github.com/andeya/ratelimit

Efficient token-bucket-based rate limiter package.


License
LGPL-3.0
Install
go get github.com/andeya/ratelimit

Documentation

ratelimit

-- import "github.com/juju/ratelimit"

The ratelimit package provides an efficient token bucket implementation. See http://en.wikipedia.org/wiki/Token_bucket.

Usage

func Reader

func Reader(r io.Reader, bucket *Bucket) io.Reader

Reader returns a reader that is rate limited by the given token bucket. Each token in the bucket represents one byte.

func Writer

func Writer(w io.Writer, bucket *Bucket) io.Writer

Writer returns a writer that is rate limited by the given token bucket. Each token in the bucket represents one byte.

type Bucket

type Bucket struct {
}

Bucket represents a token bucket that fills at a predetermined rate. Methods on Bucket may be called concurrently.

func NewBucket

func NewBucket(fillInterval time.Duration, capacity int64) *Bucket

NewBucket returns a new token bucket that fills at the rate of one token every fillInterval, up to the given maximum capacity. Both arguments must be positive. The bucket is initially full.

func NewBucketWithQuantum

func NewBucketWithQuantum(fillInterval time.Duration, capacity, quantum int64) *Bucket

NewBucketWithQuantum is similar to NewBucket, but allows the specification of the quantum size - quantum tokens are added every fillInterval.

func NewBucketWithRate

func NewBucketWithRate(rate float64, capacity int64) *Bucket

NewBucketWithRate returns a token bucket that fills the bucket at the rate of rate tokens per second up to the given maximum capacity. Because of limited clock resolution, at high rates, the actual rate may be up to 1% different from the specified rate.

func (*Bucket) Available

func (tb *Bucket) Available() int64

Available returns the number of available tokens. It will be negative when there are consumers waiting for tokens. Note that if this returns greater than zero, it does not guarantee that calls that take tokens from the buffer will succeed, as the number of available tokens could have changed in the meantime. This method is intended primarily for metrics reporting and debugging.

func (*Bucket) Rate

func (tb *Bucket) Rate() float64

Rate returns the fill rate of the bucket, in tokens per second.

func (*Bucket) Take

func (tb *Bucket) Take(count int64) time.Duration

Take takes count tokens from the bucket without blocking. It returns the time that the caller should wait until the tokens are actually available.

Note that if the request is irrevocable - there is no way to return tokens to the bucket once this method commits us to taking them.

func (*Bucket) TakeAvailable

func (tb *Bucket) TakeAvailable(count int64) int64

TakeAvailable takes up to count immediately available tokens from the bucket. It returns the number of tokens removed, or zero if there are no available tokens. It does not block.

func (*Bucket) TakeMaxDuration

func (tb *Bucket) TakeMaxDuration(count int64, maxWait time.Duration) (time.Duration, bool)

TakeMaxDuration is like Take, except that it will only take tokens from the bucket if the wait time for the tokens is no greater than maxWait.

If it would take longer than maxWait for the tokens to become available, it does nothing and reports false, otherwise it returns the time that the caller should wait until the tokens are actually available, and reports true.

func (*Bucket) Wait

func (tb *Bucket) Wait(count int64)

Wait takes count tokens from the bucket, waiting until they are available.

func (*Bucket) WaitMaxDuration

func (tb *Bucket) WaitMaxDuration(count int64, maxWait time.Duration) bool

WaitMaxDuration is like Wait except that it will only take tokens from the bucket if it needs to wait for no greater than maxWait. It reports whether any tokens have been removed from the bucket If no tokens have been removed, it returns immediately.