Feat/rate limit

[Ataba29]

feat/rateLimit

What's in this PR

This PR introduces a full rate limiting layer to protect the server from abusive traffic using the Token Bucket algorithm — the same approach used by Stripe, GitHub, and AWS.

RateLimiter (src/Limit/Limiter.h/.cpp)

• Per-IP token bucket: each IP starts with a configurable number of tokens, each request costs one token, tokens refill at a fixed rate per second
• Global request cap: limits total requests per second across all IPs regardless of source
• Lazy token refill: no background thread needed, tokens are calculated on demand based on elapsed time
• Thread safe: separate mutexes for per-IP map and global window to minimize contention

Server integration (src/Server/Server.cpp)

• Rate check happens in acceptClients before any job is handed to the thread pool
• Blocked clients are dropped immediately with CloseSocket and continue — no worker thread is consumed
• Client IP extracted using getpeername + inet_ntop (cross-platform, works on both Windows and Linux)

Unit tests (src/Tests/test_ratelimiter.cpp)

• Normal traffic from multiple IPs is allowed
• Single IP gets blocked after exhausting token bucket
• Blocked IP does not affect other IPs
• Global limit blocks all IPs once reached
• Tokens refill correctly over time
• Concurrent requests from same IP are handled safely without race conditions

Verified on both Windows (local) and Linux (Docker container)

[razimograbi]

Overall good job !
Very clean code

[razimograbi]

We reset the global bucket count each second? Why?

[Ataba29]

Because the actual hardware is fast enough to handle a 1-second reset time, with the correct max global traffic per second adjusted, it won't harm the server

[razimograbi]

Approved