From 387f7931e005f1928244748fdd63ac69613c8614 Mon Sep 17 00:00:00 2001 From: root Date: Tue, 29 Sep 2026 17:41:22 +0800 Subject: [PATCH] feat(sources): MP-aligned TMDB recall, alias registration guard, re-resolve on retitle (ID-2) recallTmdbCandidates searches cn then en names in the preferred language by type and fetches details for a bounded set of candidates (MP alias fallback); translations become aliases; bindAuto no longer registers generic or alias-only site titles; snapshot/feed title changes reparse and force a re-resolve that unbinds when it can no longer confirm. Co-Authored-By: Claude Opus 5.5 --- .../sources-identity/service/resolve.ts | 57 ++++++-------- .../sources-identity/service/snapshot.ts | 36 ++++++++- .../service/tmdb-candidate.ts | 51 +++++++++++++ .../src/modules/sources-identity/tasks.ts | 8 +- .../test/sources-identity/resolve.test.ts | 7 +- .../test/sources-identity/retitle.test.ts | 75 +++++++++++++++++++ packages/adapters/src/tmdb/README.md | 2 +- packages/adapters/src/tmdb/index.ts | 1 + packages/adapters/src/tmdb/recall.ts | 71 ++++++++++++++++++ 9 files changed, 269 insertions(+), 39 deletions(-) create mode 100644 apps/server/src/modules/sources-identity/service/tmdb-candidate.ts create mode 100644 apps/server/test/sources-identity/retitle.test.ts create mode 100644 packages/adapters/src/tmdb/recall.ts diff --git a/apps/server/src/modules/sources-identity/service/resolve.ts b/apps/server/src/modules/sources-identity/service/resolve.ts index 64c1dc4f..70263a47 100644 --- a/apps/server/src/modules/sources-identity/service/resolve.ts +++ b/apps/server/src/modules/sources-identity/service/resolve.ts @@ -1,10 +1,11 @@ import { createHash } from "node:crypto"; -import { type MetadataCandidate, TmdbError } from "@umt/adapters"; +import { type MetadataCandidate, recallTmdbCandidates, TmdbError } from "@umt/adapters"; import type { EvidenceLevel, PoolCandidate } from "@umt/contracts"; import { type SourceReleaseRow, sql } from "@umt/db"; import { type IdentityCandidate as DomainCandidate, type IdentityResolution, + isGenericAliasName, normalizeTitle, PARSER_VERSION, type ParsedRelease, @@ -35,6 +36,7 @@ import { resolveEntityId, } from "./entities.ts"; import { detachFromIdentityDecisions, openGroupDecision, refreshGroupDecision } from "./identity-decisions.ts"; +import { tmdbRecallInput, tmdbToCandidate } from "./tmdb-candidate.ts"; import { loadRecognizeCache, type RecognizeMeta, recognizeMetaOf, saveRecognizeCache } from "./tmdb-recognize-cache.ts"; // resolveObservation:parseRelease → 本地召回(别名 pg_trgm + 已有外部身份 + 人工锁)→ [按需] TMDB 候选 → domain.resolveIdentity。 @@ -97,26 +99,7 @@ export function evidenceVersion(parsed: ParsedRelease | null, ranked: RankedCand return `p${PARSER_VERSION}:${h}`; } -function yearOf(d: string | null | undefined): number | undefined { - return d && /^\d{4}/.test(d) ? Number(d.slice(0, 4)) : undefined; -} - -export function tmdbToCandidate(m: MetadataCandidate): DomainCandidate { - const titles = [m.title, ...(m.originalTitle ? [m.originalTitle] : [])]; - for (const t of m.titles) if (t.kind !== "alias" && !titles.includes(t.text)) titles.push(t.text); - const aliases = [...new Set([...m.aliases, ...m.titles.filter((t) => t.kind === "alias").map((t) => t.text)])]; - const seasonYears = m.seasons?.map((s) => yearOf(s.airDate)).filter((y): y is number => y !