4.ThreadLocal实现线程范围的共享变量

当没有ThreadLocal的时候，我们是这样实现线程范围内的变量共享的

import java.util.HashMap;import java.util.Map;import java.util.Random;/** *map为每一个线程都放入了独立的数据，线程范围内的数据共享  */public class ThreadScopeShareData {private static int data = 0;private static Map<Thread, Integer> threadData = new HashMap<Thread, Integer>();public static void main(String[] args) {for(int i=0;i<2;i++){new Thread(new Runnable(){public void run() {//保证数据不覆盖int data = new Random().nextInt();System.out.println(Thread.currentThread().getName() + " has put data :" + data);threadData.put(Thread.currentThread(), data);new A().get();new B().get();}}).start();}}static class A{public void get(){int data = threadData.get(Thread.currentThread());System.out.println("A from " + Thread.currentThread().getName() + " get data :" + data);}}static class B{public void get(){int data = threadData.get(Thread.currentThread());System.out.println("B from " + Thread.currentThread().getName() + " get data :" + data);}}}

输出的结果：

Thread-0 has put data :1313822882Thread-1 has put data :-1822590364A from Thread-0 get data :1313822882A from Thread-1 get data :-1822590364B from Thread-0 get data :1313822882B from Thread-1 get data :-1822590364

也就是AB两个模块都共享线程里的数据，且互相独立。

①.线程范围内共享数据的示意图

见上图解释ThreadLocal的作用和目的：用于实现线程内的数据共享，即对于相同的程序代码，多个模块在同一个线程中运行时要共享一份数据，而在另外线程中运行时又共享另外一份数据。
②每个线程调用全局ThreadLocal对象的set方法，就相当于往其内部的map中增加一条记录，key分别是各自的线程，value是各自的set方法传进去的值。在线程结束时可以调用ThreadLocal.clear()方法，这样会更快释放内存，不调用也可以，因为线程结束后也可以自动释放相关的ThreadLocal变量。
③ThreadLocal的应用场景：
a.订单处理包含一系列操作：减少库存量、增加一条流水台账、修改总账，这几个操作要在同一个事务中完成，通常也即同一个线程中进行处理，如果累加公司应收款的操作失败了，则应该把前面的操作回滚，否则，提交所有操作，这要求这些操作使用相同的数据库连接对象，而这些操作的代码分别位于不同的模块类中。
b. 银行转账包含一系列操作： 把转出帐户的余额减少，把转入帐户的余额增加，这两个操作要在同一个事务中完成，它们必须使用相同的数据库连接对象，转入和转出操作的代码分别是两个不同的帐户对象的方法。
c.例如Strut2的ActionContext，同一段代码被不同的线程调用运行时，该代码操作的数据是每个线程各自的状态和数据，对于不同的线程来说，getContext方法拿到的对象都不相同，对同一个线程来说，不管调用getContext方法多少次和在哪个模块中getContext方法，拿到的都是同一个。
④实验案例：定义一个全局共享的ThreadLocal变量，然后启动多个线程向该ThreadLocal变量中存储一个随机值，接着各个线程调用另外其他多个类的方法，这多个类的方法中读取这个ThreadLocal变量的值，就可以看到多个类在同一个线程中共享同一份数据。
⑤实现对ThreadLocal变量的封装，让外界不要直接操作ThreadLocal变量。
对基本类型的数据的封装，这种应用相对很少见。
对对象类型的数据的封装，比较常见，即让某个类针对不同线程分别创建一个独立的实例对象。

import java.util.Random;/** * ThreadLocal:一个线程有一份数据相互独立 * ThreadLocal其实就是一个map */public class ThreadLocalTest1 {private static ThreadLocal<Integer> x = new ThreadLocal<Integer>();public static void main(String[] args) {for(int i=0;i<2;i++){new Thread(new Runnable(){public void run() {//保证数据不覆盖int data = new Random().nextInt();System.out.println(Thread.currentThread().getName() + " has put data :" + data);x.set(data);new A().get();new B().get();}}).start();}}static class A{public void get(){int data = x.get();System.out.println("A from " + Thread.currentThread().getName() + " get data :" + data);}}static class B{public void get(){int data = x.get();System.out.println("B from " + Thread.currentThread().getName() + " get data :" + data);}}}

输出结果：

Thread-0 has put data :2080519566Thread-1 has put data :926701240A from Thread-1 get data :926701240A from Thread-0 get data :2080519566B from Thread-1 get data :926701240B from Thread-0 get data :2080519566

