Data Persistence

Patterns for saving and loading data to disk.

Choosing a Storage Method

Need	Solution	When to Use
App preferences	Preferences System	Strongly-typed settings (theme, shortcuts)
Emergency recovery	Recovery System	Crash recovery, backup before risky operations
Relational data	SQLite	User data requiring queries, relationships
External API data	TanStack Query	Remote data with caching (see external-apis.md)

Need to persist data?
├─ App settings? → Preferences (Rust struct + TanStack Query)
├─ User data with queries/relationships? → SQLite (see below)
├─ Remote API data? → external-apis.md
└─ Emergency/crash recovery? → Recovery System

All data goes through Rust for type safety and security. Use TanStack Query on the frontend for loading states and cache invalidation.

File Locations

~/Library/Application Support/com.myapp.app/  (macOS)
├── preferences.json                          # App preferences
└── recovery/                                 # Emergency data
    └── *.json

Atomic Write Pattern (Critical)

All file writes use atomic operations to prevent corruption:

// Write to temp file first, then rename (atomic)
let temp_path = file_path.with_extension("tmp");
std::fs::write(&temp_path, content)?;
std::fs::rename(&temp_path, &file_path)?;

Why: If the app crashes during write, you either have the old file or the new file - never a corrupted partial file.

Preferences System

Rust Side

#[derive(Debug, Clone, Serialize, Deserialize, Type)]
pub struct AppPreferences {
    pub theme: String,
    // Add new preferences here
}

impl Default for AppPreferences {
    fn default() -> Self {
        Self {
            theme: "system".to_string(),
        }
    }
}

React Side

// src/services/preferences.ts
export function usePreferences() {
  return useQuery({
    queryKey: ['preferences'],
    queryFn: async () => unwrapResult(await commands.loadPreferences()),
  })
}

export function useUpdatePreferences() {
  const queryClient = useQueryClient()

  return useMutation({
    mutationFn: (preferences: AppPreferences) =>
      commands.savePreferences(preferences),
    onSuccess: () => {
      queryClient.invalidateQueries({ queryKey: ['preferences'] })
    },
  })
}

Emergency Recovery System

For saving data before crashes or risky operations:

// Save emergency data
await commands.saveEmergencyData({
  filename: 'unsaved-work',
  data: { content: userContent, timestamp: Date.now() },
})

// Load on startup
const recoveryData = await commands.loadEmergencyData({
  filename: 'unsaved-work',
})
if (recoveryData.status === 'ok' && recoveryData.data) {
  // Show recovery dialog
}

Recovery files are automatically cleaned up after 7 days via cleanupOldRecoveryFiles.

Adding New Persistent Data

1. Define Rust struct

#[derive(Debug, Clone, Serialize, Deserialize, Type)]
pub struct MyData {
    pub field: String,
}

impl Default for MyData {
    fn default() -> Self {
        Self { field: "default".to_string() }
    }
}

2. Add Tauri commands

Follow the pattern in src-tauri/src/commands/preferences.rs:

• load_* command with Default fallback
• save_* command with atomic write

3. Register commands

Add to src-tauri/src/bindings.rs and regenerate bindings:

npm run rust:bindings

4. Create React hooks

export function useMyData() {
  return useQuery({
    queryKey: ['my-data'],
    queryFn: async () => unwrapResult(await commands.loadMyData()),
  })
}

Security

Filename Validation

Always validate filenames to prevent path traversal:

if filename.contains("..") || filename.contains("/") || filename.contains("\\") {
    return Err("Invalid filename".to_string());
}

Directory Permissions

Use Tauri's app_data_dir() for safe storage locations - never write to arbitrary paths.

SQLite Database (When Needed)

Note: SQLite is not installed in this app. Add it when your app needs relational data with queries.

When to Use SQLite

Use Case	Recommendation
Simple key-value settings	Preferences System
User data with relationships	SQLite
Data requiring complex queries	SQLite
Large datasets (1000+ records)	SQLite
Data needing atomic transactions	SQLite

Approach Options

Approach	Use When
rusqlite	Simpler setup, synchronous queries, smaller apps
sqlx	Async queries, compile-time SQL checking, larger apps

Both integrate with Tauri commands and tauri-specta for type safety.

Setup (rusqlite)

cd src-tauri && cargo add rusqlite --features bundled

Architecture Pattern

Tauri commands wrap database operations, TanStack Query provides frontend caching.

React Component → TanStack Query → Tauri Command (rusqlite) → SQLite

use rusqlite::{Connection, params};
use std::sync::Mutex;
use tauri::State;

// Database connection managed as Tauri state
pub struct DbConnection(pub Mutex<Connection>);

#[tauri::command]
#[specta::specta]
pub fn get_items(db: State<DbConnection>) -> Result<Vec<Item>, String> {
    let conn = db.0.lock().map_err(|e| e.to_string())?;
    let mut stmt = conn
        .prepare("SELECT id, name, created_at FROM items ORDER BY created_at DESC")
        .map_err(|e| e.to_string())?;

    let items = stmt
        .query_map([], |row| {
            Ok(Item {
                id: row.get(0)?,
                name: row.get(1)?,
                created_at: row.get(2)?,
            })
        })
        .map_err(|e| e.to_string())?
        .collect::<Result<Vec<_>, _>>()
        .map_err(|e| e.to_string())?;

    Ok(items)
}

Initialize in src-tauri/src/lib.rs:

let db_path = app.path().app_data_dir()?.join("app.db");
let conn = Connection::open(&db_path)?;

// Run migrations
conn.execute(
    "CREATE TABLE IF NOT EXISTS items (
        id INTEGER PRIMARY KEY AUTOINCREMENT,
        name TEXT NOT NULL,
        created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP
    )",
    [],
)?;

app.manage(DbConnection(Mutex::new(conn)));

// Frontend: TanStack Query for caching and loading states
export function useItems() {
  return useQuery({
    queryKey: ['items'],
    queryFn: async () => unwrapResult(await commands.getItems()),
  })
}

export function useAddItem() {
  const queryClient = useQueryClient()
  return useMutation({
    mutationFn: (item: CreateItem) => commands.addItem(item),
    onSuccess: () => queryClient.invalidateQueries({ queryKey: ['items'] }),
  })
}

Migration Rules

• Run migrations at app startup before managing database state
• Use IF NOT EXISTS / IF EXISTS for idempotent migrations
• For complex apps, consider a version table to track applied migrations