== undefined); - return { - provider: "tmdb", - entityType: m.entityType, - externalId: m.externalId, - titles, - ...(m.year !== null ? { year: m.year } : {}), - aliases, - ...(seasonYears?.length ? { seasonYears } : {}), - ...(m.externalIds?.imdb ? { linkedExternalIds: [{ provider: "imdb", id: m.externalIds.imdb }] } : {}), - }; -} +export { tmdbToCandidate }; function poolCandidate(r: RankedCandidate): PoolCandidate { const c = r.candidate; @@ -208,15 +191,8 @@ async function tmdbCandidates( return { remote: [], cacheHit: true, meta }; } } - const queries = [parsed.titleCn, parsed.titleEn, parsed.title].filter((t): t is string => !!t); - const seen = new Map(); - for (const q of [...new Set(queries)].slice(0, 2)) { - for (const m of await tmdb.searchAll(q)) { - const k = `${m.entityType}:${m.externalId}`; - if (!seen.has(k)) seen.set(k, m); - } - } - return { remote: [...seen.values()].slice(0, 20), cacheHit: false, meta }; + const remote = await recallTmdbCandidates(tmdb, tmdbRecallInput(parsed)); + return { remote, cacheHit: false, meta }; } async function tmdbByExplicitIds( @@ -512,6 +488,11 @@ function rankedSummary(r: RankedCandidate | undefined) { return r ? { ...poolCandidate(r), titleMatch: r.titleMatch, yearMatch: r.yearMatch, typeMatch: r.typeMatch } : null; } +/** 自动绑定后能否把站点标题登记为别名:标题精确命中,或别名命中且有第二种文字的名称佐证 */ +export function siteAliasAllowed(best: RankedCandidate): boolean { + return best.titleMatch === "exact" || (best.titleMatch === "alias" && best.signals.nameLangs > 1); +} + async function bindAuto( rt: SourcesRuntime, release: SourceReleaseRow, @@ -563,9 +544,17 @@ async function bindAuto( .where("id", "=", entityId) .executeTakeFirstOrThrow(); const units = await ensureUnits(trx, entityId, entity.kind, parsed); - // 站点标题作为别名:下次同名资源可直接本地召回 - if (parsed.titleCn) await addAliasIfAbsent(trx, entityId, parsed.titleCn, "site_title", "site", "zh"); - if (parsed.titleEn) await addAliasIfAbsent(trx, entityId, parsed.titleEn, "site_title", "site", "en"); + // 站点标题作为别名:下次同名资源可直接本地召回。只在名称本身命中过(不是仅靠别名单线命中)时登记, + // 且通用词(分类词 / 单字 / 停用词)不登记,避免错误别名链式扩散(ID-2) + if (siteAliasAllowed(best)) { + for (const [t, lang] of [ + [parsed.titleCn, "zh"], + [parsed.titleEn, "en"], + ] as const) { + if (t && !isGenericAliasName(normalizeTitle(t))) + await addAliasIfAbsent(trx, entityId, t, "site_title", "site", lang); + } + } const evidence = { level, @@ -784,6 +773,8 @@ async function toPool( .set({ parser_version: cols.parserVersion, parsed: jsonb(cols.parsed), + // 重新识别(标题更新 / 复查)后不再能确认:撤销之前的自动绑定(人工锁定的不走到这里) + ...(release.media_entity_id ? { media_entity_id: null, content_unit_id: null } : {}), resolution_status: toReleaseResolutionStatus(contractStatus), resolution_evidence: jsonb({ reason, diff --git a/apps/server/src/modules/sources-identity/service/snapshot.ts b/apps/server/src/modules/sources-identity/service/snapshot.ts index 9c434bf9..91358cc7 100644 --- a/apps/server/src/modules/sources-identity/service/snapshot.ts +++ b/apps/server/src/modules/sources-identity/service/snapshot.ts @@ -1,3 +1,4 @@ +import { createHash } from "node:crypto"; import { type SiteDetailPage, SiteError, type SitePage, type SiteRelease } from "@umt/adapters"; import