From e1625a717d9479f3a0b8c8b4c47cb1ea069a8001 Mon Sep 17 00:00:00 2001
From: fspecii <4722521+fspecii@users.noreply.github.com>
Date: Mon, 2 Mar 2026 17:43:53 +0200
Subject: [PATCH] Fix multiple issues: format API, Gradio availability,
 storage, UI responsiveness

- Format endpoint now calls ACE-Step /format_input REST API directly instead of
  spawning Python, fixing ENOENT errors on Windows (#44, #27, #34)
- isGradioAvailable() tries /gradio_api/info, /info, / in sequence to handle
  Gradio 4.x/5.x/6.x version differences, fixing generation fallback (#53, #20)
- Storage getUrl/getPublicUrl normalize /audio/ prefix to prevent double-prefix
  URLs when reference tracks are used for cover generation (#10)
- Gradio args: fix normalization_db default from 0.0 to -1.0 (Gradio default)
- Volume popover: add 400ms delay before hiding to prevent accidental dismissal (#51)
- Polling: skip setSongs state update when nothing changed to reduce re-renders (#51)
- FFmpeg: add jsdelivr CDN fallback when unpkg fails (#30)
- Model switching: add switchModelIfNeeded() via /v1/init REST API (#45)
---
 App.tsx                              |  24 +++---
 components/Player.tsx                |  10 ++-
 components/VideoGeneratorModal.tsx   |  25 +++++--
 server/src/routes/generate.ts        | 108 ++++++++++++++++++++++-----
 server/src/services/acestep.ts       |  66 ++++++++++++----
 server/src/services/gradio-client.ts |  29 ++++---
 server/src/services/storage/local.ts |   9 +--
 7 files changed, 208 insertions(+), 63 deletions(-)

diff --git a/App.tsx b/App.tsx
index 1a2bf0e..5233de8 100644
--- a/App.tsx
+++ b/App.tsx
@@ -707,17 +707,21 @@ function AppContent() {
           ? (Number(status.progress) > 1 ? Number(status.progress) / 100 : Number(status.progress))
           : undefined;
 
-        setSongs(prev => prev.map(s => {
-          if (s.id === tempId) {
-            return {
-              ...s,
-              queuePosition: status.status === 'queued' ? status.queuePosition : undefined,
-              progress: normalizedProgress ?? s.progress,
-              stage: status.stage ?? s.stage,
-            };
+        setSongs(prev => {
+          const song = prev.find(s => s.id === tempId);
+          if (!song) return prev;
+          const newQueuePos = status.status === 'queued' ? status.queuePosition : undefined;
+          const newProgress = normalizedProgress ?? song.progress;
+          const newStage = status.stage ?? song.stage;
+          // Skip update if nothing changed to avoid unnecessary re-renders
+          if (newProgress === song.progress && newStage === song.stage && newQueuePos === song.queuePosition) {
+            return prev;
           }
-          return s;
-        }));
+          return prev.map(s => {
+            if (s.id !== tempId) return s;
+            return { ...s, queuePosition: newQueuePos, progress: newProgress, stage: newStage };
+          });
+        });
 
