Enhance HistoryTable Component with Infinite Scroll and Cache Management

- Updated HistoryTable to implement infinite scrolling for loading history items dynamically.
- Introduced state management for accumulated history and total item count.
- Added Intersection Observer for triggering additional data fetches when scrolling.
- Implemented cache clearing functionality in the backend to manage voice prompt caches effectively.
- Improved loading indicators and user feedback for data fetching states.
- Refactored code for better readability and maintainability.

backend/backends/mlx_backend.py

@@ -175,7 +175,13 @@ class MLXTTSBackend:
             if cached_prompt is not None:
                 # Return cached prompt (should be dict format)
                 if isinstance(cached_prompt, dict):
-                    return cached_prompt, True
+                    # Validate that the cached audio file still exists
+                    cached_audio_path = cached_prompt.get("ref_audio") or cached_prompt.get("ref_audio_path")
+                    if cached_audio_path and Path(cached_audio_path).exists():
+                        return cached_prompt, True
+                    else:
+                        # Cached file no longer exists, invalidate cache
+                        print(f"Cached audio file not found: {cached_audio_path}, regenerating prompt")
         
         # MLX voice prompt format - store audio path and text
         # The model will process this during generation
@@ -263,6 +269,13 @@ class MLXTTSBackend:
             ref_audio = voice_prompt.get("ref_audio") or voice_prompt.get("ref_audio_path")
             ref_text = voice_prompt.get("ref_text", "")
             
+            # Validate that the audio file exists
+            if ref_audio and not Path(ref_audio).exists():
+                print(f"Warning: Audio file not found: {ref_audio}")
+                print("This may be due to a cached voice prompt referencing a deleted temp file.")
+                print("Regenerating without voice prompt.")
+                ref_audio = None
+            
             # Check if model supports voice cloning via generate method
             # MLX API may support ref_audio parameter directly
             try:

backend/backends/pytorch_backend.py

@@ -196,6 +196,8 @@ class PyTorchTTSBackend:
                 # Cache stores as torch.Tensor but actual prompt is dict
                 # Convert if needed
                 if isinstance(cached_prompt, dict):
+                    # For PyTorch backend, the dict should contain tensors, not file paths
+                    # So we can safely return it
                     return cached_prompt, True
                 elif isinstance(cached_prompt, torch.Tensor):
                     # Legacy cache format - convert to dict