NovaSonic/backend/lib/metadata.js

const https = require('https');
const path  = require('path');
const fs    = require('fs');

function sanitize(s) {
  return (s || 'Unknown').replace(/[\\/:*?"<>|]/g, '_').trim().slice(0, 100);
}

function navidromeArtist(name) {
  if (!name) return name;
  return name
    .replace(/\s+feat\.?\s+/gi, ';')
    .replace(/\s+ft\.?\s+/gi, ';')
    .replace(/\s+featuring\s+/gi, ';')
    .replace(/\s+&\s+/g, ';')

    .replace(/,\s*/g, ';');
}

// Download hi-res iTunes artwork (1000x1000) to cover.jpg, return path or null
async function downloadArtwork(url, dir) {
  if (!url) return null;
  const dest  = path.join(dir, 'cover.jpg');
  const hiRes = url.replace(/\d+x\d+bb\./, '1000x1000bb.');
  return new Promise(resolve => {
    const tryUrl = (u, fallback) => {
      const file = fs.createWriteStream(dest);
      https.get(u, res => {
        if (res.statusCode !== 200) {
          res.resume();
          file.close();
          try { fs.unlinkSync(dest); } catch {}
          if (fallback) tryUrl(fallback, null);
          else resolve(null);
          return;
        }
        res.pipe(file);
        file.on('finish', () => resolve(dest));
        file.on('error',  () => resolve(null));
      }).on('error', () => {
        if (fallback) tryUrl(fallback, null);
        else resolve(null);
      });
    };
    tryUrl(hiRes, url !== hiRes ? url : null);
  });
}

const AUDIO_EXTS = ['mp3', 'flac', 'opus', 'm4a', 'ogg', 'wav'];

// Strips release-type suffixes like "- Single", "- EP", "2024" from end of title/stem.
const _RELEASE_SUFFIX_RE = /\s*[-–]\s*(?:(?:official\s+)?(?:music\s+)?(?:single|ep|lp|album|ost|soundtrack|remaster(?:ed)?|deluxe(?:\s+edition)?|bonus\s+tracks?|b-?sides?|version|edit)\b.*|(?:20|19)\d{2})\s*$/i;

// Robust duplicate detection: search case-insensitively across all album folders
// under the artist folder matching firstArtist, compare normalized title ignoring
// release suffixes and track prefixes. Handles DJ OKAWARI vs DJ Okawari, and
// "recognize - Single.mp3" vs "recognize.mp3" (drift between runs).
function findExistingTrack(musicRoot, firstArtist, cleanTitle) {
  const norm = s => (s || '').toLowerCase()
    .replace(/[\[(（][^\]）)]*[\]）)]/g, '')
    // Filesystem-sanitized chars (/ \ : * ? " < > |) become "_" on disk — map those AND "_"
    // to a space so a title tag "Kwik Flip | YTPMV" matches the file "Kwik Flip _ YTPMV".
    .replace(/[/\\:*?"<>|_]/g, ' ')
    .replace(/[,.'!]/g, '')
    .replace(/\s+/g, ' ')
    .trim();
  const normArtist = norm(firstArtist);
  const normTitle  = norm(cleanTitle);
  // Accept both the raw cleanTitle and version without release suffix (e.g. "recognize" matches "recognize - Single")
  const titleVariants = new Set([normTitle, norm(cleanTitle.replace(_RELEASE_SUFFIX_RE, ''))]);

  try {
    for (const ad of fs.readdirSync(musicRoot, { withFileTypes: true })) {
      if (!ad.isDirectory() || norm(ad.name) !== normArtist) continue;
      const artistPath = path.join(musicRoot, ad.name);
      try {
        for (const abd of fs.readdirSync(artistPath, { withFileTypes: true })) {
          if (!abd.isDirectory()) continue;
          const albumPath = path.join(artistPath, abd.name);
          try {
            for (const f of fs.readdirSync(albumPath)) {
              const ext = path.extname(f).slice(1).toLowerCase();
              if (!AUDIO_EXTS.includes(ext)) continue;
              // Strip extension + leading track prefix ("01 - ", "01-15 – " etc.)
              const stem = f.slice(0, -(ext.length + 1)).replace(/^\d[\d-]*\s*[-–—]\s*/, '');
              // Stem may be "Artist - Title" or just "Title"
              const di = stem.indexOf(' - ');
              const rawTitle = di >= 0 ? stem.slice(di + 3) : stem;
              const normStemTitle = norm(rawTitle.replace(_RELEASE_SUFFIX_RE, ''));
              if (titleVariants.has(normStemTitle)) return path.join(albumPath, f);
            }
          } catch {}
        }
      } catch {}
    }
  } catch {}
  return null;
}