         if (status.status === 'succeeded' && status.result) {
           cleanupJob(jobId, tempId);
diff --git a/components/Player.tsx b/components/Player.tsx
index 331abbb..2fd3950 100644
--- a/components/Player.tsx
+++ b/components/Player.tsx
@@ -69,6 +69,7 @@ export const Player: React.FC<PlayerProps> = ({
     const progressBarRef = useRef<HTMLDivElement>(null);
     const fullscreenProgressRef = useRef<HTMLDivElement>(null);
     const [isHoveringVolume, setIsHoveringVolume] = useState(false);
+    const volumeHideTimer = useRef<ReturnType<typeof setTimeout> | null>(null);
     const [showDropdown, setShowDropdown] = useState(false);
     const [isFullscreen, setIsFullscreen] = useState(false);
     const [shareModalOpen, setShareModalOpen] = useState(false);
@@ -755,8 +756,13 @@ export const Player: React.FC<PlayerProps> = ({
                     {/* Volume Control with Vertical Slider */}
                     <div
                         className="relative group hidden md:block"
-                        onMouseEnter={() => setIsHoveringVolume(true)}
-                        onMouseLeave={() => setIsHoveringVolume(false)}
+                        onMouseEnter={() => {
+                            if (volumeHideTimer.current) clearTimeout(volumeHideTimer.current);
+                            setIsHoveringVolume(true);
+                        }}
+                        onMouseLeave={() => {
+                            volumeHideTimer.current = setTimeout(() => setIsHoveringVolume(false), 400);
+                        }}
                     >
                         <button
                             onClick={() => onVolumeChange(volume === 0 ? 0.8 : 0)}
diff --git a/components/VideoGeneratorModal.tsx b/components/VideoGeneratorModal.tsx
index 0f1462d..f10c544 100644
--- a/components/VideoGeneratorModal.tsx
+++ b/components/VideoGeneratorModal.tsx
@@ -228,17 +228,30 @@ export const VideoGeneratorModal: React.FC<VideoGeneratorModalProps> = ({ isOpen
         }
       });
 
-      const baseURL = 'https://unpkg.com/@ffmpeg/core@0.12.6/dist/esm';
-      await ffmpeg.load({
-        coreURL: await toBlobURL(`${baseURL}/ffmpeg-core.js`, 'text/javascript'),
-        wasmURL: await toBlobURL(`${baseURL}/ffmpeg-core.wasm`, 'application/wasm'),
-      });
+      const cdnBases = [
+        'https://unpkg.com/@ffmpeg/core@0.12.6/dist/esm',
+        'https://cdn.jsdelivr.net/npm/@ffmpeg/core@0.12.6/dist/esm',
+      ];
+      let loaded = false;
+      for (const baseURL of cdnBases) {
+        try {
+          await ffmpeg.load({
+            coreURL: await toBlobURL(`${baseURL}/ffmpeg-core.js`, 'text/javascript'),
+            wasmURL: await toBlobURL(`${baseURL}/ffmpeg-core.wasm`, 'application/wasm'),
+          });
+          loaded = true;
+          break;
+        } catch {
+          console.warn(`FFmpeg load failed from ${baseURL}, trying next CDN...`);
+        }
+      }
+      if (!loaded) throw new Error('All CDN sources failed');
 
       ffmpegRef.current = ffmpeg;
       setFfmpegLoaded(true);
     } catch (error) {
       console.error('Failed to load FFmpeg:', error);
-      alert('Failed to load video encoder. Please refresh and try again.');
+      alert('Failed to load video encoder. Check your internet connection and try again.');
     } finally {
       setFfmpegLoading(false);
     }
diff --git a/server/src/routes/generate.ts b/server/src/routes/generate.ts
index aee3b59..dc60bd3 100644
--- a/server/src/routes/generate.ts
+++ b/server/src/routes/generate.ts
@@ -22,6 +22,30 @@ import { getStorageProvider } from '../services/storage/factory.js';
 
 const router = Router();
 
+// Auto-generate a song title from lyrics or style when none is provided
+function autoTitle(params: { title?: string; lyrics?: string; instrumental?: boolean; style?: string; songDescription?: string }): string {
+  if (params.title?.trim()) return params.title.trim();
+
+  // Try first meaningful lyric line (skip section markers like [verse], [chorus])
+  if (!params.instrumental && params.lyrics) {
+    for (const line of params.lyrics.split('\n')) {
+      const t = line.trim();
+      if (t && !/^\[.*\]$/.test(t)) {
+        return t.length > 40 ? t.slice(0, 40).trimEnd() + '…' : t;
+      }
+    }
+  }
+
+  // Fall back to first 4 words of style or description
+  const source = params.style || params.songDescription || '';
+  if (source) {
+    const words = source.trim().split(/\s+/).slice(0, 4).join(' ');
+    return words.charAt(0).toUpperCase() + words.slice(1);
+  }
+
+  return 'Untitled';
+}
+
 const audioUpload = multer({
   storage: multer.memoryStorage(),
   limits: { fileSize: 25 * 1024 * 1024 }, // 25MB max
@@ -125,6 +149,9 @@ interface GenerateBody {
   trackName?: string;
   completeTrackClasses?: string[];
   isFormatCaption?: boolean;
+
+  // Model selection
+  ditModel?: string;
 }
 
