共计 8087 个字符,预计需要花费 21 分钟才能阅读完成。
手撸 golang 行为型设计模式 状态模式
缘起
最近温习设计模式
拜读谭勇德的 << 设计模式就该这样学 >>
本系列笔记拟采纳 golang 练习之
状态模式
状态模式(State Pattern)也叫作状态机模式(State Machine Pattern),容许对象在外部状态产生扭转时扭转它的行为,对象看起来如同批改了它的类,属于行为型设计模式。状态模式次要蕴含 3 个角色。(1)环境类角色(Context):定义客户端须要的接口,外部保护一个以后状态实例,并负责具体状态的切换。(2)形象状态角色(IState):定义该状态下的行为,能够有一个或多个行为。(3)具体状态角色(ConcreteState):具体实现该状态对应的行为,并且在须要的状况下进行状态切换。(摘自 谭勇德 << 设计模式就该这样学 >>)
场景
- 某业务零碎, 须要对接某对象存储系统, 以对立治理大量的文件数据
- 因为该对象存储系统只提供了 http api 文档, 因而须要自行封装文件读写 api
-
码农王二狗接到 Leader 张阿蛋下达的开发工作, 啃了几遍文档, 感觉胸有成竹
- 王二狗: 张哥, 这个读写 api, 是不是就写一个门面, 一个办法封装一个 api 就行了, 感觉性能不简单
- 张阿蛋: 基本上是这样的. 不过你有没有留神到, 文档上有注意事项, 就是关上文件的时候, 必须确定文件的应用形式?
- 王二狗: 啥? 哦 … 我看看, 还真是有这么一条, 不过这个有影响吗, 不就是多传一个参数吗?
- 张阿蛋: 当然有影响, 如果你以读形式关上, 那么后续是不承受写入的. 如果你以写形式关上, 后续是不能读取的.
- 王二狗: 哦, 那预计要好好封装一下 http api 的报错信息?
- 张阿蛋: 这个是必定要的, 通信不免有出错的时候. 然而, 兴许咱们能够用 状态模式, 提前回绝一些非法操作
- 王二狗: … 不明觉厉, 能对我间接点吗?
- 张阿蛋: 将文件流对象的办法委托给 ” 未关上 ”, “ 读取中 ”, “ 写入中 ”, 和 ” 已敞开 ” 四种状态. 每种状态只响应以后状态下容许的操作, 其余操作间接返回 error.
- 王二狗: 张哥, 强!
设计
- IFileStream: 文件流 API 接口
- iFileStreamContext: 文件流上下文接口, 仅外部应用
- iFileStreamState: 文件流状态接口, 仅外部应用
- tMockFileStream: 虚构的文件流 API 实现类, 实现 iFileStream 接口 + iFileStreamContext 接口
- tDefaultState: 默认状态 – 未关上状态. 该状态下只容许关上 / 敞开操作
- tReadingState: 读取状态. 该状态下只容许读取 / 敞开操作
- tWritingState: 写入状态. 该状态下只容许写入 / 敞开操作
- tClosedState: 敞开状态. 该状态下只容许敞开操作
单元测试
state_pattern_test.go
package behavioral_patterns
import (
"learning/gooop/behavioral_patterns/state"
"testing"
)
func Test_StatePattern(t *testing.T) {fnTestFileStream := func(fs state.IFileStream, readonly bool) {
if readonly {e := fs.OpenRead()
if e != nil {t.Log(e)
}
e = fs.OpenWrite()
if e != nil {t.Log(e)
}
} else {e := fs.OpenWrite()
if e != nil {t.Log(e)
}
e = fs.OpenRead()
if e != nil {t.Log(e)
}
}
buffer := make([]byte, 8192)
n,e := fs.Read(buffer)
if e != nil {t.Log(e)
} else {t.Logf("%v bytes read", n)
}
n, e = fs.Write(buffer)
if e != nil {t.Log(e)
} else {t.Logf("%v bytes written", n)
}
e = fs.Close()
if e != nil {t.Log(e)
}
}
fnTestFileStream(state.NewMockFileStream("read-only.txt"), true)
fnTestFileStream(state.NewMockFileStream("write-only.txt"), false)
}
测试输入
$ go test -v state_pattern_test.go
=== RUN Test_StatePattern
tDefaultState.OpenRead, file=read-only.txt
tMockFileStream.Switch, *state.tDefaultState => *state.tReadingState
state_pattern_test.go:18: tReadingState.OpenWrite, already reading read-only.txt
tReadingState.Read, file=read-only.txt, iBytesRead=8192
state_pattern_test.go:37: 8192 bytes read
state_pattern_test.go:42: tReadingState.Write, cannot write to read-only.txt
tReadingState.Close, file=read-only.txt, iBytesRead=8192
tMockFileStream.Switch, *state.tReadingState => *state.tClosedState
tDefaultState.OpenWrite, file=write-only.txt
tMockFileStream.Switch, *state.tDefaultState => *state.tWritingState
state_pattern_test.go:28: tWritingState.OpenRead, already writing write-only.txt
state_pattern_test.go:35: tWritingState.Read, cannot read write-only.txt
tWritingState.Write, file=write-only.txt, written=8192
state_pattern_test.go:44: 8192 bytes written
