Minimalist flow programming in golang

The disadvantages brought by the traditional procedural coding method are obvious. We often have this experience. It takes a long time to suddenly pull back the code that has not been maintained for a period of time or other people's code to understand the original idea. If If there is a document or well-written annotation at this time, it may take a shorter time, but even so, many calling relationships must be confirmed repeatedly before dare to change.

The following is a piece of pseudo code describing the procedural coding method:

func A(){
    B()
    C()
}

func B(){    do something
    D()
}
func C(){    do something    
}
func D(){    do something
}
func main(){
    A()
}

Compare the writing method of the streaming style:

NewStream().
Next(A).
Next(B).
Next(D).
Next(C).
Go()

When the procedural style code call relationship is complex, the program Engineers need to act cautiously and carefully. Compared with flow-style code, it is cleaner and has a clear backbone, which has obvious advantages especially when responding to changes in requirements.

Java8 borrows lambda expressions to implement a relatively perfect streaming programming style. As a simple language, golang does not have an official streaming style package (maybe it already exists, maybe I am ignorant) ) is a bit of a pity.

I referred to the code of gorequest and implemented a relatively common streaming style package. The implementation principle is to form a task linked list, and each node saves the first node, the next node and the node. The callback function pointer that should be executed. After the streaming task is started, it will be executed one by one starting from the first node. If an exception is encountered, the streaming task will be terminated until the last one is executed, ending the task chain. Let’s take a look at the code first:

package Stream

import (    "errors"
    "fmt")/**
流式工作原理：
各个任务都过指针链表的方式组成一个任务链，这个任务链从第一个开始执行，直到最后一个
每一个任务节点执行完毕会将结果带入到下一级任务节点中。
每一个任务是一个Stream节点，每个任务节点都包含首节点和下一个任务节点的指针,
除了首节点，每个节都会设置一个回调函数的指针，用本节点的任务执行,
最后一个节点的nextStream为空,表示任务链结束。
**///定回调函数指针的类型type CB func(interface{}) (interface{}, error)//任务节点结构定义type Stream struct {    //任务链表首节点,其他非首节点此指针永远指向首节点
    firstStream *Stream    //任务链表下一个节点，为空表示任务结束
    nextStream *Stream    //当前任务对应的执行处理函数，首节点没有可执行任务，处理函数指针为空    cb CB
}/**
创建新的流
**/func NewStream() *Stream {    //生成新的节点
    stream := &Stream{}    //设置第一个首节点，为自己    //其他节点会调用run方法将从firs指针开始执行，直到next为空
    stream.firstStream = stream    //fmt.Println("new first", stream)
    return stream
}/**
流结束
arg为流初始参数，初始参数放在End方法中是考虑到初始参数不需在任务链中传递
**/func (this *Stream) Go(arg interface{}) (interface{}, error) {    //设置为任务链结束
    this.nextStream = nil    //fmt.Println("first=", this.firstStream, "second=", this.firstStream.nextStream)    //检查是否有任务节点存在，存在则调用run方法    //run方法是首先执行本任务回调函数指针，然后查找下一个任务节点，并调用run方法
    if this.firstStream.nextStream != nil {        return this.firstStream.nextStream.run(arg)
    } else {        //流式任务终止
        return nil, errors.New("Not found execute node.")
    }
}
func (this *Stream) run(arg interface{}) (interface{}, error) {    //fmt.Println("run,args=", args)    //执行本节点函数指针
    result, err := this.cb(arg)    //然后调用下一个节点的Run方法
    if this.nextStream != nil && err == nil {        return this.nextStream.run(result)
    } else {        //任务链终端，流式任务执行完毕
        return result, err
    }
}
func (this *Stream) Next(cb CB) *Stream {    //创建新的Stream，将新的任务节点Stream连接在后面
    this.nextStream = &Stream{}    //设置流式任务链的首节点
    this.nextStream.firstStream = this.firstStream    //设置本任务的回调函数指针
    this.nextStream.cb = cb    //fmt.Println("next=", this.nextStream)
    return this.nextStream
}

The following is an example of streaming. Here is the process from getting up in the morning to going out to work:

//起床func GetUP(arg interface{}) (interface{}, error) {
    t, _ := arg.(string)
    fmt.Println("铃铃.......", t, "###到时间啦，再不起又要迟到了！")    return "醒着的状态", nil
}//蹲坑func GetPit(arg interface{}) (interface{}, error) {
    s, _ := arg.(string)
    fmt.Println(s, "###每早必做的功课，蹲坑！")    return "舒服啦", nil
}//洗脸func GetFace(arg interface{}) (interface{}, error) {
    s, _ := arg.(string)
    fmt.Println(s, "###洗脸很重要！")    return "脸已经洗干净了，可以去见人了", nil
}//刷牙func GetTooth(arg interface{}) (interface{}, error) {
    s, _ := arg.(string)
    fmt.Println(s, "###刷牙也很重要！")    return "牙也刷干净了，可以放心的大笑", nil
}//吃早饭func GetEat(arg interface{}) (interface{}, error) {
    s, _ := arg.(string)
    fmt.Println(s, "###吃饭是必须的(需求变更了，原来的流程里没有，这次加上)")    return "吃饱饱了", nil
}//换衣服func GetCloth(arg interface{}) (interface{}, error) {
    s, _ := arg.(string)
    fmt.Println(s, "###还要增加一个换衣服的流程！")    return "找到心仪的衣服了", nil
}//出门func GetOut(arg interface{}) (interface{}, error) {
    s, _ := arg.(string)
    fmt.Println(s, "###一切就绪，可以出门啦！")    return "", nil

}
func main() {
    NewStream().
        Next(GetUP).
        Next(GetPit).
        Next(GetTooth).
        Next(GetFace).
        Next(GetEat).//需求变更了后加上的        Next(GetCloth).
        Next(GetOut).
        Go("2018年1月28日8点10分")
}

From the above code, streaming coding After a large task is decomposed into multiple small tasks, the style is very intuitive at the code level. You no longer have to work hard to find out which one calls which one. In addition, it is easier to change requirements. In the above example, eating breakfast is What was not implemented in the first version is that the customer said that he must eat in the morning, otherwise he would easily develop gallstones. The second version needs to be added. We need to complete the eating function and then add it to the response position. Relative to procedural coding, it is much simpler.

For more golang related technical articles, please visit the golang tutorial column!

The above is the detailed content of Minimalist flow programming in golang. For more information, please follow other related articles on the PHP Chinese website!