type { SiteSnapshotKind } from "@umt/contracts"; import { type SiteRow, sql } from "@umt/db"; @@ -41,7 +42,14 @@ export async function upsertSiteRelease( const parsed = parseRelease(r.title, r.subtitle ? { subtitle: r.subtitle } : {}); const c = siteReleaseToSourceRelease(r, { siteId: site.id, parsed }); const row = ( - await sql<{ id: string; inserted: boolean }>` + await sql<{ + id: string; + inserted: boolean; + resolution_locked: boolean; + prev_title: string | null; + prev_subtitle: string | null; + }>` + with prev as (select title, subtitle from source_release where site_id = ${site.id} and site_release_id = ${c.siteReleaseId}) insert into source_release (site_id, site_release_id, title, subtitle, category_raw, tags_raw, size_bytes, infohash_v1, infohash_v2, promotion, promotion_expires_at, seeders, leechers, source_published_at, parsed, parser_version, raw, detail_fetched_at) @@ -51,6 +59,12 @@ export async function upsertSiteRelease( ${jsonb(c.parsed ?? {})}::jsonb, ${c.parserVersion ?? null}, ${jsonb(c.raw ?? {})}::jsonb, ${null}) on conflict (site_id, site_release_id) do update set title = excluded.title, subtitle = coalesce(excluded.subtitle, source_release.subtitle), + parsed = case when source_release.title is distinct from excluded.title + or coalesce(excluded.subtitle, source_release.subtitle) is distinct from source_release.subtitle + then excluded.parsed else source_release.parsed end, + parser_version = case when source_release.title is distinct from excluded.title + or coalesce(excluded.subtitle, source_release.subtitle) is distinct from source_release.subtitle + then excluded.parser_version else source_release.parser_version end, category_raw = coalesce(excluded.category_raw, source_release.category_raw), tags_raw = excluded.tags_raw, size_bytes = coalesce(excluded.size_bytes, source_release.size_bytes), infohash_v1 = coalesce(excluded.infohash_v1, source_release.infohash_v1), @@ -60,7 +74,8 @@ export async function upsertSiteRelease( source_published_at = coalesce(source_release.source_published_at, excluded.source_published_at), raw = source_release.raw || excluded.raw, last_seen_at = now(), availability = case when source_release.availability = 'expired' then 'active' else source_release.availability end - returning id, (xmax = 0) as inserted`.execute(trx) + returning id, (xmax = 0) as inserted, resolution_locked, + (select p.title from prev p) as prev_title, (select p.subtitle from prev p) as prev_subtitle`.execute(trx) ).rows[0]!; if (row.inserted) { await rt.ctx.outbox.emit( @@ -69,8 +84,23 @@ export async function upsertSiteRelease( { sourceReleaseId: row.id, observationId: null, siteId: site.id, isNew: true }, { idempotencyKey: `release.observed:${row.id}` }, ); + } else if ( + !row.resolution_locked && + (row.prev_title !== c.title || (c.subtitle != null && row.prev_subtitle !