// Fetch plain + synced lyrics from LRCLIB
async function fetchLyrics(artist, title, album) {
  try {
    const p = new URLSearchParams({ artist_name: artist, track_name: title });
    if (album) p.set('album_name', album);
    const r = await fetch(`https://lrclib.net/api/get?${p}`, {
      headers: { 'User-Agent': 'NovaSonic/1.0' },
      signal: AbortSignal.timeout(8000),
    });
    if (!r.ok) return null;
    const d = await r.json();
    return { plain: d.plainLyrics || null, synced: d.syncedLyrics || null };
  } catch { return null; }
}

// Normalize a track title for comparison: unify variant suffixes so that
// "don't rush - slowed" and "don't rush (slowed)" are treated as equal.
function normalizeTitle(s) {
  return (s || '')
    .toLowerCase()
    .trim()
    // convert "title - variant" → "title (variant)" so both forms compare equal
    .replace(/\s+-\s+([^-]+)$/, ' ($1)')
    // strip all remaining parens/brackets content for the base comparison
    .replace(/[\[(（][^\]）)]*[\]）)]/g, '')
    // strip punctuation so "Mr. Lawrence" == "Mr Lawrence" and comma variants match
    .replace(/[,.'!?]/g, '')
    .replace(/\s+/g, ' ')
    .trim();
}

// iTunes JP store returns localized genre names — map them to English equivalents
const JP_GENRE_MAP = {
  'アニメ': 'Anime', 'イージーリスニング': 'Easy Listening', 'ポップ': 'Pop',
  'J-Pop': 'J-Pop', 'ロック': 'Rock', 'クラシック': 'Classical', 'ジャズ': 'Jazz',
  'エレクトロニック': 'Electronic', 'R&B/ソウル': 'R&B/Soul', 'ヒップホップ/ラップ': 'Hip-Hop/Rap',
  'オルタナティブ': 'Alternative', 'サウンドトラック': 'Soundtrack', 'ボーカル': 'Vocal',
  'インストゥルメンタル': 'Instrumental', 'ワールド': 'World', 'フォーク': 'Folk',
  'カントリー': 'Country', 'レゲエ': 'Reggae', 'ソウル': 'Soul', 'ブルース': 'Blues',
  'メタル': 'Metal', 'パンク': 'Punk', 'ダンス': 'Dance', 'コメディ': 'Comedy',
  'スピーチ': 'Speech', 'キッズ/ファミリー': 'Kids/Family',
};
function toEnglishGenre(g) { return JP_GENRE_MAP[g] || g; }

// iTunes track lookup — fills in album, year, trackNum, totalTracks, genre when missing
async function lookupItunesTrackMeta(artist, title) {
  try {
    // Search with base title (strip variant suffix) for better iTunes hit rate —
    // e.g. "don't rush - slowed" → search "don't rush" so iTunes finds the single
    const baseTitle = title.replace(/\s*[-–]\s*\S[^-]*$/, '').trim() || title;
    const searchTitle = baseTitle !== title ? baseTitle : title;
    // Strip leading aesthetic punctuation many artists use (".diedlonely", "†", "✝", "_x")
    // so the iTunes search actually surfaces them instead of a same-titled track by someone else.
    const stripLeadSym = s => (s || '').replace(/^[^\p{L}\p{N}]+/u, '').trim();
    const searchArtist = stripLeadSym(artist.split(/[;,&]/)[0].trim());
    const fetchItunes = (country) => new Promise((resolve, reject) => {
      const url = `https://itunes.apple.com/search?term=${encodeURIComponent(searchArtist + ' ' + searchTitle)}&media=music&entity=musicTrack&limit=10&country=${country}&explicit=Yes`;
      https.get(url, res => {
        let d = ''; res.on('data', c => d += c);
        res.on('end', () => { try { resolve(JSON.parse(d).results || []); } catch { reject(new Error('parse')); } });
      }).on('error', reject);
    });
    const normT = normalizeTitle(title);
    const normA = (searchArtist || '').toLowerCase().trim();
    // Split by space OR comma/semicolon/slash so "VXLLAIN, iGRES" → "vxllain" not "vxllain,"
    const firstWord = s => stripLeadSym((s || '').toLowerCase()).split(/[\s,;/]/)[0];
    const strictMatch = results => results.find(t =>
      normalizeTitle(t.trackName) === normT && firstWord(t.artistName) === firstWord(normA)
    );
    const titleOnlyMatch = results => results.find(t => normalizeTitle(t.trackName) === normT);

