gitdataai/libs/git/http/rate_limit.rs
2026-04-14 19:02:01 +08:00

143 lines
4.1 KiB
Rust

//! HTTP rate limiting for git operations.
//!
//! Uses a token-bucket approach with per-IP and per-repo-write limits.
//! Cleanup runs every 5 minutes to prevent unbounded memory growth.
use std::collections::HashMap;
use std::sync::Arc;
use std::time::{Duration, Instant};
use tokio::sync::RwLock;
use tokio::time::interval;
#[derive(Debug, Clone)]
pub struct RateLimitConfig {
/// Requests allowed per window for read operations.
pub read_requests_per_window: u32,
/// Requests allowed per window for write operations.
pub write_requests_per_window: u32,
/// Window duration in seconds.
pub window_secs: u64,
}
impl Default for RateLimitConfig {
fn default() -> Self {
Self {
read_requests_per_window: 120,
write_requests_per_window: 30,
window_secs: 60,
}
}
}
#[derive(Debug)]
struct RateLimitBucket {
read_count: u32,
write_count: u32,
reset_time: Instant,
}
pub struct RateLimiter {
buckets: Arc<RwLock<HashMap<String, RateLimitBucket>>>,
config: RateLimitConfig,
}
impl RateLimiter {
pub fn new(config: RateLimitConfig) -> Self {
Self {
buckets: Arc::new(RwLock::new(HashMap::new())),
config,
}
}
pub async fn is_ip_read_allowed(&self, ip: &str) -> bool {
let key = format!("ip:read:{}", ip);
self.is_allowed(&key, self.config.read_requests_per_window)
.await
}
pub async fn is_ip_write_allowed(&self, ip: &str) -> bool {
let key = format!("ip:write:{}", ip);
self.is_allowed(&key, self.config.write_requests_per_window)
.await
}
pub async fn is_repo_write_allowed(&self, ip: &str, repo_path: &str) -> bool {
let key = format!("repo:write:{}:{}", ip, repo_path);
self.is_allowed(&key, self.config.write_requests_per_window)
.await
}
async fn is_allowed(&self, key: &str, limit: u32) -> bool {
let now = Instant::now();
let mut buckets = self.buckets.write().await;
let bucket = buckets
.entry(key.to_string())
.or_insert_with(|| RateLimitBucket {
read_count: 0,
write_count: 0,
reset_time: now + Duration::from_secs(self.config.window_secs),
});
if now >= bucket.reset_time {
bucket.read_count = 0;
bucket.write_count = 0;
bucket.reset_time = now + Duration::from_secs(self.config.window_secs);
}
// Use read_count for both read/write since we don't distinguish in bucket
if bucket.read_count >= limit {
return false;
}
bucket.read_count += 1;
true
}
pub async fn retry_after(&self, ip: &str) -> u64 {
let key_read = format!("ip:read:{}", ip);
let now = Instant::now();
let buckets = self.buckets.read().await;
if let Some(bucket) = buckets.get(&key_read) {
if now < bucket.reset_time {
return bucket.reset_time.saturating_duration_since(now).as_secs() as u64;
}
}
0
}
/// Start a background cleanup task that removes expired entries.
/// Should be spawned once at startup.
pub fn start_cleanup(self: Arc<Self>) -> tokio::task::JoinHandle<()> {
tokio::spawn(async move {
let mut ticker = interval(Duration::from_secs(300)); // every 5 minutes
loop {
ticker.tick().await;
let now = Instant::now();
let mut buckets = self.buckets.write().await;
buckets.retain(|_, bucket| now < bucket.reset_time);
}
})
}
}
#[cfg(test)]
mod tests {
use super::*;
#[tokio::test]
async fn test_rate_limit_allows_requests_up_to_limit() {
let limiter = Arc::new(RateLimiter::new(RateLimitConfig {
read_requests_per_window: 3,
write_requests_per_window: 1,
window_secs: 60,
}));
for _ in 0..3 {
assert!(limiter.is_ip_read_allowed("1.2.3.4").await);
}
assert!(!limiter.is_ip_read_allowed("1.2.3.4").await);
}
}