== c.subtitle)) + ) { + // 标题 / 副标题更新(RSS 标题 → 列表标题等):按新标题重新解析并重新识别,不沿用旧解析(ID-2) + const digest = createHash("sha1") + .update(`${c.title}\n${c.subtitle ?? ""}`) + .digest("hex") + .slice(0, 16); + await rt.ctx.outbox.emit( + trx, + "release.observed", + { sourceReleaseId: row.id, observationId: null, siteId: site.id, isNew: false }, + { idempotencyKey: `release.observed:${row.id}:${digest}` }, + ); } - return row; + return { id: row.id, inserted: row.inserted }; } /** 凭据失效:账号标失效 + 站点健康 failing + 待处理(不重试) */ diff --git a/apps/server/src/modules/sources-identity/service/tmdb-candidate.ts b/apps/server/src/modules/sources-identity/service/tmdb-candidate.ts new file mode 100644 index 00000000..15c997fb --- /dev/null +++ b/apps/server/src/modules/sources-identity/service/tmdb-candidate.ts @@ -0,0 +1,51 @@ +import type { MetadataCandidate } from "@umt/adapters"; +import { type IdentityCandidate, isGenericAliasName, normalizeTitle, type ParsedRelease } from "@umt/domain"; + +function yearOf(d: string | null | undefined): number | undefined { + return d && /^\d{4}/.test(d) ? Number(d.slice(0, 4)) : undefined; +} + +/** + * TMDB 候选 → 身份候选。标题 = 主标题 / 原标题 / 其他语言的本地化标题(精确命中算 exact); + * 别名 = alternative_titles + translations(对照 MP __get_names,命中算 alias,ID-2 起译名不再算主标题)。 + * 剧集带各季首播年份与季号(取了详情才有)。 + */ +export function tmdbToCandidate(m: MetadataCandidate): IdentityCandidate { + const titles = [m.title, ...(m.originalTitle ? [m.originalTitle] : [])].filter(Boolean); + for (const t of m.titles) + if ((t.kind === "localized" || t.kind === "original") && !titles.includes(t.text)) titles.push(t.text); + const aliases = [ + ...new Set([ + ...m.aliases, + ...m.titles.filter((t) => t.kind === "alias" || t.kind === "translation").map((t) => t.text), + ]), + ].filter((a) => !titles.includes(a)); + const seasons = (m.seasons ?? []).filter((s) => s.seasonNumber > 0); + const seasonYears = seasons.map((s) => yearOf(s.airDate)).filter((y): y is number => y !== undefined); + return { + provider: "tmdb", + entityType: m.entityType, + externalId: m.externalId, + titles, + ...(m.year !== null ? { year: m.year } : {}), + aliases, + ...(seasonYears.length ? { seasonYears } : {}), + ...(seasons.length ? { seasonNumbers: seasons.map((s) => s.seasonNumber) } : {}), + ...(m.externalIds?.imdb ? { linkedExternalIds: [{ provider: "imdb", id: m.externalIds.imdb }] } : {}), + }; +} + +/** + * 召回输入(ID-2):中文名(含副标题补出的)在前、英文名在后,通用词不搜;剧集线索只搜 tv;年份用于挑选拉详情的候选。 + * 对照 MP _prepare_search_names(cn_name → en_name)。 + */ +export function tmdbRecallInput(parsed: ParsedRelease) { + const names = [parsed.titleCn, parsed.titleEn, parsed.title].filter( + (t): t is string => !!t && !isGenericAliasName(normalizeTitle(t)), + ); + return { + names, + ...(parsed.mediaTypeHint === "tv" ? { type: "tv" as const } : {}), + ...(parsed.year !== undefined ? { year: parsed.year } : {}), + }; +} diff --git a/apps/server/src/modules/sources-identity/tasks.ts b/apps/server/src/modules/sources-identity/tasks.ts index dba99c8e..a245c681 100644 --- a/apps/server/src/modules/sources-identity/tasks.ts +++ b/apps/server/src/modules/sources-identity/tasks.ts @@ -230,9 +230,15 @@ export const sourcesOutboxRoutes: Record = { }, "release.observed": (ev) => { const p = ev.payload as P; + // 标题更新(isNew=false):按新标题重新解析并重新评估,已识别的也不短路(ID-2);触发来源仍是 observed(DEC-1 不变) + const retitled = p.isNew === false; return { task: "resolveObservation", - input: { sourceReleaseId: p.sourceReleaseId, observationId: p.observationId }, + input: { + sourceReleaseId: p.sourceReleaseId, + observationId: p.observationId, + ...