 router.post('/upload-audio', authMiddleware, (req: AuthenticatedRequest, res: Response, next: Function) => {
@@ -237,6 +264,7 @@ router.post('/', authMiddleware, async (req: AuthenticatedRequest, res: Response
       trackName,
       completeTrackClasses,
       isFormatCaption,
+      ditModel,
     } = req.body as GenerateBody;
 
     if (!customMode && !songDescription) {
@@ -304,6 +332,7 @@ router.post('/', authMiddleware, async (req: AuthenticatedRequest, res: Response
       trackName,
       completeTrackClasses,
       isFormatCaption,
+      ditModel,
     };
 
     // Create job record in database
@@ -392,7 +421,7 @@ router.get('/status/:jobId', authMiddleware, async (req: AuthenticatedRequest, r
             for (let i = 0; i < audioUrls.length; i++) {
               const audioUrl = audioUrls[i];
               const variationSuffix = audioUrls.length > 1 ? ` (v${i + 1})` : '';
-              const songTitle = (params.title || 'Untitled') + variationSuffix;
+              const songTitle = autoTitle(params) + variationSuffix;
 
               const songId = generateUUID();
 
@@ -416,7 +445,7 @@ router.get('/status/:jobId', authMiddleware, async (req: AuthenticatedRequest, r
                     params.style,
                     params.style,
                     storedPath,
-                    aceStatus.result.duration && aceStatus.result.duration > 0 ? aceStatus.result.duration : (params.duration && params.duration > 0 ? params.duration : 120),
+                    aceStatus.result.duration && aceStatus.result.duration > 0 ? aceStatus.result.duration : (params.duration && params.duration > 0 ? params.duration : 0),
                     aceStatus.result.bpm || params.bpm,
                     aceStatus.result.keyScale || params.keyScale,
                     aceStatus.result.timeSignature || params.timeSignature,
@@ -442,7 +471,7 @@ router.get('/status/:jobId', authMiddleware, async (req: AuthenticatedRequest, r
                     params.style,
                     params.style,
                     audioUrl,
-                    aceStatus.result.duration && aceStatus.result.duration > 0 ? aceStatus.result.duration : (params.duration && params.duration > 0 ? params.duration : 120),
+                    aceStatus.result.duration && aceStatus.result.duration > 0 ? aceStatus.result.duration : (params.duration && params.duration > 0 ? params.duration : 0),
                     aceStatus.result.bpm || params.bpm,
                     aceStatus.result.keyScale || params.keyScale,
                     aceStatus.result.timeSignature || params.timeSignature,
@@ -750,8 +779,63 @@ router.post('/format', authMiddleware, async (req: AuthenticatedRequest, res: Re
       return;
     }
 
-    const { spawn } = await import('child_process');
+    const ACESTEP_API_URL = config.acestep.apiUrl;
 
