GOF:运用共享技术有效地支持大量细粒度的对象。
解释一下概念:也就是说在一个系统中如果有多个相同的对象,那么只共享一份就可以了,不必每个都去实例化一个对象。比如说(这里引用GOF书中的例子)一个文本系统,每个字母定一个对象,那么大小写字母一共就是52个,那么就要定义52个对象。如果有一个1M的文本,那么字母是何其的多,如果每个字母都定义一个对象那么内存早就爆了。那么如果要是每个字母都共享一个对象,那么就大大节约了资源。
在Flyweight模式中,由于要产生各种各样的对象,所以在Flyweight(享元)模式中常出现Factory模式。Flyweight的内部状态是用来共享的,Flyweight factory负责维护一个对象存储池(Flyweight Pool)来存放内部状态的对象。Flyweight模式是一个提高程序效率和性能的模式,会大大加快程序的运行速度.应用场合很多,下面举个例子:
先定义一个抽象的Flyweight类:
package Flyweight;
public abstract class Flyweight
...
{
public abstract void operation();
}//end abstract class Flyweight
在实现一个具体类:
package Flyweight;
public class ConcreteFlyweight extends Flyweight
...
{
private String string;
public ConcreteFlyweight(String str)
...
{
string = str;
}//end ConcreteFlyweight(...)
public void operation()
...
{
System.out.println("Concrete---Flyweight : " + string);
}//end operation()
}//end class ConcreteFlyweight
实现一个工厂方法类:
package Flyweight;
import java.util.Hashtable;
public class FlyweightFactory
...
{
private Hashtable flyweights = new Hashtable();//----------------------------1
public FlyweightFactory() ...{}
public Flyweight getFlyWeight(Object obj)
...
{
Flyweight flyweight = (Flyweight) flyweights.get(obj);//----------------2
if(flyweight == null) ...{//---------------------------------------------------3
//产生新的ConcreteFlyweight
flyweight = new ConcreteFlyweight((String)obj);
flyweights.put(obj, flyweight);//--------------------------------------5
}
return flyweight;//---------------------------------------------------------6
}//end GetFlyWeight(...)
public int getFlyweightSize()
...
{
return flyweights.size();
}
}//end class FlyweightFactory
这个工厂方法类非常关键,这里详细解释一下:
在1处定义了一个Hashtable用来存储各个对象;在2处选出要实例化的对象,在6处将该对象返回,如果在Hashtable中没有要选择的对象,此时变量flyweight为null,产生一个新的flyweight存储在Hashtable中,并将该对象返回。
最后看看Flyweight的调用:
package Flyweight;
import java.util.Hashtable;
public class FlyweightPattern ...{
FlyweightFactory factory = new FlyweightFactory();
Flyweight fly1;
Flyweight fly2;
Flyweight fly3;
Flyweight fly4;
Flyweight fly5;
Flyweight fly6;
/** *//** Creates a new instance of FlyweightPattern */
public FlyweightPattern() ...{
fly1 = factory.getFlyWeight("Google");
fly2 = factory.getFlyWeight("Qutr");
fly3 = factory.getFlyWeight("Google");
fly4 = factory.getFlyWeight("Google");
fly5 = factory.getFlyWeight("Google");
fly6 = factory.getFlyWeight("Google");
}//end FlyweightPattern()
public void showFlyweight()
...
{
fly1.operation();
fly2.operation();
fly3.operation();
fly4.operation();
fly5.operation();
fly6.operation();
int objSize = factory.getFlyweightSize();
System.out.println("objSize = " + objSize);
}//end showFlyweight()
public static void main(String[] args)
...
{
System.out.println("The FlyWeight Pattern!");
FlyweightPattern fp = new FlyweightPattern();
fp.showFlyweight();
}//end main(...)
}//end class FlyweightPattern
下面是运行结果:
Concrete---Flyweight : Google
Concrete---Flyweight : Qutr
Concrete---Flyweight : Google
Concrete---Flyweight : Google
Concrete---Flyweight : Google
Concrete---Flyweight : Google
objSize = 2
我们定义了6个对象,其中有5个是相同的,按照Flyweight模式的定义“Google”应该共享一个对象,在实际的对象数中我们可以看出实际的对象却是只有2个。
Flyweight(享元)模式是如此的重要,因为它能帮你在一个复杂的系统中大量的节省内存空间。在GOF的书中举了文本处理的例子,我觉得非常恰当。那么,在Java中String这个类型比较特殊,为什么呢:
String a = "hello";
String b = "hello";
if(a == b)
System.out.println("OK");
else
System.out.println("Error");
输出结果是:OK。可以看出if条件比较的是两a和b的地址,也可以说是内存空间。
可以共享的对象,也就是说返回的同一类型的对象其实是同一实例,当客户端要求生成一个对象时,工厂会检测是否存在此对象的实例,如果存在那么直接返回此对象实例,如果不存在就创建一个并保存起来,这点有些单例模式的意思。通常工厂类会有一个集合类型的成员变量来用以保存对象,如hashtable,vector等。在java中,数据库连接池,线程池等即是用享元模式的应用。 |