(retitled ? { force: true, trigger: "observed" } : {}), + }, }; }, "feed.gap_suspected": (ev) => { diff --git a/apps/server/test/sources-identity/resolve.test.ts b/apps/server/test/sources-identity/resolve.test.ts index 1ccbc8fa..05f0249b 100644 --- a/apps/server/test/sources-identity/resolve.test.ts +++ b/apps/server/test/sources-identity/resolve.test.ts @@ -1,5 +1,6 @@ import { CapabilityRegistry } from "@umt/agent"; import { sql } from "@umt/db"; +import { PARSER_VERSION } from "@umt/domain"; import { createMockTmdb, type MockTmdb, TMDB_FIXTURES, type TmdbFixture } from "@umt/testkit"; import { afterAll, beforeAll, describe, expect, it } from "vitest"; import { refreshMetadata } from "../../src/modules/sources-identity/service/metadata.ts"; @@ -82,7 +83,11 @@ describe.skipIf(!up)("resolveObservation / 身份与元数据", () => { const created = await h.db.selectFrom("media_entity").selectAll().where("id", "=", oppId).executeTakeFirstOrThrow(); expect((created.metadata as Record).overview?.source).toBe("tmdb"); const rel = await h.db.selectFrom("source_release").selectAll().where("id", "=", r.id).executeTakeFirstOrThrow(); - expect(rel).toMatchObject({ resolution_status: "resolved", media_entity_id: oppId, parser_version: "1" }); + expect(rel).toMatchObject({ + resolution_status: "resolved", + media_entity_id: oppId, + parser_version: String(PARSER_VERSION), + }); expect((rel.parsed as { year: number }).year).toBe(2023); const ev = (await h.outbox("release.resolved")).find((x) => x.payload.sourceReleaseId === r.id); expect(ev?.payload).toMatchObject({ mediaEntityId: oppId, evidenceLevel: "rule_confirmed", contentUnitIds: [] }); diff --git a/apps/server/test/sources-identity/retitle.test.ts b/apps/server/test/sources-identity/retitle.test.ts new file mode 100644 index 00000000..eb2f80ac --- /dev/null +++ b/apps/server/test/sources-identity/retitle.test.ts @@ -0,0 +1,75 @@ +/** + * ID-2:站点资源标题更新(RSS 带分类前缀 / 内嵌描述的标题 → 列表标题)后按新标题重新解析并重新识别, + * 不沿用旧解析;已识别的资源也不短路。 + */ +import type { SiteRelease } from "@umt/adapters"; +import { PARSER_VERSION } from "@umt/domain"; +import { afterAll, beforeAll, describe, expect, it } from "vitest"; +import { upsertSiteRelease } from "../../src/modules/sources-identity/service/snapshot.ts"; +import { sourcesOutboxRoutes } from "../../src/modules/sources-identity/tasks.ts"; +import { createHarness, type Harness, pgAvailable, seedSite } from "./harness.ts"; + +const up = await pgAvailable(); + +const rel = (title: string, subtitle?: string): SiteRelease => ({ + siteId: "x", + siteReleaseId: "9001", + title, + ...(subtitle ? { subtitle } : {}), + category: { id: "402", name: "剧集" }, + sizeBytes: null, + seeders: null, + leechers: null, + snatched: null, + promotion: { type: "none", downloadFactor: 1, uploadFactor: 1, expiresAt: null } as SiteRelease["promotion"], + hr: false, + uploadedAt: null, + detailUrl: "https://hhan.example/details.php?id=9001", + tags: [], + raw: {}, + observedAt: new Date().toISOString(), +}); + +describe.skipIf(!up)("标题更新后重新解析 + 重新识别", () => { + let h: Harness; + let siteId: string; + beforeAll(async () => { + h = await createHarness("retitle"); + siteId = (await seedSite(h, { baseUrl: "https://hhan.example/" })).site.id; + }); + afterAll(async () => { + await h?.close(); + }); + + it("RSS 标题 → 列表标题:parsed 按新标题重算,发 release.observed(isNew=false)并强制重新识别", async () => { + const rss = + "[电视剧]Irreplaceable