+    // Build param_obj for the REST API
+    const paramObj: Record<string, unknown> = {};
+    if (bpm && bpm > 0) paramObj.bpm = bpm;
+    if (duration && duration > 0) paramObj.duration = duration;
+    if (keyScale) paramObj.key = keyScale;
+    if (timeSignature) paramObj.time_signature = timeSignature;
+
+    // Primary path: call ACE-Step's /format_input REST endpoint (avoids Python spawn ENOENT on Windows)
+    try {
+      console.log(`[Format] Calling REST API: ${ACESTEP_API_URL}/format_input`);
+      const apiRes = await fetch(`${ACESTEP_API_URL}/format_input`, {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({
+          prompt: caption,
+          lyrics: lyrics || '',
+          temperature: temperature ?? 0.85,
+          param_obj: paramObj,
+        }),
+        signal: AbortSignal.timeout(300_000), // 5 min — LLM may need to init first
+      });
+
+      const apiData = await apiRes.json() as any;
+
+      if (!apiRes.ok || apiData.code !== 200) {
+        const errMsg = apiData.error || apiData.detail || `Format API returned ${apiRes.status}`;
+        console.error('[Format] API error:', errMsg);
+        res.status(500).json({ success: false, error: errMsg });
+        return;
+      }
+
+      const d = apiData.data;
+      res.json({
+        caption: d.caption,
+        lyrics: d.lyrics,
+        bpm: d.bpm,
+        duration: d.duration,
+        key_scale: d.key_scale,
+        time_signature: d.time_signature,
+        vocal_language: d.vocal_language,
+      });
+      return;
+    } catch (fetchErr: any) {
+      // Only fall back to Python spawn on network errors (service not yet reachable)
+      if (fetchErr?.name !== 'AbortError' && (fetchErr?.code === 'ECONNREFUSED' || fetchErr?.cause?.code === 'ECONNREFUSED')) {
+        console.warn('[Format] REST API unreachable, falling back to Python spawn');
+      } else {
+        console.error('[Format] REST API request failed:', fetchErr?.message);
+        res.status(500).json({ success: false, error: fetchErr?.message || 'Format request failed' });
+        return;
+      }
+    }
+
+    // Fallback: Python spawn (only reached when REST API is unreachable)
+    const { spawn } = await import('child_process');
     const ACESTEP_DIR = process.env.ACESTEP_PATH || path.resolve(path.dirname(fileURLToPath(import.meta.url)), '../../../ACE-Step-1.5');
     const __filename = fileURLToPath(import.meta.url);
     const __dirname = path.dirname(__filename);
@@ -759,12 +843,7 @@ router.post('/format', authMiddleware, async (req: AuthenticatedRequest, res: Re
     const FORMAT_SCRIPT = path.join(SCRIPTS_DIR, 'format_sample.py');
     const pythonPath = resolvePythonPath(ACESTEP_DIR);
 
-    const args = [
-      FORMAT_SCRIPT,
-      '--caption', caption,
-      '--json',
-    ];
-
+    const args = [FORMAT_SCRIPT, '--caption', caption, '--json'];
     if (lyrics) args.push('--lyrics', lyrics);
     if (bpm && bpm > 0) args.push('--bpm', String(bpm));
     if (duration && duration > 0) args.push('--duration', String(duration));
@@ -776,15 +855,11 @@ router.post('/format', authMiddleware, async (req: AuthenticatedRequest, res: Re
     if (lmModel) args.push('--lm-model', lmModel);
     if (lmBackend) args.push('--lm-backend', lmBackend);
 
-    console.log(`[Format] Running: ${pythonPath} ${args.join(' ')}`);
-    console.log(`[Format] CWD: ${ACESTEP_DIR}`);
+    console.log(`[Format] Fallback spawn: ${pythonPath} ${args.join(' ')}`);
     const result = await new Promise<{ success: boolean; data?: any; error?: string }>((resolve) => {
       const proc = spawn(pythonPath, args, {
         cwd: ACESTEP_DIR,
-        env: {
-          ...process.env,
-          ACESTEP_PATH: ACESTEP_DIR,
-        },
+        env: { ...process.env, ACESTEP_PATH: ACESTEP_DIR },
       });
 
       let stdout = '';
@@ -795,7 +870,6 @@ router.post('/format', authMiddleware, async (req: AuthenticatedRequest, res: Re
 
       proc.on('close', (code) => {
         if (code === 0 && stdout) {
-          // stdout may contain log lines before the JSON — extract last JSON line
           const lines = stdout.trim().split('\n');
           let jsonStr = '';
           for (let i = lines.length - 1; i >= 0; i--) {
diff --git a/server/src/services/acestep.ts b/server/src/services/acestep.ts
index 08e2213..c37a6f6 100644
--- a/server/src/services/acestep.ts
+++ b/server/src/services/acestep.ts
@@ -77,7 +77,7 @@ const SCRIPTS_DIR = path.join(__dirname, '../../scripts');
 const PYTHON_SCRIPT = path.join(SCRIPTS_DIR, 'simple_generate.py');
 
