巧妙使用channel实现限流器

普通Limiter

特点:无时间限制,只要不超过数量就可通过

// 基于channel阻塞实现
// 缺点:阻塞无时间限制
type (
 ChannelLimiter struct {
 bufferChannel chan golang.PlaceholderType
 }
)
func NewChannelLimiter(limit int) *ChannelLimiter {
 return &ChannelLimiter{bufferChannel: make(chan golang.PlaceholderType, limit)}
}
func (l *ChannelLimiter) Allow() bool {
 select {
 case l.bufferChannel <- golang.Placeholder:
 return true
 default:
 return false
 }
}
func (l *ChannelLimiter) Release() bool {
 <-l.bufferChannel
 return true
}
func (l *ChannelLimiter) Close() {
 close(l.bufferChannel)
}

令牌桶Limiter

特点:令牌桶

// 令牌桶算法
// 产生令牌:均为间隔时间内(1秒)向指定桶中产生指定数量的令牌
// 消费令牌:从桶中获取令牌并消费
// 思路:通过channel阻塞原理来实现
type (
 TokenBucketLimiter struct {
 t *time.Ticker
 bucket chan golang.PlaceholderType
 doneC channel.DoneChan
 limit int
 rate int
 stop func()
 }
)
// rate:token put rates per second
// limit:max limit
func NewTokenBucketLimiter(rate, limit int) *TokenBucketLimiter {
 t := time.NewTicker(time.Second)
 doneC := channel.NewDoneChan()
 bucket := make(chan golang.PlaceholderType, limit)
 tbl := &TokenBucketLimiter{
 rate: rate,
 t: t,
 bucket: bucket,
 doneC: doneC,
 limit: limit,
 // only stop once
 stop: routine.DoOnce(func() {
 doneC.Stop()
 close(bucket)
 t.Stop()
 }),
 }
 // 定时放置令牌
 tbl.asyncPutTokens()
 return tbl
}
// 通过尝试put golang.Placeholder 来达到是否有令牌可消费
func (tbl *TokenBucketLimiter) Allow() bool {
 select {
 case tbl.bucket <- golang.Placeholder:
 return true
 case <-tbl.doneC.Done():
 return false
 default:
 return false
 }
}
func (tbl *TokenBucketLimiter) Close() {
 tbl.stop()
}
// 通过排空channel达到放置token的目的
func (tbl *TokenBucketLimiter) asyncPutTokens() {
 safe.GoRun(func() {
 for {
 select {
 case <-tbl.t.C:
 tbl.drain()
 case <-tbl.doneC.Done():
 return
 }
 }
 })
}
func (tbl *TokenBucketLimiter) drain() {
 for i := 0; i < tbl.limit; i++ {
 select {
 case <-tbl.bucket:
 default:
 return
 }
 }
}

本文来自:Segmentfault

感谢作者:Keson

查看原文:巧妙使用channel实现限流器

• 编辑
• 预览