S01 2026 1080p YK WEB-DL H.264 DTS5.1-HHWEB[无可替代 | 第03集 | 1080p | 类型: 剧情][510.37 MB]"; + const first = await h.db.transaction().execute((trx) => upsertSiteRelease(h.rt, trx, { id: siteId }, rel(rss))); + expect(first.inserted).toBe(true); + let row = await h.db.selectFrom("source_release").selectAll().where("id", "=", first.id).executeTakeFirstOrThrow(); + // 分类前缀不进片名;内嵌描述里的中文名补作中文名 + expect(row.parsed).toMatchObject({ titleCn: "无可替代", titleEn: "Irreplaceable", category: "电视剧" }); + + const listTitle = "Irreplaceable S01 2026 1080p YK WEB-DL H.264 DTS5.1-HHWEB"; + const again = await h.db + .transaction() + .execute((trx) => upsertSiteRelease(h.rt, trx, { id: siteId }, rel(listTitle, "无可替代 | 第04集 | 1080p"))); + expect(again).toEqual({ id: first.id, inserted: false }); + row = await h.db.selectFrom("source_release").selectAll().where("id", "=", first.id).executeTakeFirstOrThrow(); + expect(row.title).toBe(listTitle); + expect(row.parser_version).toBe(String(PARSER_VERSION)); + expect(row.parsed).toMatchObject({ titleCn: "无可替代", episodes: { start: 4, end: 4 } }); + + const events = (await h.outbox("release.observed")).filter((e) => e.payload.sourceReleaseId === first.id); + expect(events.map((e) => e.payload.isNew)).toEqual([true, false]); + const route = sourcesOutboxRoutes["release.observed"]!({ payload: events[1]!.payload } as never); + expect(route).toMatchObject({ task: "resolveObservation", input: { force: true, trigger: "observed" } }); + + // 同标题再次快照:不再发重新识别 + await h.db + .transaction() + .execute((trx) => upsertSiteRelease(h.rt, trx, { id: siteId }, rel(listTitle, "无可替代 | 第04集 | 1080p"))); + const after = (await h.outbox("release.observed")).filter((e) => e.payload.sourceReleaseId === first.id); + expect(after).toHaveLength(2); + }); +}); diff --git a/packages/adapters/src/tmdb/README.md b/packages/adapters/src/tmdb/README.md index 9fb7ebc5..22919407 100644 --- a/packages/adapters/src/tmdb/README.md +++ b/packages/adapters/src/tmdb/README.md @@ -1,5 +1,5 @@ # tmdb -TMDB v3 查询:多语言搜索(默认 zh-CN + en-US,按 ID 合并标题)、`searchMulti`、详情(别名 alternative_titles、译名 translations、外部 ID、季列表)、季集列表;内存 LRU + TTL 缓存(默认 1024 条/24h,空结果 30min)与并发合并,并发/间隔限流,429 按 Retry-After 退避。输出归一化候选 `MetadataCandidate`(provider `tmdb`,entityType,externalId,titles,year,aliases,seasons?)。`matchTmdb`/`matchTmdbMulti` 迁移 MoviePilot 的匹配链(名称规范化后完全相等、年份 ±1、季年份优先、别名/译名回退)。错误为 `TmdbError`(auth/network/timeout/notfound/rate/protocol),不会把查询失败当成无结果。 +TMDB v3 查询:多语言搜索(默认 zh-CN + en-US,按 ID 合并标题)、`searchMulti`、详情(别名 alternative_titles、译名 translations、外部 ID、季列表)、季集列表;内存 LRU + TTL 缓存(默认 1024 条/24h,空结果 30min)与并发合并,并发/间隔限流,429 按 Retry-After 退避。输出归一化候选 `MetadataCandidate`(provider `tmdb`,entityType,externalId,titles,year,aliases,seasons?)。`matchTmdb`/`matchTmdbMulti` 迁移 MoviePilot 的匹配链(名称规范化后完全相等、年份 ±1、季年份优先、别名/译名回退)。`recallTmdbCandidates` 是身份识别的候选召回(ID-2,对照 MP 名称顺序与别名回退:中文名/英文名按类型单语言搜索,标题对不上的按年份相近度拉前几个详情比对别名/译名),打分交给 `domain.resolveIdentity`。错误为 `TmdbError`(auth/network/timeout/notfound/rate/protocol),不会把查询失败当成无结果。 仅对 mock 与官方 API 文档验证,未连接真实实例。 