mirror of
https://github.com/ZHANGTIANYAO1/teamspeak-music-bot.git
synced 2026-10-02 04:52:50 +08:00
saopig1andClaude Opus 5
74ea8d26d4
按 id 精确播放原本要写 `!play id:<id>`，冒号在一堆 `!<命令> <子命令> <参数>` 的命令里显得很突兀。现在空格写法 `!play id <id>` 也可以，`!add` / `!playnext` 共用同一个解析器，一起生效。 `id:<id>` 继续支持，不做废弃：用户的聊天记录、旧文档和 !search 输出里都是 这个写法。 冒号是个明确的标记，所以 `id:<任意内容>` 一律当 id。空格不是——「ID 4」和 「ID Bruno」都是真实存在的歌名，而且 `id <链接>` 原本会落到 URL 分支正常解析。 所以空格写法只认「长得像 id」的 token（纯数字 / BV 号 / 11 位以上的 id 字符）， 其余照旧继续走 URL 识别，最后落到普通搜索，不会把搜索词误当成 id。 同步更新 !search 输出的提示、三条 Usage、!help 和 README。 Closes #139 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
101 lines
4.2 KiB
TypeScript
101 lines
4.2 KiB
TypeScript
/**
|
|
* Parsing helpers for picking an EXACT song in a !play / !add / !playnext query,
|
|
* so same-name songs can be disambiguated instead of always getting the single
|
|
* most-popular search hit (issue #90).
|
|
*
|
|
* Two mechanisms:
|
|
* - A song reference: an explicit id / platform URL → play that exact song.
|
|
* - A selection index: "#N" → the Nth result of the previous !search.
|
|
*/
|
|
|
|
export interface SongRef {
|
|
id: string;
|
|
/**
|
|
* Platform inferred from a URL. `null` means the platform wasn't encoded in
|
|
* the reference (e.g. a bare `id:`), so the caller should fall back to the
|
|
* command's flags / default provider.
|
|
*/
|
|
platform: "netease" | "qq" | "bilibili" | null;
|
|
}
|
|
|
|
/**
|
|
* Could this token plausibly BE an id on a supported platform?
|
|
* - NetEase / Kugou numeric ids → all digits
|
|
* - BiliBili → BV + 8-12 alphanumerics
|
|
* - QQ mid (14), YouTube (11), Spotify (22), Jellyfin GUID / Kugou hash (32)
|
|
* → 11+ chars from the id alphabet
|
|
* Deliberately conservative: anything rejected here just stays an ordinary
|
|
* search term, which is what it almost certainly was.
|
|
*/
|
|
function looksLikeSongId(token: string): boolean {
|
|
return /^(?:\d+|BV[0-9A-Za-z]{8,12}|[0-9A-Za-z_-]{11,})$/i.test(token);
|
|
}
|
|
|
|
/**
|
|
* Detect an explicit song reference in a query. Recognizes:
|
|
* - `id <id>` / `id:<id>` → platform from flags/default
|
|
* - NetEase song URL → music.163.com/song?id=N (also /#/song?id=N, /song/N)
|
|
* - QQ song URL → y.qq.com/.../songDetail/MID (or ?songmid=MID)
|
|
* - BiliBili BVID (bare or in a URL) → bilibili.com/video/BVxxxx, b23.tv, or BVxxxx
|
|
* Returns `null` for a plain search term (the common case).
|
|
*/
|
|
export function parseSongRef(raw: string): SongRef | null {
|
|
const q = (raw ?? "").trim();
|
|
if (!q) return null;
|
|
|
|
// Explicit id — platform decided by the command's flags/default. The
|
|
// separator is a colon or plain whitespace, so `id <id>` matches the
|
|
// `!<cmd> <sub> <arg>` shape of every other command (issue #139) while the
|
|
// older `id:<id>` keeps working. Strip trailing punctuation that tags along
|
|
// from a chat paste ("id:12345." / "id:12345)") — no supported id
|
|
// (numeric / BVID / mid) ends in those.
|
|
//
|
|
// The colon is an unambiguous sigil, so `id:<anything>` is always an id. A
|
|
// space is not: "ID 4" and "ID Bruno" are real track titles, and `id <url>`
|
|
// has to keep resolving as a URL. So the space form only claims tokens that
|
|
// could actually be an id; anything else falls through to the URL branches
|
|
// below and ultimately to a plain search.
|
|
const idPrefix = /^id(:\s*|\s+)(\S+)$/i.exec(q);
|
|
if (idPrefix) {
|
|
const id = idPrefix[2].replace(/[.,;)\]]+$/, "");
|
|
if (idPrefix[1].startsWith(":") || looksLikeSongId(id)) {
|
|
return { id, platform: null };
|
|
}
|
|
}
|
|
|
|
// BiliBili BV id, bare or inside a bilibili URL (NetEase ids are numeric, so
|
|
// a "BV..." token never collides with them).
|
|
const bv = /BV[0-9A-Za-z]{8,12}/.exec(q);
|
|
if (bv && (/^BV[0-9A-Za-z]{8,12}$/.test(q) || /bilibili\.com|b23\.tv/i.test(q))) {
|
|
return { id: bv[0], platform: "bilibili" };
|
|
}
|
|
|
|
// NetEase song URL. Only treat `id=N` as a SONG id when the URL is not a
|
|
// collection page (playlist/album/artist/toplist/djradio) — those reuse the
|
|
// same `id=` param but are NOT songs; getSongDetail() would 404 them into a
|
|
// confusing "no song" error instead of falling back to a normal search.
|
|
if (/music\.163\.com/i.test(q) && !/(playlist|album|artist|toplist|djradio)/i.test(q)) {
|
|
const m = /[?&#/]id=(\d+)/.exec(q) ?? /\/song\/(\d+)/.exec(q);
|
|
if (m) return { id: m[1], platform: "netease" };
|
|
}
|
|
|
|
// QQ song URL.
|
|
if (/y\.qq\.com/i.test(q)) {
|
|
const m = /songDetail\/([0-9A-Za-z]+)/.exec(q) ?? /[?&]songmid=([0-9A-Za-z]+)/i.exec(q);
|
|
if (m) return { id: m[1], platform: "qq" };
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
/**
|
|
* Detect a "#N" selection token (1-based) referencing the previous !search.
|
|
* Returns the positive integer, or `null` when the query isn't a selection.
|
|
*/
|
|
export function parseSelectionIndex(raw: string): number | null {
|
|
const m = /^#\s*(\d+)$/.exec((raw ?? "").trim());
|
|
if (!m) return null;
|
|
const n = parseInt(m[1], 10);
|
|
return Number.isFinite(n) && n > 0 ? n : null;
|
|
}
|