jansuchanek/nchat

Reusable real-time admin chat extension for Nette Framework with WebSocket support (Soketi/Pusher), typing indicators, presence, read receipts, and optional AI responder.

v1.1.0 2026-03-19 09:59 UTC

Requires

• php: >=8.1
• nette/application: ^3.2
• nette/di: ^3.1
• nette/http: ^3.2

Suggests

• doctrine/dbal: For raw SQL queries
• doctrine/orm: For entity management

MIT

README

Reusable real-time chat extension for Nette Framework admin panels with WebSocket support (Soketi/Pusher), typing indicators, presence detection, read receipts, emoji reactions, sound alerts, and browser push notifications.

Features

• 💬 Broadcast + DM + Group chat
• ⚡ WebSocket real-time (Soketi/Pusher) with polling fallback
• ✍️ Typing indicators — "Jan píše…"
• 👥 Online presence — instant online/offline
• ✓✓ Read receipts — DM delivery confirmation
• 😂 Emoji reactions — 👍❤️😂😮😢🙏
• 🔔 Sound alerts — Web Audio API ping (toggleable)
• 📢 Push notifications — Browser Notification API
• 🤖 AI Bot integration — pluggable via AiResponderInterface
• 🗄️ Database agnostic — ChatStorageInterface (DBAL included)

Installation

{
    "repositories": [
        {"type": "vcs", "url": "[e-mail]:JanSuchanek/nchat.git"}
    ],
    "require": {
        "jansuchanek/nchat": "dev-main"
    }
}

Configuration

extensions:
    nchat: NChat\ChatExtension

nchat:
    websocket:
        host: soketi
        port: 6001
        appId: my-app
        key: my-key
        secret: my-secret
    storage: NChat\Storage\DbalChatStorage  # or your own
    ai:
        enabled: true
        responder: App\Model\MyAiResponder  # implements AiResponderInterface

Usage

Add the trait to any presenter:

final class AdminPresenter extends BasePresenter
{
    use NChat\Presenter\ChatPresenterTrait;
}

Then add chat action names to your allowed actions:

$openActions = array_merge(['dashboard'], self::getChatActionNames());

Custom Database Storage

Implement ChatStorageInterface for any database:

class SqliteChatStorage implements ChatStorageInterface
{
    public function saveMessage(array $data): int { /* ... */ }
    public function fetchMessages(int $userId, array $criteria = []): array { /* ... */ }
    // ...8 methods total
}

Custom AI Responder

class MyAiResponder implements AiResponderInterface
{
    public function respond(string $context): string { /* call OpenAI/Claude/Gemini */ }
    public function getSystemPrompt(): string { return 'You are...'; }
    public function getBotName(): string { return '🤖 Bot'; }
}

License

MIT