diff --git a/packages/adapters/src/tmdb/index.ts b/packages/adapters/src/tmdb/index.ts index 2f38d109..87ff4d67 100644 --- a/packages/adapters/src/tmdb/index.ts +++ b/packages/adapters/src/tmdb/index.ts @@ -1,3 +1,4 @@ export { RequestLimiter, TtlLruCache } from "./cache.ts"; export * from "./client.ts"; export * from "./match.ts"; +export * from "./recall.ts"; diff --git a/packages/adapters/src/tmdb/recall.ts b/packages/adapters/src/tmdb/recall.ts new file mode 100644 index 00000000..50f7f737 --- /dev/null +++ b/packages/adapters/src/tmdb/recall.ts @@ -0,0 +1,71 @@ +/** + * 身份识别的 TMDB 候选召回(ID-2),对照 MoviePilot TheMovieDbModule._prepare_search_names + TmdbApi 的 + * __search_movie_by_name / __search_tv_by_name / _candidate_resolution: + * - 名称顺序 [中文名, 英文名](MP:cn_name → 简体 cn_name → en_name); + * - 按类型搜索(剧集线索只搜 tv,否则 movie + tv),单语言(首选语言,与 MP 相同); + * - 搜索结果的标题/原标题都对不上名称时,按年份相近度取前几个拉详情,用别名 + 译名比对(MP 对全部候选逐个拉详情, + * 这里设上限,打分交给 domain.resolveIdentity)。 + * 与 MP 的差异:MP 在第一个命中处短路,这里收集候选池交给证据打分(同名多候选 → 进待识别池而不是取第一个)。 + */ +import { type MediaEntityType, type MetadataCandidate, type TmdbClient, TmdbError } from "./client.ts"; +import { normalizeTitleForCompare } from "./match.ts"; + +export interface TmdbRecallInput { + /** 查询名称,按优先级(中文名在前);空值与重复会被忽略 */ + names: (string | null | undefined)[]; + /** 有剧集线索时只搜 tv */ + type?: MediaEntityType; + year?: number; + /** 最多用几个名称搜索,默认 2 */ + maxNames?: number; + /** 最多为几个候选拉详情(别名比对),默认 3 */ + maxDetails?: number; + /** 候选池上限,默认 20 */ + limit?: number; +} + +const yearDistance = (c: MetadataCandidate, year: number | undefined, type: MediaEntityType): number => { + if (year === undefined || c.year === null) return 50; + const d = year - c.year; + // 剧集:发布标题的年份可能是后续季的年份(晚于首播) + if (type === "tv" && d > 0) return Math.min(d, 5); + return Math.abs(d); +}; + +export async function recallTmdbCandidates(tmdb: TmdbClient, input: TmdbRecallInput): Promise { + const names: string[] = []; + for (const n of input.names) { + const t = n?.trim(); + if (t && !names.some((x) => normalizeTitleForCompare(x) === normalizeTitleForCompare(t))) names.push(t); + } + const types: MediaEntityType[] = input.type ? [input.type] : ["movie", "tv"]; + const seen = new Map(); + for (const name of names.slice(0, input.maxNames ?? 2)) { + for (const type of types) { + for (const m of await tmdb.searchRaw(type, name)) { + const k = `${m.entityType}:${m.externalId}`; + if (!seen.has(k)) seen.set(k, m); + } + } + } + const pool = [...seen.values()]; + const keys = new Set(names.map(normalizeTitleForCompare)); + const direct = (c: MetadataCandidate) => + [c.title, c.originalTitle].some((t) => !!t && keys.has(normalizeTitleForCompare(t))); + // 年份相容(±1,剧集允许季年份更晚)的候选按年份相近度排序:标题对不上的拉详情比对别名; + // 直接命中的前两个也拉详情(自动绑定本来就要取详情),看另一种文字的名称能否佐证(中文名命中一部、英文名命中另一部时靠它区分) + const compatible = pool + .filter((c) => yearDistance(c, input.year, c.entityType) <= (c.entityType === "tv" ? 5 : 1)) + .sort((a, b) => yearDistance(a, input.year, a.entityType) - yearDistance(b, input.year, b.entityType)); + const directs = compatible.filter(direct); + const needDetail = [...directs.slice(0, 2), ...compatible.filter((c) => !direct(c)).slice(0, input.maxDetails ?? 3)]; + for (const c of needDetail) { + try { + seen.set(`${c.entityType}:${c.externalId}`, await tmdb.details(c.entityType, c.externalId)); + } catch (e) { + // 对照 MP get_info 失败返回 None 后继续:详情只用于补别名,取不到保留搜索结果 + if (!(e instanceof TmdbError)) throw e; + } + } + return [...seen.values()].slice(0, input.limit ?? 20); +}