    // US store only — one consistent naming source (English/romaji) across search + download, so the
    // same artist always resolves the same way instead of "sometimes Japanese, sometimes romaji". A
    // single request instead of JP+US also halves the download's iTunes calls (helps the rate limit).
    const usResults = await fetchItunes('us');
    const JP_RE = /[぀-ヿ㐀-䶿一-鿿]/;
    const isJpSearch = JP_RE.test(normT) || JP_RE.test(normA);
    // titleOnlyMatch (no artist check) is too permissive when the title is Japanese — common anime/
    // OST titles exist under different artists — so only use it as a fallback for non-Japanese titles.
    const match = strictMatch(usResults) || (!isJpSearch && titleOnlyMatch(usResults));
    if (!match) return {};
    const naming = match;
    const rawGenre = match.primaryGenreName;
    const genre = rawGenre ? toEnglishGenre(rawGenre) : undefined;
    return {
      album:            naming.collectionName       || '',
      albumArtist:      naming.collectionArtistName  || undefined,
      trackName:        naming.trackName             || undefined,
      year:             match.releaseDate ? new Date(match.releaseDate).getFullYear() : undefined,
      trackNum:         match.trackNumber   || undefined,
      totalTracks:      match.trackCount    || undefined,
      discNum:          match.discNumber    || undefined,
      discTotal:        match.discCount     || undefined,
      genre:            genre,
      artworkUrl:       (naming.artworkUrl100 || match.artworkUrl100 || '').replace('100x100bb', '1000x1000bb'),
      durationMs:       match.trackTimeMillis  || undefined,
      itunesArtistName: naming.artistName            || undefined,
    };
  } catch { return {}; }
}

// Fetch individual tracks for an iTunes collectionId
async function fetchItunesTracks(collectionId) {
  return new Promise((resolve, reject) => {
    const url = `https://itunes.apple.com/lookup?id=${collectionId}&entity=song&explicit=Yes`;
    https.get(url, res => {
      let data = '';
      res.on('data', d => data += d);
      res.on('end', () => {
        try {
          const parsed = JSON.parse(data);
          resolve((parsed.results || []).filter(r => r.wrapperType === 'track'));
        } catch { reject(new Error('iTunes API parse error')); }
      });
    }).on('error', reject);
  });
}

// Search iTunes for album tracks by artist + album name (fallback when no collectionId)
async function searchItunesAlbumTracks(artist, album) {
  return new Promise(resolve => {
    const q = encodeURIComponent(`${artist} ${album}`);
    https.get(`https://itunes.apple.com/search?term=${q}&media=music&entity=album&limit=5&country=us&explicit=Yes`, res => {
      let d = '';
      res.on('data', c => d += c);
      res.on('end', () => {
        try {
          const results = JSON.parse(d).results || [];
          const match = results.find(r =>
            r.collectionName?.toLowerCase().includes(album.toLowerCase()) ||
            r.artistName?.toLowerCase().includes(artist.toLowerCase())
          );
          if (match) fetchItunesTracks(match.collectionId).then(resolve).catch(() => resolve([]));
          else resolve([]);
        } catch { resolve([]); }
      });
    }).on('error', () => resolve([]));
  });
}

module.exports = {
  sanitize, navidromeArtist,
  downloadArtwork, findExistingTrack,
  lookupItunesTrackMeta, fetchItunesTracks, searchItunesAlbumTracks,
};