设计模式(十三)-State

设计动机

1. 在软件构建过程中，某些对象的状态如果改变，其行为也会随之，而发生变化，比如文档处于只读状态，其支持的行为和读写状态支持的行为就可能完全不同。
2. 如何在运行时根据对象的状态来透明地更改对象的行为?而不会为对象操作和状态转化之间引入紧耦合?

定义

允许一个对象在其内部状态改变时改变它的行为。从而使对象看起来似乎修改了其行为。

结构

例子

传统

enum NetworkState{
    NetworkOpen,
    NetworkClose,
    NetworkConnect,
};
class NetworkProcess{
    NetworkState state;
public:
    void operator1(){
        if(state == NetworkOpen){
            //do something
            /*
             *******
            */
            //state change
            state  = NetworkClose;
        }else if(state == NetworkClose){
            //do something
            /*
             *******
            */
            //state change
            state  = NetworkConnect;
        }else if(state == NetworkConnect){
            //do something
            /*
             *******
            */
            //state change
            state  = NetworkOpen;
        }
    }
    void operator2(){
        if(state == NetworkOpen){
            /*
             *******
            */
        }else if(state == NetworkClose){
            /*
             *******
            */
            state  = NetworkConnect;
        }else if(state == NetworkConnect){
            /*
             *******
            */
            state  = NetworkOpen;
        }
    }
}

状态模式

class OpenState;
class CloseState;
class ConnectState;
class NetworkState{
public:
    NetworkState* pNext; //next state

    //提供接口，等待子类重写
    virtual void operator1() = 0;
    virtual void operator2() = 0;
    
    virtual ~NetworkState(){}
}
//开启状态
class OpenState:public NetworkState{
    //单例，只有一个状态实例
    static NetworkState* m_instance;
public:
    static NetworkState* getInstance(){
        if(m_instance == nullptr){
            m_instance = new OpenState();
        }
        return m_instance;
    }
    void operator1(){
        //do something
            /*
             *******
            */
        //state change
        pNext  = CloseState::getInstance();
    }
}

//关闭状态
class CloseState:public NetworkState{
    //单例，只有一个状态实例
    static NetworkState* m_instance;
public:
    static NetworkState* getInstance(){
        if(m_instance == nullptr){
            m_instance = new CloseState();
        }
        return m_instance;
    }
    void operator1(){
        //do something
            /*
             *******
            */
        //state change
        pNext  = ConnectState::getInstance();
    }
}

//连接状态
class ConnectState:public NetworkState{
    //单例，只有一个状态实例
    static NetworkState* m_instance;
public:
    static NetworkState* getInstance(){
        if(m_instance == nullptr){
            m_instance = new ConnectState();
        }
        return m_instance;
    }
    void operator1(){
        //do something
            /*
             *******
            */
        //state change
        pNext  = OpenState::getInstance();
    }
}

NetworkState* OpenState::m_instance = nullptr;
NetworkState* CloseState::m_instance = nullptr;
NetworkState* ConnectState::m_instance = nullptr;

//使用
class NetworkProcessor{
    NetworkState* pState;
public:
    // 构造函数
    NetworkProcessor(NetworkState* state):pState(state){

    }
    void operator1(){
        //...
        pState->operator1();
        pState = pState->pNext;
    }
}

总结

1. State模式将所有与一个特定状态相关的行为都放入一个State的子类对象中，在对象状态切换时，切换相应的对象;但同时维持State的接口，这样实现了具体操作与状态转换之间的解耦。
2. 为不同的状态引入不同的对象使得状态转换变得更加明确，而且可以保证不会出现状态不一致的情况，因为转换是原子性的——即要么彻底转换过来，要么不转换。
3. 如果State对象没有实例变量，那么各个上下文可以共享同一个State对象，从而节省对象开销。