ratelimit

new_token_bucket(rate_per_sec: 100.0, burst: 20) creates a thread-safe token bucket. The bucket starts full; allow() consumes one token or returns false when empty. Tokens refill at the positive, finite rate_per_sec, capped at the positive burst capacity. stats() returns the numbers of allowed and denied requests.

The default clock is monotonic with millisecond resolution. new_token_bucket_with_clock(config, now) accepts a millisecond clock for deterministic tests. Backward clock readings do not refill the bucket or move its last-refill timestamp backward. The clock callback runs while the bucket mutex is locked; it must not reenter the same bucket.

fn new_token_bucket(config TokenBucketConfig) &TokenBucket

new_token_bucket creates a full bucket reading a monotonic millisecond clock.

fn new_token_bucket_with_clock(config TokenBucketConfig, now fn () i64) &TokenBucket

new_token_bucket_with_clock is new_token_bucket with an injectable millisecond clock, for deterministic tests and custom time sources.

struct BucketStats {
pub:
	allowed u64
	denied  u64
}

BucketStats counts allowed and denied requests.

struct TokenBucket {
mut:
	mutex   &sync.Mutex
	rate    f64
	cap     f64
	tokens  f64
	last    i64 // millisecond timestamp of the last refill
	now     fn () i64
	allowed u64
	denied  u64
}

TokenBucket is a thread-safe token-bucket rate limiter: tokens refill at rate_per_sec up to burst, and each allowed request costs one token. It is the V-idiomatic equivalent of .NET's token-bucket limiter, for veb/fasthttp handlers and metered API clients.

fn (mut b TokenBucket) allow() bool

allow consumes one token if available, refilling first from the elapsed time. It returns false (counted in stats) when empty.

fn (b &TokenBucket) stats() BucketStats

stats returns the allowed/denied counters.

struct TokenBucketConfig {
pub:
	// sustained rate in tokens (requests) per second
	rate_per_sec f64
	// maximum burst: bucket capacity and initial tokens
	burst int
}

TokenBucketConfig configures a TokenBucket.