tWritingState.Close, file=write-only.txt, written=8192
tMockFileStream.Switch, *state.tWritingState => *state.tClosedState
--- PASS: Test_StatePattern (0.00s)
PASS
ok command-line-arguments 0.002s
IFileStream.go
文件流 API 接口
package state
import "io"
type IFileStream interface {
io.ReadWriteCloser
OpenRead() error
OpenWrite() error}
iFileStreamContext.go
文件流上下文接口, 仅外部应用
package state
type iFileStreamContext interface {File() string
Switch(state iFileStreamState)
}
iFileStreamState.go
文件流状态接口, 仅外部应用
package state
type iFileStreamState interface {IFileStream}
tMockFileStream.go
虚构的文件流 API 实现类, 实现 iFileStream 接口 + iFileStreamContext 接口
package state
import (
"fmt"
"reflect"
)
type tMockFileStream struct {
state iFileStreamState
file string
}
func NewMockFileStream(file string) IFileStream {
fs := &tMockFileStream{
nil,
file,
}
fs.state = newDefaultState(fs)
return fs
}
func (me *tMockFileStream) File() string {return me.file}
func (me *tMockFileStream) Switch(st iFileStreamState) {fmt.Printf("tMockFileStream.Switch, %s => %s\n", reflect.TypeOf(me.state).String(), reflect.TypeOf(st).String())
me.state = st
}
func (me *tMockFileStream) OpenRead() error {return me.state.OpenRead()
}
func (me *tMockFileStream) OpenWrite() error {return me.state.OpenWrite()
}
func (me *tMockFileStream) Read(p []byte) (n int, e error) {return me.state.Read(p)
}
func (me *tMockFileStream) Write(p []byte) (n int, e error) {return me.state.Write(p)
}
func (me *tMockFileStream) Close() error {return me.state.Close()
}
tDefaultState.go
默认状态 – 未关上状态. 该状态下只容许关上 / 敞开操作
package state
import (
"errors"
"fmt"
)
type tDefaultState struct {context iFileStreamContext}
func newDefaultState(context iFileStreamContext) iFileStreamState {
return &tDefaultState{context,}
}
func (me *tDefaultState) OpenRead() error {fmt.Printf("tDefaultState.OpenRead, file=%s\n", me.context.File())
me.context.Switch(newReadingState(me.context))
return nil
}
func (me *tDefaultState) OpenWrite() error {fmt.Printf("tDefaultState.OpenWrite, file=%s\n", me.context.File())
me.context.Switch(newWritingState(me.context))
return nil
}
func (me *tDefaultState) Read(p []byte) (n int, err error) {return 0, errors.New(fmt.Sprintf("tDefaultState.Read, file not opened, %s", me.context.File()))
}
func (me *tDefaultState) Write(p []byte) (n int, err error) {return 0, errors.New(fmt.Sprintf("tDefaultState.Write, file not opened, %s", me.context.File()))
}
func (me *tDefaultState) Close() error {fmt.Printf("tDefaultState.Close, file=%s\n", me.context.File())
me.context.Switch(newClosedState(me.context))
return nil
}
tReadingState.go
读取状态. 该状态下只容许读取 / 敞开操作
package state
import (
"errors"
"fmt"
)
type tReadingState struct {
context iFileStreamContext
iBytesRead int
}
func newReadingState(context iFileStreamContext) iFileStreamState {
return &tReadingState{
context,
0,
}
}
func (me *tReadingState) OpenRead() error {return errors.New(fmt.Sprintf("tReadingState.OpenRead, already reading %s", me.context.File()))
}