 // ---------------------------------------------------------------------------
-// Gradio generation: map params to the 45 positional args for /generation_wrapper
+// Gradio generation: map params to the 51 positional args for /generation_wrapper
 // ---------------------------------------------------------------------------
 
 /**
@@ -186,16 +186,16 @@ async function buildGradioArgs(params: GenerationParams): Promise<unknown[]> {
     params.constrainedDecodingDebug ?? false,                     // 38: Constrained Decoding Debug
     params.allowLmBatch ?? true,                                  // 39: ParallelThinking
     params.getScores ?? false,                                    // 40: Auto Score
-    // Note: auto_lrc (getLrc) is a hidden Gradio state param — NOT included in the public API args
-    params.scoreScale ?? 0.5,                                     // 41: Quality Score Sensitivity (0.01-1.0)
-    params.lmBatchChunkSize ?? 8,                                 // 42: LM Batch Chunk Size
-    params.trackName || null,                                     // 43: Track Name
-    params.completeTrackClasses || [],                            // 44: Track Names
-    false,                                                        // 45: Enable Normalization (ACE-Step v1.5 new param, default false)
-    0.0,                                                          // 46: Normalization DB (ACE-Step v1.5 new param, default 0.0)
-    0.0,                                                          // 47: Latent Shift (ACE-Step v1.5 new param, default 0.0)
-    1.0,                                                          // 48: Latent Rescale (ACE-Step v1.5 new param, default 1.0)
-    params.autogen ?? false,                                      // 49: AutoGen (visible last param)
+    params.getLrc ?? false,                                       // 41: Auto LRC (timestamped lyrics)
+    params.scoreScale ?? 0.5,                                     // 42: Quality Score Sensitivity (0.01-1.0)
+    params.lmBatchChunkSize ?? 8,                                 // 43: LM Batch Chunk Size
+    params.trackName || null,                                     // 44: Track Name
+    params.completeTrackClasses || [],                            // 45: Track Names
+    true,                                                         // 46: Enable Normalization (ACE-Step v1.5, default true)
+    -1.0,                                                         // 47: Normalization DB (ACE-Step v1.5, default -1.0)
+    0.0,                                                          // 48: Latent Shift (ACE-Step v1.5, default 0.0)
+    1.0,                                                          // 49: Latent Rescale (ACE-Step v1.5, default 1.0)
+    params.autogen ?? false,                                      // 50: AutoGen
     // Note: current_batch_index, total_batches, batch_queue, generation_params_state
     // are hidden Gradio state variables and must NOT be passed via client.predict()
   ];
@@ -365,6 +365,40 @@ export async function checkSpaceHealth(): Promise<boolean> {
   return isGradioAvailable();
 }
 
+// ---------------------------------------------------------------------------
+// Model switching — call /v1/init to change the active DiT model
+// ---------------------------------------------------------------------------
+
+async function getActiveModel(): Promise<string | null> {
+  try {
+    const res = await fetch(`${ACESTEP_API}/v1/models`);
+    if (!res.ok) return null;
+    const data = await res.json() as any;
+    const models = data?.data?.models || data?.models || [];
+    return models[0]?.name || null;
+  } catch {
+    return null;
+  }
+}
+
+async function switchModelIfNeeded(ditModel: string): Promise<void> {
+  const activeModel = await getActiveModel();
+  if (activeModel === ditModel) return; // already loaded, no-op
+
+  console.log(`[Model] Switching from '${activeModel ?? 'unknown'}' to '${ditModel}'`);
+  const res = await fetch(`${ACESTEP_API}/v1/init`, {
+    method: 'POST',
+    headers: { 'Content-Type': 'application/json' },
+    body: JSON.stringify({ model: ditModel, init_llm: false }),
+  });
+
+  if (!res.ok) {
+    const err = await res.text().catch(() => '');
+    throw new Error(`Model switch to '${ditModel}' failed: ${res.status} ${err}`);
+  }
+  console.log(`[Model] Switched to '${ditModel}'`);
+}
+
 // Discover endpoints (for compatibility)
 export async function discoverEndpoints(): Promise<unknown> {
   return { provider: 'acestep-gradio', endpoint: ACESTEP_API };
@@ -472,6 +506,12 @@ async function processGenerationViaGradio(
   params: GenerationParams,
   job: ActiveJob,
 ): Promise<void> {
+  // Switch DiT model if a specific one was requested
+  if (params.ditModel) {
+    job.stage = `Loading model ${params.ditModel}...`;
+    await switchModelIfNeeded(params.ditModel);
+  }
+
   const client = await getGradioClient();
   const args = await buildGradioArgs(params);
 
