feat: enhance application functionality and telemetry monitoring

- Added new scripts in `package.json` for Python dependency installation and sidecar compilation.
- Integrated telemetry fetching in `src/App.tsx` to monitor CPU, RAM, GPU, and disk usage.
- Implemented model downloading functionality with progress tracking in `src/App.tsx`.
- Updated `Footer` and `Header` components to display telemetry data and settings options.
- Enhanced Tauri configuration for sidecar execution and application window settings.
- Updated Python requirements in `requirements.txt` to include additional dependencies for improved functionality.

src-python/__init__.py

@@ -0,0 +1 @@
+# Package initialization

src-python/app/__init__.py

@@ -0,0 +1 @@
+# Package initialization

src-python/app/api_server.py

@@ -0,0 +1,139 @@
+"""
+Servidor FastAPI para expor telemetria e download de modelos.
+Utiliza Server-Sent Events (SSE) para streaming de progresso em tempo real.
+"""
+import sys
+import os
+import json
+from pathlib import Path
+from typing import Dict, List
+from fastapi import FastAPI, HTTPException
+from fastapi.responses import StreamingResponse, JSONResponse
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel
+import asyncio
+
+# Detecta se está rodando como binário ou script
+if getattr(sys, 'frozen', False):
+    # Caminho do executável (PyInstaller)
+    BASE_DIR = os.path.dirname(sys.executable)
+    CONFIG_PATH = os.path.join(BASE_DIR, "modelConfig.yaml")
+    MODELS_DIR = os.path.join(BASE_DIR, "models")
+else:
+    # Rodando como script Python
+    # Ajusta PYTHONPATH antes dos imports relativos
+    script_dir = os.path.dirname(os.path.abspath(__file__))
+    project_root = os.path.dirname(os.path.dirname(script_dir))  # Sobe para raiz do projeto
+    if project_root not in sys.path:
+        sys.path.insert(0, project_root)
+    
+    BASE_DIR = project_root
+    CONFIG_PATH = os.path.join(BASE_DIR, "modelConfig.yaml")
+    MODELS_DIR = os.path.join(BASE_DIR, "src-python", "app", "models")
+
+# Importações relativas (funcionam quando executado como módulo)
+from .telemetry import SystemTelemetry
+from .model_downloader import ModelDownloader
+
+app = FastAPI(title="LocalAI API", version="1.0.0")
+
+# CORS para permitir comunicação com frontend Tauri
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # Em produção, restringir ao localhost
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Instâncias globais
+downloader = ModelDownloader(models_dir=MODELS_DIR)
+
+# Modelos de dados
+class ModelInfo(BaseModel):
+    id: str
+    name: str
+    quant: str
+    size_gb: float
+    repo: str
+    filename: str
+    description: str
+
+class DownloadRequest(BaseModel):
+    model_id: str
+    repo_id: str
+    filename: str
+
+# Cache de modelos disponíveis (pode ser carregado de JSON)
+MODELS_REGISTRY: List[Dict] = [
+    {
+        "id": "lfm-2.5-1.2b",
+        "name": "Liquid LFM 2.5",
+        "quant": "Q6_K",
+        "size_gb": 1.12,
+        "repo": "LiquidAI/LFM2.5-1.2B-Instruct-GGUF",
+        "filename": "LFM2.5-1.2B-Instruct-Q6_K.gguf",
+        "description": "Otimizado para arquitetura sequencial de alta performance."
+    }
+]
+
+@app.get("/api/telemetry")
+async def get_telemetry():
+    """Retorna telemetria do sistema em tempo real."""
+    return SystemTelemetry.get_stats()
+
+@app.get("/api/models")
+async def list_models():
+    """Lista todos os modelos disponíveis com status de download."""
+    models_with_status = []
+    for model in MODELS_REGISTRY:
+        model_path = downloader.get_model_path(model["repo"], model["filename"])
+        model_copy = model.copy()
+        model_copy["downloaded"] = model_path.exists() if model_path else False
+        models_with_status.append(model_copy)
+    return {"models": models_with_status}
+
+@app.post("/api/models/download")
+async def download_model(request: DownloadRequest):
+    """
+    Inicia download de modelo com progresso via SSE.
+    O frontend deve consumir como EventSource.
+    """
+    async def generate_progress():
+        try:
+            async for progress_event in downloader.download_with_progress(
+                repo_id=request.repo_id,
+                filename=request.filename
+            ):
+                yield f"data: {json.dumps(progress_event)}\n\n"
+        except Exception as e:
+            error_event = {"type": "error", "message": str(e)}
+            yield f"data: {json.dumps(error_event)}\n\n"
+    
+    return StreamingResponse(
+        generate_progress(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no"
+        }
+    )
+
+@app.get("/api/models/{model_id}/status")
+async def get_model_status(model_id: str):
+    """Verifica se um modelo específico está baixado."""
+    model = next((m for m in MODELS_REGISTRY if m["id"] == model_id), None)
+    if not model:
+        raise HTTPException(status_code=404, detail="Modelo não encontrado")
+    
+    model_path = downloader.get_model_path(model["repo"], model["filename"])
+    return {
+        "model_id": model_id,
+        "downloaded": model_path.exists() if model_path else False,
+        "path": str(model_path) if model_path else None
+    }
+
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="127.0.0.1", port=8000)

