Fix/online user count

[soham04010]

Description

This PR implements a robust Real-Time Online User Count system, replacing the previous database-polling method which caused stale data and rate-limiting issues.

The new system uses a Shared Memory Architecture to track active WebSocket connections instantly. It also resolves critical stability bugs including "ghost users" (users appearing online after leaving) and frontend connection loops.

🎯 Key Features & Fixes

1. Real-Time "Shared Memory" Architecture

• New File: Created backend/src/lib/roomState.ts to store live user counts in memory (Map<string, number>).
• Why: This allows the API to serve user counts instantly (0ms latency) without hitting the database, preventing performance bottlenecks and stale data.

2. Backend Stability (ChatServer.ts)

• Ghost User Fix: Implemented "Lazy Cleanup" logic. The server now actively checks for dead sockets whenever counting users, preventing the count from getting stuck (e.g., showing 2 users when only 1 is online).
• Strict Disconnect Handling: Added logic to ensure we only remove the specific socket that disconnected. This prevents a user from accidentally deleting their own new connection if they refresh the page quickly.
• Safe Iteration: Fixed a bug where iterating over a Set while deleting from it caused users to be skipped. Replaced with safe Array.from() iteration.

3. Frontend Reliability (useWebSocket.ts)

• Stale Closure Fix: Refactored the hook to use useRef for event callbacks (onMessage, onConnect). This prevents the WebSocket from disconnecting and reconnecting on every React render.
• Explicit Leave Signal: Added a cleanup function that sends a type: 'leave' message immediately before the component unmounts, ensuring the count drops the moment a user closes the tab.

4. Infrastructure (index.ts)

• Rate Limit Adjustment: Increased the API rate limit to 2000 requests per 15 minutes (previously ~100). This safely accommodates the Dashboard polling (every 3 seconds) without triggering false-positive 429 Too Many Requests errors.

Type of Change

• 🐛 Bug fix (non-breaking change which fixes an issue)
• ✨ New feature (non-breaking change which adds functionality)
• ⚡ Performance improvement
• 🧹 Code refactoring

How Has This Been Tested?

I have validated these changes with the following scenarios:

1. Instant Leave: Opened a room in a new tab and closed it immediately. Verified the count dropped instantly on the dashboard.
2. Refresh Persistence: Refreshed the chat page multiple times. Confirmed the user count remained stable (did not duplicate to "2 online").
3. Notifications: Verified that "User joined" and "User left" system messages appear correctly in the chat window.
4. Network Stability: Monitored the Network tab to ensure polling requests return 200 OK and no longer trigger 429 errors.

Screenshots

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• My changes generate no new warnings

Learning Outcomes

Implemented a hybrid state management system syncing WebSocket events with REST API polling using in-memory storage for high-performance real-time updates.