fix: strip parens from keywords, top-10 search, clean search query

- buildTitleKeywords: strip parenthetical tags before picking words so
  "Eleven Night (Jersey Club)" → keywords ["Eleven","Night"] instead of
  ["Eleven","Jersey"] — "Jersey" was never in the YT video title
- Cascade: add top-10 steps (3+4) with duration still active before
  dropping duration filter, so "Whispers" (position 6-10) can be found
  without falling through to the unfiltered last resort
- search query: strip non-ASCII chars from artist name so "☆NoAki☆"
  becomes "NoAki" in the query → cleaner YTM search ranking

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

backend/novasonic-server.js

@@ -80,7 +80,10 @@ function buildTitleKeywords(title) {
   const STOPWORDS = new Set(['the','a','an','and','or','of','in','on','at','to','for','feat','ft','with']);
   const MIX_QUALIFIERS = ['radio', 'extended', 'club', 'vip', 'original', 'edit'];
   const lower = (title || '').toLowerCase();
-  const words = (title || '')
+  // Strip parenthetical genre/style tags like "(Jersey Club)", "(feat. X)" before picking keywords
+  // so "Eleven Night (Jersey Club)" → keywords from "Eleven Night", not "Jersey"
+  const strippedTitle = (title || '').replace(/\([^)]*\)/g, '').trim();
+  const words = strippedTitle
     .replace(/[^\w\s]/g, ' ')
     .split(/\s+/)
     .filter(w => w.length >= 3 && !VARIANT_WORDS.has(w.toLowerCase()) && !STOPWORDS.has(w.toLowerCase()));
@@ -239,7 +242,9 @@ async function downloadTrack({ jobId, artist, albumArtist, title, album, year, t
   }
 
   const outputTemplate = path.join(dir, `${trackPrefix}${safeTitle}.%(ext)s`);
-  const query = `${mainArtist} - ${title}`;
+  // Strip non-ASCII chars (e.g. ☆) from artist name for cleaner search ranking
+  const searchArtist = mainArtist.replace(/[^\x00-\x7F]/g, '').trim();
+  const query = `${searchArtist} - ${title}`;
 
   // Step 1: yt-dlp — audio only, no thumbnail (we embed iTunes artwork ourselves)
   const baseArgs = [
@@ -251,10 +256,11 @@ async function downloadTrack({ jobId, artist, albumArtist, title, album, year, t
     '--no-warnings',
     '--newline',
   ];
-  // Filtered searches scan top 5 results and stop at first match.
+  // Filtered searches: top 5 (fast) and top 10 (wider net when top 5 misses).
   // Unfiltered fallback only needs item 1 (no filter to skip past).
-  const PI_FILTERED  = ['--playlist-items', '1-5', '--max-downloads', '1'];
+  const PI_FILTERED    = ['--playlist-items', '1-5',  '--max-downloads', '1'];
-  const PI_UNFILTERED = ['--playlist-items', '1'];
+  const PI_FILTERED_10 = ['--playlist-items', '1-10', '--max-downloads', '1'];
+  const PI_UNFILTERED  = ['--playlist-items', '1'];
 
   const spawnYtdlp = (args) => new Promise((resolve, reject) => {
     const proc = spawn(CONFIG.ytdlpBin, args, { env: { ...process.env, PYTHONUNBUFFERED: '1' } });
@@ -313,33 +319,42 @@ async function downloadTrack({ jobId, artist, albumArtist, title, album, year, t
     const dur = { min: Math.max(30, durationSec - margin), max: durationSec + margin };
 
     log(jobId, `🔍 Suche YT Music (~${durationSec}s ±${margin}s): "${query}"`);
-    // 1. YTM — Dauer + Titel + Varianten-Filter
+    // 1. YTM top-5 — Dauer + Titel + Varianten-Filter
     await tryYtdlp([ytmUrl, ...PI_FILTERED, ...mf(dur, true), ...baseArgs]);
     if (!hasAudio()) {
-      // 2. YT — Dauer + Titel + Varianten-Filter
+      // 2. YT top-5 — Dauer + Titel + Varianten-Filter
       log(jobId, '⚠️  Kein YTM-Treffer — Fallback zu YouTube…');
       await tryYtdlp([`ytsearch5:${query}`, ...PI_FILTERED, ...mf(dur, true), ...baseArgs]);
     }
     if (!hasAudio()) {
-      // 3. YTM — Titel + Varianten (kein Dauer-Check)
+      // 3. YTM top-10 — Dauer + Titel (correcte Version evtl. außerhalb top-5)
+      log(jobId, '⚠️  Erweitere Suche auf top 10 mit Dauerbeschränkung…');
+      await tryYtdlp([ytmUrl, ...PI_FILTERED_10, ...mf(dur, true), ...baseArgs]);
+    }
+    if (!hasAudio()) {
+      // 4. YT top-10 — Dauer + Titel
+      await tryYtdlp([`ytsearch10:${query}`, ...PI_FILTERED_10, ...mf(dur, true), ...baseArgs]);
+    }
+    if (!hasAudio()) {
+      // 5. YTM top-5 — Titel + Varianten (kein Dauer-Check)
       log(jobId, '⚠️  Kein Treffer im Dauerbereich — suche ohne Dauerbeschränkung…');
       await tryYtdlp([ytmUrl, ...PI_FILTERED, ...mf(null, true), ...baseArgs]);
     }
     if (!hasAudio()) {
-      // 4. YT — Titel + Varianten
+      // 6. YT top-5 — Titel + Varianten
       await tryYtdlp([`ytsearch5:${query}`, ...PI_FILTERED, ...mf(null, true), ...baseArgs]);
     }
     if (!hasAudio()) {
-      // 5. YTM — nur Titel (kein Varianten-Filter)
+      // 7. YTM top-5 — nur Titel (kein Varianten-Filter)
       log(jobId, '⚠️  Letzter gefilterter Versuch…');
       await tryYtdlp([ytmUrl, ...PI_FILTERED, ...mf(null, false), ...baseArgs]);
     }
     if (!hasAudio()) {
-      // 6. YT — nur Titel
+      // 8. YT top-5 — nur Titel
       await tryYtdlp([`ytsearch5:${query}`, ...PI_FILTERED, ...mf(null, false), ...baseArgs]);
     }
     if (!hasAudio()) {
-      // 7+8. Letzter Ausweg: komplett ungefiltert
+      // 9+10. Letzter Ausweg: komplett ungefiltert
       log(jobId, '⚠️  Letzter Versuch: keine Filter…');
       await tryYtdlp([ytmUrl, ...PI_UNFILTERED, ...baseArgs]);
     }