func (me *tReadingState) OpenWrite() error {return errors.New(fmt.Sprintf("tReadingState.OpenWrite, already reading %s", me.context.File()))
}
func (me *tReadingState) Read(p []byte) (n int, err error) {size := len(p)
me.iBytesRead += size
fmt.Printf("tReadingState.Read, file=%s, iBytesRead=%v\n", me.context.File(), me.iBytesRead)
return size, nil
}
func (me *tReadingState) Write(p []byte) (n int, err error) {return 0, errors.New(fmt.Sprintf("tReadingState.Write, cannot write to %s", me.context.File()))
}
func (me *tReadingState) Close() error {fmt.Printf("tReadingState.Close, file=%s, iBytesRead=%v\n", me.context.File(), me.iBytesRead)
me.context.Switch(newClosedState(me.context))
return nil
}
tWritingState.go
写入状态. 该状态下只容许写入 / 敞开操作
package state
import (
"errors"
"fmt"
)
type tWritingState struct {
context iFileStreamContext
written int
}
func newWritingState(context iFileStreamContext) iFileStreamState {
return &tWritingState{
context,
0,
}
}
func (me *tWritingState) OpenRead() error {return errors.New(fmt.Sprintf("tWritingState.OpenRead, already writing %s", me.context.File()))
}
func (me *tWritingState) OpenWrite() error {return errors.New(fmt.Sprintf("tWritingState.OpenWrite, already writing %s", me.context.File()))
}
func (me *tWritingState) Read(p []byte) (n int, err error) {return 0, errors.New(fmt.Sprintf("tWritingState.Read, cannot read %s", me.context.File()))
}
func (me *tWritingState) Write(p []byte) (n int, err error) {size := len(p)
me.written += size
fmt.Printf("tWritingState.Write, file=%s, written=%v\n", me.context.File(), me.written)
return size, nil
}
func (me *tWritingState) Close() error {fmt.Printf("tWritingState.Close, file=%s, written=%v\n", me.context.File(), me.written)
me.context.Switch(newClosedState(me.context))
return nil
}
tClosedState.go
敞开状态. 该状态下只容许敞开操作
package state
import (
"errors"
"fmt"
)
type tClosedState struct {context iFileStreamContext}
func newClosedState(context iFileStreamContext) iFileStreamState {
return &tClosedState{context,}
}
func (me *tClosedState) OpenRead() error {return errors.New(fmt.Sprintf("tClosedState.OpenRead, file(%s) already closed", me.context.File()))
}
func (me *tClosedState) OpenWrite() error {return errors.New(fmt.Sprintf("tClosedState.OpenWrite, file(%s) already closed", me.context.File()))
}
func (me *tClosedState) Read(p []byte) (n int, e error) {return 0, errors.New(fmt.Sprintf("tClosedState.Read, file(%s) already closed", me.context.File()))
}
func (me *tClosedState) Write(p []byte) (n int, e error) {return 0, errors.New(fmt.Sprintf("tClosedState.Write, file(%s) already closed", me.context.File()))
}
func (me *tClosedState) Close() error {return nil}
状态模式小结
状态模式的长处(1)构造清晰:将状态独立为类,打消了冗余的 if...else 或 switch...case 语句。(2)将状态转换显示化:通常对象外部都是应用数值类型来定义状态的,状态的切换通过赋值进行体现,不够直观;而应用状态类,当切换状态时,是以不同的类进行示意的,转换目标更加明确。(3)状态类职责明确且具备扩展性。状态模式的毛病(1)类收缩:如果一个事物具备很多状态,则会造成状态类太多。(2)状态模式的构造与实现都较为简单,如果使用不当,将导致程序结构和代码的凌乱。(3)状态模式对开闭准则的反对并不太好,对于能够切换状态的状态模式,减少新的状态类须要批改那些负责状态转换的源码,否则无奈切换到新增状态,而且批改某个状态类的行为也须要批改对应类的源码。(摘自 谭勇德 << 设计模式就该这样学 >>)
(end)
正文完