@@ -552,7 +592,7 @@ async function processGenerationViaGradio(
 
   const finalDuration = actualDuration > 0
     ? actualDuration
-    : (metas.duration || params.duration || 60);
+    : (metas.duration || params.duration || 0);
 
   job.status = 'succeeded';
   job.result = {
@@ -703,7 +743,7 @@ async function processGenerationViaPython(
       console.warn(`Job ${jobId}: Failed to cleanup output dir`, cleanupError);
     }
 
-    const finalDuration = actualDuration > 0 ? actualDuration : (params.duration && params.duration > 0 ? params.duration : 60);
+    const finalDuration = actualDuration > 0 ? actualDuration : (params.duration && params.duration > 0 ? params.duration : 0);
 
     job.status = 'succeeded';
     job.result = {
diff --git a/server/src/services/gradio-client.ts b/server/src/services/gradio-client.ts
index b527311..51822a1 100644
--- a/server/src/services/gradio-client.ts
+++ b/server/src/services/gradio-client.ts
@@ -41,17 +41,26 @@ export function resetGradioClient(): void {
 
 /**
  * Check if the Gradio app is reachable.
+ * Tries multiple well-known endpoints to handle version differences.
  */
 export async function isGradioAvailable(): Promise<boolean> {
-  try {
-    const controller = new AbortController();
-    const timeout = setTimeout(() => controller.abort(), 3000);
-    const response = await fetch(`${config.acestep.apiUrl}/gradio_api/info`, {
-      signal: controller.signal,
-    });
-    clearTimeout(timeout);
-    return response.ok;
-  } catch {
-    return false;
+  const baseUrl = config.acestep.apiUrl;
+  const candidates = [
+    `${baseUrl}/gradio_api/info`, // Gradio 5+
+    `${baseUrl}/info`,            // Gradio 4.x fallback
+    `${baseUrl}/`,                // Any HTTP response means server is up
+  ];
+
+  for (const url of candidates) {
+    try {
+      const controller = new AbortController();
+      const timer = setTimeout(() => controller.abort(), 5000);
+      const response = await fetch(url, { signal: controller.signal });
+      clearTimeout(timer);
+      if (response.ok || response.status < 500) return true;
+    } catch {
+      // Try next candidate
+    }
   }
+  return false;
 }
diff --git a/server/src/services/storage/local.ts b/server/src/services/storage/local.ts
index 63814d4..903582c 100644
--- a/server/src/services/storage/local.ts
+++ b/server/src/services/storage/local.ts
@@ -22,14 +22,13 @@ export class LocalStorageProvider implements StorageProvider {
   }
 
   async getUrl(key: string, _expiresIn?: number): Promise<string> {
-    return `/audio/${key}`;
+    const cleanKey = key.startsWith('/audio/') ? key.slice('/audio/'.length) : key;
+    return `/audio/${cleanKey.replace(/^\/+/, '')}`;
   }
 
   getPublicUrl(key: string): string {
-    if (key.startsWith('/audio/')) {
-      return key;
-    }
-    return `/audio/${key}`;
+    const cleanKey = key.startsWith('/audio/') ? key.slice('/audio/'.length) : key;
+    return `/audio/${cleanKey.replace(/^\/+/, '')}`;
   }
 
   async delete(key: string): Promise<void> {