总结：一个ThreadLocal代表一个变量，故其里面只能放一份数据，你有两个变量都要在线程范围内的共享，则要定义两个ThreadLocal对象。如果有100个变量要实现线程共享呢？那么此时就要先定义一个实体对象来封装这些变量，然后在ThreadLocal中存储这一个对象。如下示例

import java.util.Random;public class ThreadLocalTest2 {private static ThreadLocal<Integer> x = new ThreadLocal<Integer>();private static ThreadLocal<MyThreadScopeData> myThreadScopeData = new ThreadLocal<MyThreadScopeData>();public static void main(String[] args) {for(int i=0;i<2;i++){new Thread(new Runnable(){public void run() {//保证数据不覆盖int data = new Random().nextInt();System.out.println(Thread.currentThread().getName() + " has put data :" + data);x.set(data);MyThreadScopeData data2 = new MyThreadScopeData();data2.setName("name"+data);data2.setAge(data);myThreadScopeData.set(data2);new A().get();new B().get();}}).start();}}static class A{public void get(){int data = x.get();System.out.println("A from " + Thread.currentThread().getName() + " get data :" + data);MyThreadScopeData myData = myThreadScopeData.get();;System.out.println("A from " + Thread.currentThread().getName() + " getMyData: " + myData.getName() + "," +myData.getAge());}}static class B{public void get(){int data = x.get();System.out.println("B from " + Thread.currentThread().getName() + " get data :" + data);MyThreadScopeData myData = myThreadScopeData.get();;System.out.println("B from " + Thread.currentThread().getName() + " getMyData: " + myData.getName() + "," +myData.getAge());}}}class MyThreadScopeData{private String name;private int age;public String getName() {return name;}public void setName(String name) {this.name = name;}public int getAge() {return age;}public void setAge(int age) {this.age = age;}}

输出结果：

Thread-0 has put data :982174899A from Thread-0 get data :982174899A from Thread-0 getMyData: name982174899,982174899B from Thread-0 get data :982174899B from Thread-0 getMyData: name982174899,982174899Thread-1 has put data :811081634A from Thread-1 get data :811081634A from Thread-1 getMyData: name811081634,811081634B from Thread-1 get data :811081634B from Thread-1 getMyData: name811081634,811081634

但是我们看到上面的程序结构很糟糕，该怎么优化呢？

import java.util.Random;public class ThreadLocalTest {private static ThreadLocal<Integer> x = new ThreadLocal<Integer>();public static void main(String[] args) {for(int i=0;i<2;i++){new Thread(new Runnable(){public void run() {int data = new Random().nextInt();System.out.println(Thread.currentThread().getName() + " has put data :" + data);x.set(data);/*MyThreadScopeData myData = new MyThreadScopeData();myData.setName("name" + data);myData.setAge(data);myThreadScopeData.set(myData);*//** * MyThreadScopeData.getThreadInstance()拿到就是本线程相关的MyThreadScopeData的实例对象 * 存在就获取本对象，不存在就创建，反正就是获取本线程相关的 */MyThreadScopeData.getThreadInstance().setName("name" + data);MyThreadScopeData.getThreadInstance().setAge(data);new A().get();new B().get();}}).start();}}static class A{public void get(){int data = x.get();System.out.println("A from " + Thread.currentThread().getName() + " get data :" + data);/*MyThreadScopeData myData = myThreadScopeData.get();;System.out.println("A from " + Thread.currentThread().getName() + " getMyData: " + myData.getName() + "," +myData.getAge());*///MyThreadScopeData.getThreadInstance()得到与本线程相关的实例MyThreadScopeData myData = MyThreadScopeData.getThreadInstance();System.out.println("A from " + Thread.currentThread().getName() + " getMyData: " + myData.getName() + "," +myData.getAge());}}static class B{public void get(){int data = x.get();System.out.println("B from " + Thread.currentThread().getName() + " get data :" + data);MyThreadScopeData myData = MyThreadScopeData.getThreadInstance();System.out.println("B from " + Thread.currentThread().getName() + " getMyData: " + myData.getName() + "," +myData.getAge());}}}class MyThreadScopeData{private MyThreadScopeData(){}/** * 饿汉式单例模式 * 加synchronized的原因是，当一个线程创建了MyThreadScopeData()对象还未赋值给 * instance时，CPU切换另一个线程可能进来又创建了，导致线程的不安全性。 *  * 不加synchronized的原因是：instance是独立的(各拿各的)，不存在两个线程去拿一份数据的情形 */public static /*synchronized*/ MyThreadScopeData getThreadInstance(){MyThreadScopeData instance = map.get();if(instance == null){instance = new MyThreadScopeData();map.set(instance);}return instance;}//private static MyThreadScopeData instance = null;//new MyThreadScopeData();饱汉式单例模式private static ThreadLocal<MyThreadScopeData> map = new ThreadLocal<MyThreadScopeData>();private String name;private int age;public String getName() {return name;}public void setName(String name) {this.name = name;}public int getAge() {return age;}public void setAge(int age) {this.age = age;}}

输出结果：

Thread-0 has put data :1439565896Thread-1 has put data :369006026A from Thread-0 get data :1439565896A from Thread-1 get data :369006026A from Thread-0 getMyData: name1439565896,1439565896A from Thread-1 getMyData: name369006026,369006026B from Thread-1 get data :369006026B from Thread-0 get data :1439565896B from Thread-1 getMyData: name369006026,369006026B from Thread-0 getMyData: name1439565896,1439565896