01 / S02 / S03、Vol 原盘、无年份标题)都属于", () => {
|
||||
expect(samples).toHaveLength(109);
|
||||
expect(out(apothecary).sort()).toEqual([S01_S24, ALBUM].sort());
|
||||
// 逐类抽查不被误伤
|
||||
for (const [id, why] of [
|
||||
["1139946", "S01E25-E48 绝对集号"],
|
||||
["728125", "S03E01"],
|
||||
["954567", "S02 自购自抓原盘,中文名解析成「自购自抓自扫」,靠英文名 / 副标题"],
|
||||
["1021945", "Vol.3 BDMV,英文名带 Vol 3、没识别"],
|
||||
["768303", "Vol 2 原盘"],
|
||||
["994487", "没有年份没有季号(VCB BDRip)"],
|
||||
["1157450", "S02 2026(蓝光发行年份)"],
|
||||
] as const) {
|
||||
const r = byId(id);
|
||||
expect(r, why).toBeTruthy();
|
||||
expect(belongsToWork(apothecary, r, null), `${why}:${r.title}`).toBe(true);
|
||||
}
|
||||
});
|
||||
|
||||
it("没拿到 TMDB 季信息时不核对季集范围,只排除专辑", () => {
|
||||
expect(out({ ...apothecary, seasons: null })).toEqual([ALBUM]);
|
||||
});
|
||||
|
||||
it("专辑没有分类也靠名字排除:简介里『』括起的作品名不算命中", () => {
|
||||
const album = { ...byId(ALBUM), category: { id: null, name: null } };
|
||||
expect(belongsToWork(apothecary, album, null)).toBe(false);
|
||||
});
|
||||
|
||||
it("已识别:同一作品或同一 TMDB 身份属于;识别到别的作品不属于(哪怕名字对得上)", () => {
|
||||
const album = byId(ALBUM);
|
||||
const s1 = rows[0]!;
|
||||
expect(belongsToWork(apothecary, album, { entityId: "e-dungeon", tmdb: "tv:209867" })).toBe(false);
|
||||
expect(belongsToWork(apothecary, s1, { entityId: "e-dungeon", tmdb: null })).toBe(false);
|
||||
expect(belongsToWork(apothecary, s1, { entityId: "e-apothecary", tmdb: null })).toBe(true);
|
||||
expect(belongsToWork(apothecary, byId(S01_S24), { entityId: "e-apothecary", tmdb: null })).toBe(true);
|
||||
// 本地有重复建档,但 TMDB 身份相同
|
||||
expect(belongsToWork(apothecary, s1, { entityId: "e-dup", tmdb: "tv:220542" })).toBe(true);
|
||||
// 只在 TMDB 上的作品:按 TMDB 身份比
|
||||
const tmdbOnly = { ...apothecary, entityId: null };
|
||||
expect(belongsToWork(tmdbOnly, s1, { entityId: "e-any", tmdb: "tv:220542" })).toBe(true);
|
||||
expect(belongsToWork(tmdbOnly, s1, { entityId: "e-any", tmdb: null })).toBe(false);
|
||||
});
|
||||
|
||||
it("不属于的不进筛选与分面:原盘只剩这部作品的", () => {
|
||||
const mine = rows.filter((r) => belongsToWork(apothecary, r, null));
|
||||
const q = { sort: "default", order: "desc", limit: 50, qf: "releaseType:disc" } as ReleaseSearchQuery;
|
||||
const all = quickFilter(rows, q, Date.now());
|
||||
const kept = quickFilter(mine, q, Date.now());
|
||||
expect(all.items.map((r) => r.siteReleaseId)).toContain(ALBUM);
|
||||
expect(kept.items.map((r) => r.siteReleaseId)).not.toContain(ALBUM);
|
||||
// 「自购自抓自扫」那条按原盘计
|
||||
expect(kept.items.map((r) => r.siteReleaseId)).toContain("954567");
|
||||
expect(kept.items).toHaveLength(all.items.length - 1);
|
||||
});
|
||||
});
|
||||
|
||||
describe("按作品判定归属(规则)", () => {
|
||||
const mk = (title: string, subtitle: string | null = null, category: string | null = null) =>
|
||||
