除了在goroutine之间安全的传递数据之外,在看了《Concurrency in Go》之后,感慨channel还有那么多模式可供使用,在个人的学习中总结了以下几种常用的模式
pipeline
概念
我们以爬虫为例,一般爬虫分为如下步骤:
抓取页面 -> 解析页面 -> 整合数据分析 -> 分析结果入库
如果你把上面所有的步骤都放在一个函数里面处理,那会是多难看,多难以维护,从解耦角度考虑,我们可以起四个进程,分别承担不同的角色,例如,进程1负责抓取页面, 进程2负责解析页面,等等,各个进程拿到一个数据后,交给下一个进程来处理,这就是pipeline的基本思想,每个角色只负责关心自己的东西
示例
给定一个数n,执行 (n2 + 1) 2的操作
func pipeline() {
generator := func(done chan interface{}, intergers ...int) <-chan int {
inStream := make(chan int)
go func() {
defer close(inStream)
for _, i := range intergers {
select {
case <-done:
return
case inStream <- i:
}
}
}()
return inStream
}
add := func(done <-chan interface{}, inStream <-chan int, increment int) <-chan int {
addInStream := make(chan int)
go func() {
defer close(addInStream)
for i := range inStream {
select {
case <-done:
return
case addInStream <- i + increment:
}
}
}()
return addInStream
}
multiply := func(done <-chan interface{}, inStream <-chan int, increment int) <-chan int {
multiplyInStream := make(chan int)
go func() {
defer close(multiplyInStream)
for i := range inStream {
select {
case <-done:
return
case multiplyInStream <- i * increment:
}
}
}()
return multiplyInStream
}
done := make(chan interface{})
defer close(done)
inStream := generator(done, []int{1, 2, 3, 4, 5, 6, 7}...)
pipeline := multiply(done, add(done, multiply(done, inStream, 2), 1), 2)
for v := range pipeline {
fmt.Println(v)
}
}
扇入扇出
在pipeline模型中,是一种高效的流式处理,但是假如pipeline中有a,b,c三个环节,b环节处理的特别慢,这时候就会影响到c环节的处理,如果增加b环节进程处理的数量,也就可以减弱b环节的慢处理对整个pipeline的影响,那么a->多个b的过程就是 扇入, 多个b环节输出数据到c环节,就是扇出
示例
func FanInFanOut() {
producer := func(intergers ...int) <-chan interface{} {
inStream := make(chan interface{})
go func() {
defer close(inStream)
for _, v := range intergers {
time.Sleep(5 * time.Second)
inStream <- v
}
}()
return inStream
}
fanIn := func(channels ...<-chan interface{},
) <-chan interface{} {
var wg sync.WaitGroup
multiplexStream := make(chan interface{})
multiplex := func(c <-chan interface{}) {
defer wg.Done()
for i := range c {
multiplexStream <- i
}
}
wg.Add(len(channels))
for _, c := range channels {
go multiplex(c)
}
go func() {
wg.Wait()
close(multiplexStream)
}()
return multiplexStream
}
consumer := func(inStream <-chan interface{}) {
for v := range inStream {
fmt.Println(v)
}
}
nums := runtime.NumCPU()
producerStreams := make([]<-chan interface{}, nums)
for i := 0; i < nums; i++ {
producerStreams[i] = producer(i)
}
consumer(fanIn(producerStreams...))
}
tee- channel
概念
假如你从channel中拿到了一条sql语句,这时候,你想对这条sql记录,分析并执行,那你就需要将这条sql分别转发给这三个任务对应的channel,tee-channel 就是做这个事情的
示例
func teeChannel() {
producer := func(intergers ...int) <-chan interface{} {
inStream := make(chan interface{})
go func() {
defer close(inStream)
for _, v := range intergers {
inStream <- v
}
}()
return inStream
}
tee := func(in <-chan interface{}) (_, _ <-chan interface{}) {
out1 := make(chan interface{})
out2 := make(chan interface{})
go func() {
defer close(out1)
defer close(out2)
for val := range in {
out1, out2 := out1, out2
for i := 0; i < 2; i++ {
select {
case out1 <- val:
out1 = nil
case out2 <- val:
out2 = nil
}
}
}
}()
return out1, out2
}
out1, out2 := tee(producer(1, 2, 3, 4, 5))
for val1 := range out1 {
fmt.Printf("out1: %v, out2: %v", val1, <-out2)
}
}
桥接channel
概念
无论是前面提到的pipeline还是扇入扇出,每个goroutine都是对一个channel进行消费,但是实际场景中,可能会有多个channel来供给我们消费,而作为消费者,我们不关心这些值是来自于哪个channel,这种情况下,处理一个充满channel的channel可能会很多。如果我们定义一个功能,可以将充满channel的channel拆解为一个简单的channel,这将使消费者更专注于手头的工作,这就是桥接channel的思想
示例
func bridge() {
gen := func() <-chan <-chan interface{} {
in := make(chan (<-chan interface{}))
go func() {
defer close(in)
for i := 0; i < 10; i++ {
stream := make(chan interface{}, 1)
stream <- i
close(stream)
in <- stream
}
}()
return in
}
bridge := func(in <-chan (<-chan interface{})) <-chan interface{} {
valStream := make(chan interface{})
go func() {
defer close(valStream)
for {
stream := make(<-chan interface{})
select {
case maybeStream, ok := <-in:
if ok == false {
return
}
stream = maybeStream
}
for val := range stream {
valStream <- val
}
}
}()
return valStream
}
for val := range bridge(gen()) {
fmt.Println(val)
}
}