src-python/app/model_downloader.py

@@ -0,0 +1,98 @@
+"""
+Módulo de download de modelos GGUF do HuggingFace com progresso em tempo real.
+Utiliza huggingface_hub com callbacks para reportar progresso via async generator.
+"""
+import os
+from pathlib import Path
+from typing import AsyncGenerator, Dict
+from huggingface_hub import hf_hub_download
+import asyncio
+
+class ModelDownloader:
+    """Gerencia downloads de modelos com progresso assíncrono."""
+    
+    def __init__(self, models_dir: str = "./src-python/app/models"):
+        self.models_dir = Path(models_dir)
+        self.models_dir.mkdir(parents=True, exist_ok=True)
+        self._downloads: Dict[str, float] = {}  # Cache de progresso por modelo
+    
+    def get_model_path(self, repo_id: str, filename: str) -> Path:
+        """Retorna o caminho esperado do modelo."""
+        # huggingface_hub salva em cache_dir/repo_id/filename
+        # Simplificamos para models_dir/repo_id/filename
+        repo_name = repo_id.split("/")[-1]
+        return self.models_dir / repo_name / filename
+    
+    async def download_with_progress(
+        self, 
+        repo_id: str, 
+        filename: str
+    ) -> AsyncGenerator[Dict, None]:
+        """
+        Download assíncrono com progresso via callback.
+        Gera eventos: {"type": "progress", "percent": float} ou {"type": "complete"}
+        """
+        model_path = self.get_model_path(repo_id, filename)
+        
+        # Verifica se já existe
+        if model_path.exists():
+            yield {"type": "complete", "percent": 100, "path": str(model_path)}
+            return
+        
+        # Callback para progresso
+        last_percent = 0
+        
+        def progress_callback(downloaded: int, total: int):
+            nonlocal last_percent
+            if total > 0:
+                percent = min(100, int((downloaded / total) * 100))
+                if percent != last_percent:
+                    last_percent = percent
+                    # Armazena para acesso assíncrono
+                    self._downloads[f"{repo_id}/{filename}"] = percent
+        
+        try:
+            # Executa download em thread separada para não bloquear
+            loop = asyncio.get_event_loop()
+            
+            def run_download():
+                return hf_hub_download(
+                    repo_id=repo_id,
+                    filename=filename,
+                    local_dir=str(self.models_dir),
+                    local_dir_use_symlinks=False,
+                    resume_download=True,
+                    # huggingface_hub não expõe callback direto, então monitoramos o arquivo
+                )
+            
+            # Inicia download em executor
+            download_task = loop.run_in_executor(None, run_download)
+            
+            # Monitora progresso enquanto download ocorre
+            while not download_task.done():
+                key = f"{repo_id}/{filename}"
+                percent = self._downloads.get(key, 0)
+                
+                # Tenta estimar progresso pelo tamanho do arquivo parcial
+                partial_path = self.get_model_path(repo_id, filename)
+                if partial_path.exists():
+                    # Estima baseado no tamanho (aproximado, pois não temos total exato)
+                    size_mb = partial_path.stat().st_size / (1024 * 1024)
+                    # Assumindo ~1GB para modelos pequenos (ajustar conforme necessário)
+                    estimated_total_mb = 1200  # ~1.2GB
+                    if size_mb < estimated_total_mb:
+                        percent = min(95, int((size_mb / estimated_total_mb) * 100))
+                
+                yield {"type": "progress", "percent": percent}
+                await asyncio.sleep(0.5)  # Atualiza a cada 500ms
+            
+            # Download completo
+            final_path = download_task.result()
+            yield {"type": "complete", "percent": 100, "path": final_path}
+            
+            # Limpa cache
+            self._downloads.pop(f"{repo_id}/{filename}", None)
+            
+        except Exception as e:
+            yield {"type": "error", "message": str(e)}
+            raise

src-python/app/telemetry.py

@@ -0,0 +1,44 @@
+import psutil
+import os
+import shutil
+
+# GPUtil é opcional (requer distutils, removido no Python 3.12+)
+try:
+    import GPUtil
+    HAS_GPUTIL = True
+except ImportError:
+    HAS_GPUTIL = False
+
+class SystemTelemetry:
+    @staticmethod
+    def get_stats():
+        # CPU
+        cpu_usage = psutil.cpu_percent(interval=None)
+        
+        # RAM
+        ram = psutil.virtual_memory()
+        
+        # GPU (Pega a primeira disponível)
+        gpu_data = {"usage": 0, "vram_used": 0, "vram_total": 0}
+        if HAS_GPUTIL:
+            try:
+                gpus = GPUtil.getGPUs()
+                if gpus:
+                    gpu_data = {
+                        "usage": gpus[0].load * 100,
+                        "vram_used": gpus[0].memoryUsed,
+                        "vram_total": gpus[0].memoryTotal
+                    }
+            except Exception:
+                pass  # Fallback silencioso se GPUtil falhar
+
+        # Armazenamento (Partição do App)
+        total, used, free = shutil.disk_usage("./")
+
+        return {
+            "cpu": cpu_usage,
+            "ram_used": ram.used / (1024**3),
+            "ram_total": ram.total / (1024**3),
+            "gpu": gpu_data,
+            "disk_free_gb": free / (1024**3)
+        }

src-python/requirements.txt

@@ -1,3 +1,9 @@
 # requirements.txt gerado (estático, baseado na importação visível)
 llama-cpp-python>=0.2.0
-huggingface-hub>=0.20.0
+huggingface-hub>=0.20.0
+fastapi>=0.104.0
+uvicorn>=0.24.0
+pydantic>=2.0.0
+psutil>=5.9.0
+# gputil>=1.4.0  # Opcional: requer distutils (removido no Python 3.12+)
+pyyaml>=6.0