kimdb

High-performance document database with CRDT real-time sync.

npm install kimdb

Features

• Document Database: SQLite-based, JSON documents
• Real-time Sync: WebSocket + CRDT (Conflict-free Replicated Data Types)
• Offline-first: Local-first with automatic sync
• CRDT Primitives: VectorClock, LWWSet, LWWMap, RGA, RichText
• Presence: Real-time user presence and cursors
• Undo/Redo: Built-in undo/redo support
• TypeScript: Full TypeScript support
• Redis Clustering: Optional multi-server support

Quick Start

CLI

# Initialize project
npx kimdb init

# Start server
npx kimdb start

Server

import { KimDBServer } from 'kimdb/server';

const server = new KimDBServer();
await server.start();
// Server running on http://localhost:40000

Client

import { KimDBClient } from 'kimdb/client';

const client = new KimDBClient({
  url: 'ws://localhost:40000/ws',
  apiKey: 'your-api-key',
});

await client.connect();

// Subscribe to collection
client.subscribe('posts');

// Open document
const doc = await client.openDocument('posts', 'post-1');

// Set values
client.set('posts', 'post-1', 'title', 'Hello World');
client.set('posts', 'post-1', ['author', 'name'], 'John');

// Listen for sync events
client.onSync = (collection, event, data) => {
  console.log('Sync:', collection, event, data);
};

CRDT Direct Usage

import { CRDTDocument, LWWSet, VectorClock } from 'kimdb/crdt';

// Document with automatic conflict resolution
const doc = new CRDTDocument('node1', 'doc-1');
doc.set('title', 'My Document');
doc.listInsert('tags', 0, 'crdt');
doc.setAdd('authors', 'user1');

// Get as plain object
const obj = doc.toObject();

// Serialize for network/storage
const json = doc.toJSON();
const restored = CRDTDocument.fromJSON(json);

// Standalone CRDT types
const set = new LWWSet('node1');
set.add('item1');
set.remove('item2');

Configuration

Environment variables:

| Variable | Default | Description | |----------|---------|-------------| | KIMDB_API_KEY | (required in prod) | API key for authentication | | KIMDB_PORT | 40000 | Server port | | KIMDB_HOST | 0.0.0.0 | Server host | | KIMDB_DATA_DIR | ./data | Data directory | | REDIS_ENABLED | false | Enable Redis for clustering | | MARIADB_ENABLED | false | Enable MariaDB for logging |

REST API

Health Check

GET /health

Collections

GET /api/collections
GET /api/c/:collection
GET /api/c/:collection/:id

SQL API

POST /api/sql
{
  "sql": "SELECT * FROM table WHERE id = ?",
  "params": [1],
  "collection": "my_collection"
}

WebSocket API

Connect to ws://host:port/ws

Messages

// Subscribe
{ type: 'subscribe', collection: 'posts' }

// Get CRDT state
{ type: 'crdt_get', collection: 'posts', docId: 'post-1' }

// Set value
{ type: 'crdt_set', collection: 'posts', docId: 'post-1', path: 'title', value: 'Hello' }

// Batch operations
{ type: 'crdt_ops', collection: 'posts', docId: 'post-1', operations: [...] }

// Presence
{ type: 'presence_join', collection: 'posts', docId: 'post-1', user: { name: 'John' } }

CRDT Types

VectorClock

Logical clock for causal ordering.

const clock = new VectorClock('node1');
clock.tick();
clock.merge(otherClock);
clock.compare(otherClock); // -1, 0, 1

LWWSet (Last-Writer-Wins Set)

Set with timestamps for conflict resolution.

const set = new LWWSet('node1');
set.add('item');
set.remove('item');
set.has('item');
set.toArray();

LWWMap

Map with LWW conflict resolution.

const map = new LWWMap('node1');
map.set('key', 'value');
map.get('key');
map.delete('key');
map.toObject();

RGA (Replicated Growable Array)

Ordered list/text with character-level sync.

const rga = new RGA('node1');
rga.insert(0, 'a');
rga.insert(1, 'b');
rga.delete(0);
rga.toArray();
rga.toString();

RichText

Rich text with formatting support.

const rt = new RichText('node1');
rt.insert(0, 'H', { bold: true });
rt.format(0, 5, { italic: true });
rt.toDelta(); // Quill-compatible

Production Deployment

PM2

# Initialize
npx kimdb init

# Start with PM2
pm2 start ecosystem.config.cjs

Docker

FROM node:18-alpine
WORKDIR /app
RUN npm install kimdb
COPY .env .
CMD ["npx", "kimdb", "start"]

Clustering

Multi-server synchronization via Redis Pub/Sub:

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│  Server 1   │────▶│    Redis    │◀────│  Server 2   │
│  :40000     │◀────│   Pub/Sub   │────▶│   :40001    │
└─────────────┘     └─────────────┘     └─────────────┘

# Server 1
KIMDB_PORT=40000 REDIS_ENABLED=true REDIS_HOST=redis-server npx kimdb start

# Server 2
KIMDB_PORT=40001 REDIS_ENABLED=true REDIS_HOST=redis-server npx kimdb start

Security

• Set KIMDB_API_KEY environment variable (required in production)
• API key is validated on all HTTP and WebSocket requests
• Use HTTPS/WSS in production

License

MIT