type RuntimeMode = "online" | "local";

interface PersistedSettings {
  runtimeMode: RuntimeMode;       // Legacy: overall runtime mode
  sttRuntimeMode: RuntimeMode;    // Speech-to-text runtime mode
  aiRuntimeMode: RuntimeMode;     // AI model runtime mode
}

const DEFAULT_RUNTIME_MODE: RuntimeMode = "online";

interface PersistedSettings {
  apiKey: string;              // API key for cloud provider
  apiBaseUrl: string;          // Base URL for API (e.g., OpenAI)
  sttModelName: string;        // Cloud STT model name
  aiModelName: string;         // Cloud AI model name
  rememberApiKey: boolean;     // Persist API key securely
}

const DEFAULT_API_BASE_URL = "";
const DEFAULT_STT_MODEL_NAME = "";
const DEFAULT_AI_MODEL_NAME = "";

const LOCAL_STT_MODEL_SIZE_LABELS: Record<string, string> = {
  "nvidia/parakeet-tdt-0.6b-v3": "Parakeet v3 (478 MB)",
  "nvidia/parakeet-tdt_ctc-110m": "Parakeet v2 (473 MB)",
  "openai/whisper-large-v3": "Whisper Large (1.1 GB)",
  "openai/whisper-medium": "Whisper Medium (492 MB)",
  "openai/whisper-small": "Whisper Small (487 MB)",
  "UsefulSensors/moonshine-base": "Moonshine Base (58.0 MB)",
  "openai/whisper-large-v3-turbo": "Whisper Turbo (1.6 GB)",
  "nvidia/parakeet-tdt-0.6b-v2": "Parakeet v2 (473 MB)",
  "FunAudioLLM/SenseVoiceSmall": "SenseVoice (160 MB)",
};

interface PersistedSettings {
  localOllamaBaseUrl: string;  // Ollama API URL
  localOllamaModel: string;    // Selected Ollama model
}

const DEFAULT_LOCAL_OLLAMA_BASE_URL = "http://127.0.0.1:11434";

interface LocalSttHardwareAdviceResponse {
  cpuName: string;                    // CPU model detected
  logicalCores: number;               // CPU core count
  totalRamGb: number;                 // System RAM
  nvidiaGpuDetected: boolean;         // NVIDIA GPU present
  gpuName: string;                    // GPU model
  gpuVramGb: number;                  // GPU VRAM
  performanceTier: string;            // "high", "medium", "low"
  slasshySuggestionModel: string;     // Recommended model
  suggestedModels: string[];          // All suitable models
  cautionModels: string[];            // Models that may struggle
  selectedModelWarning: string;       // Warning for current selection
}

interface PersistedSettings {
  sttRuntimeMode: RuntimeMode;    // "online" or "local"
  aiRuntimeMode: RuntimeMode;     // "online" or "local"
}

interface PersistedSettings {
  incognitoMode: boolean;  // Force offline/local processing
}

interface LocalSttDownloadResponse {
  model: string;         // Model identifier
  provider: string;      // Model provider
  method: string;        // Download method
  localPath: string;     // Storage location
  details: string;       // Status message
}

interface LocalSttDownloadStatusResponse {
  active: boolean;           // Download in progress
  completed: boolean;        // Download finished
  success: boolean;          // Download succeeded
  model: string;            // Model being downloaded
  stage: string;            // Current stage
  currentFile: string;      // File being downloaded
  downloadedBytes: number;  // Bytes downloaded
  totalBytes: number;       // Total bytes
  filesCompleted: number;   // Files completed
  filesTotal: number;       // Total files
  progressPercent: number;  // Overall progress %
}

interface LocalSttWarmupResponse {
  model: string;      // Model warmed up
  provider: string;   // Model provider
  warmed: boolean;    // Warmup succeeded
  details: string;    // Status details
}

interface LocalSttRuntimeStateResponse {
  loaded: boolean;            // Runtime is loaded
  daemonCount: number;        // Total model daemons
  loadedDaemonCount: number;  // Active daemons
  details: string;            // Runtime status
}

interface AssistantPipelineResponse {
  sttLatencyMs: number;      // STT processing time
  aiLatencyMs: number;       // AI inference time
  ttsLatencyMs: number;      // TTS generation time
  totalLatencyMs: number;    // Total pipeline time
}