native-keyshare v1.1.5

native-keyshare

A high-performance shared key-value store implementation designed for multi-threaded environments. This library leverages SharedArrayBuffer to enable efficient data sharing between worker threads.

Features

• Shared Data Access: Allows multiple worker threads to share and manipulate data using named stores
• High Performance: Optimized for fast reads and writes using TypedArrays and efficient buffer handling
• Pattern Operations: Built-in support for wildcards and regex patterns
• Simplified Communication: Uses BroadcastChannel for seamless thread communication
• No Dependencies: Core functionality works without external dependencies
• Optional msgpackr: 2-4x performance boost when using msgpackr

Prerequisites

• Node.js >= 16.0.0
• msgpackr (optional): Improves serialization/deserialization performance by 2-4x

Installation

Install the library using npm:

npm install native-keyshare

If you want to enable performance improvements, also install msgpackr:

npm install msgpackr

Usage

Basic Operations

const { createStore } = require('native-keyshare');

// Create a named store
const store = createStore('mystore');

// Basic operations
store.set('user:1', { name: 'John' });
console.log(store.get('user:1'));  // { name: 'John' }
store.delete('user:1');

// Named stores are isolated
const cacheStore = createStore('cache');
const userStore = createStore('users');

// Get same store instance
const sameStore = createStore('mystore');

Worker Thread Example

Main Thread (index.js)

const { Worker } = require('worker_threads');
const { createStore } = require('native-keyshare');

const store = createStore('shared');
store.set('sharedKey', { value: 'Hello from main!' });

const worker1 = new Worker('./worker.js');
const worker2 = new Worker('./worker.js');

Worker Thread (worker.js)

const { createStore } = require('native-keyshare');

const store = createStore('shared');
console.log(store.get('sharedKey'));  // { value: 'Hello from main!' }
store.set('workerKey', { data: 'Hello from worker!' });

Locks

// Initialize counter - minBufferSize to reuse the same buffer even when increasing length
store.set('counter', 0, { minBufferSize: 20 });
// the following code could be executed from every thread
if (store.lock('counter')) {
  try {
    const x = store.get('counter', true);
    store.set('counter', x + 1, { skipLock: true });
  } finally {
    store.release('counter');
  }
}

API

Store Creation

createStore(storeName?: string)

Creates or retrieves a named key-value store instance.

const store = createStore('mystore');    // Named store
const defaultStore = createStore();      // Default store

Store Methods

set(key: string, value: any, options?: Options): boolean

Sets a key-value pair in the store.

• options.minBufferSize: Initial buffer size in bytes if you expect value to grow
• options.immutable: Dont allow rewriting the buffer. create a new one on update.
• options.ttl: TTL in seconds.

get(key: string, skipLock: boolean = false): any

Retrieves a value from the store.

delete(key: string): boolean

Deletes a value. Supports patterns.

listKeys(pattern?: string): string[]

Lists all keys, optionally filtered by pattern.

lock(key: string, timeout = 1000): boolean

Locks a key.

release(key: string): boolean

Releases a locked key.

clear(): void

Clear the store.

close(): void

Close the store. cleanup local maps and buffer references.

Pattern Operations

The store supports two pattern matching styles for delete() and listKeys():

// Glob-style wildcards
store.delete('users:*');     // Matches: users:123, users:abc, etc
store.delete('session:?');   // Matches: session:1, session:a
store.delete('cache:??');    // Matches: cache:12, cache:ab

// Regular expressions (enclosed in forward slashes)
store.delete('/^user_\d+$/');   // Matches: user_1, user_123
store.delete('/test_.+/');      // Matches: test_abc, test_123

// List keys matching patterns
const userKeys = store.listKeys('user:*');
const logKeys = store.listKeys('/log_\d+/');

Performance Tips

1. Use msgpackr for better serialization (2-4x faster)
2. Set appropriate minBufferSize when you know data will grow
3. Batch operations when possible instead of individual calls
4. Use pattern operations sparingly on large stores

Benchmarks

Performance test for get:

const store = createStore();
store.set('test', { value: 'Benchmark' });

console.time('Benchmark');
for (let i = 0; i < 10000000; i++) {
  store.get('test');
}
console.timeEnd('Benchmark');