base({ id: title, title, subtitle, category });
|
||||
|
||||
it("别的作品:名字对不上的不属于(迷宫饭)", () => {
|
||||
expect(
|
||||
belongsToWork(apothecary, mk("Delicious in Dungeon S01 2024 1080p NF WEB-DL", "迷宫饭 / ダンジョン飯"), null),
|
||||
).toBe(false);
|
||||
});
|
||||
|
||||
it("剧集:年份与任一季首播相差 1 年以内;季号超出范围不属于;集号可按绝对集号", () => {
|
||||
expect(belongsToWork(apothecary, mk("Kusuriya no Hitorigoto S01 2019 1080p WEB-DL"), null)).toBe(false);
|
||||
expect(belongsToWork(apothecary, mk("Kusuriya no Hitorigoto S01 2024 1080p BluRay x265"), null)).toBe(true);
|
||||
expect(belongsToWork(apothecary, mk("Kusuriya no Hitorigoto S05 1080p WEB-DL"), null)).toBe(false);
|
||||
expect(belongsToWork(apothecary, mk("Kusuriya no Hitorigoto S01E30 1080p WEB-DL"), null)).toBe(true);
|
||||
expect(belongsToWork(apothecary, mk("Kusuriya no Hitorigoto S01E60 1080p WEB-DL"), null)).toBe(false);
|
||||
// 没有季信息时,只排除早于首播一年以上的
|
||||
const noSeasons = { ...apothecary, seasons: null };
|
||||
expect(belongsToWork(noSeasons, mk("Kusuriya no Hitorigoto 2030 1080p WEB-DL"), null)).toBe(true);
|
||||
expect(belongsToWork(noSeasons, mk("Kusuriya no Hitorigoto 2019 1080p WEB-DL"), null)).toBe(false);
|
||||
});
|
||||
|
||||
it("名字:标题拆词只比非英文词与作品名 / 原名;副标题拆词也比别名;英文单词不拆着比", () => {
|
||||
// 标题里的中文名
|
||||
expect(belongsToWork(apothecary, mk("[药屋少女的呢喃][S02E05][1080p]"), null)).toBe(true);
|
||||
// 副标题里的别名
|
||||
expect(belongsToWork(apothecary, mk("KnH 2023 S01 1080p WEB-DL", "药师少女的独语 | 全24集"), null)).toBe(true);
|
||||
// 英文单词片段不算
|
||||
expect(belongsToWork(apothecary, mk("Diaries of a Pharmacist S01 2023 1080p WEB-DL", "The Diaries"), null)).toBe(
|
||||
false,
|
||||
);
|
||||
// 非音视频分类
|
||||
expect(belongsToWork(apothecary, mk("Kusuriya no Hitorigoto OST 2023 FLAC", null, "音乐"), null)).toBe(false);
|
||||
});
|
||||
|
||||
it("电影:年份上下 1 年,没写年份不属于(MP 同);剧集资源或剧集分类不属于电影", () => {
|
||||
const dune: WorkTarget = {
|
||||
entityId: "e-dune",
|
||||
tmdb: "movie:693134",
|
||||
movie: true,
|
||||
titles: set(["沙丘2", "Dune: Part Two"]),
|
||||
names: set(["Dune 2", "沙丘:第二部"]),
|
||||
year: 2024,
|
||||
seasons: null,
|
||||
};
|
||||
expect(belongsToWork(dune, mk("Dune.Part.Two.2024.2160p.WEB-DL"), null)).toBe(true);
|
||||
expect(belongsToWork(dune, mk("Dune.Part.Two.2025.2160p.BluRay"), null)).toBe(true);
|
||||
expect(belongsToWork(dune, mk("Dune.Part.Two.2160p.WEB-DL"), null)).toBe(false);
|
||||
expect(belongsToWork(dune, mk("Dune 1984 1080p BluRay", "沙丘2 前作"), null)).toBe(false);
|
||||
expect(belongsToWork(dune, mk("Dune Part Two S01 2024 1080p WEB-DL"), null)).toBe(false);
|
||||
expect(belongsToWork(dune, mk("Dune.Part.Two.2024.1080p.WEB-DL", null, "剧集"), null)).toBe(false);
|
||||
expect(belongsToWork(dune, mk("Some Movie 2024 1080p WEB-DL", "沙丘2"), null)).toBe(true);
|
||||
});
|
||||
});
|
||||
Reference in new issue
Block a user