Callable+ThreadPoolExecutor实现多线程并发并获得返回值(转)

出处:https://blog.csdn.net/kity9420/article/details/80740466

 

前言
  经常会遇到一些性能问题,比如调用某个接口,可能要循环调用100次,并且需要拿到每一次调用的返回结果,通常我们都是放在for循环中一次次的串行调用,这种方式可想而知道有多慢,那怎么解决这个问题呢?

 

多线程
  为了解决以上问题,我使用的方式是多线程。多线程常规的有两种实现方式,即继承Tread类,实现Runnable接口,但是这两种实现方式,有一个共同的问题,就是没有返回值,对于我们来说,获得每个线程的返回值,是个很困难的问题,因此不能用Tread类或Runnable接口,我用的是Callable和ThreadPoolExecutor,Callable的process方法可以允许有返回值,ThreadPoolExecutor的invokeAll或submit方法可以拿到线程的执行结果

 

案例
  假设需要给100个用户发送邮件,并需要每个用户的返回结果,先看下代码结构

               

  CallableTemplate.java 

package com.gdut.thread.multiThread;

import java.util.concurrent.Callable;

/**
 * 多线程模板类
 * @author yang.han
 *
 * @param <V>
 */
public abstract class CallableTemplate<V> implements Callable<V>{
    
    /**
     * 前置处理,子类可以Override该方法
     */
    public void beforeProcess() {
        System.out.println("before process");
    }
    
    /**
     * 处理业务逻辑的方法,需要子类去Override
     * @param <V>
     * @return
     */
    public abstract V process();
    
    /**
     * 后置处理,子类可以Override该方法
     */
    public void afterProcess() {
        System.out.println("after process");
    }

    @Override
    public V call() throws Exception {
        beforeProcess();
        V result = process();
        afterProcess();
        return result;
    }

}

  CallableTemplate类实现了Callable接口,并实现了process方法,该类是一个抽象类,接收任意返回值的类型,beforeProcess方法为前置处理,afterProcess的后置处理,process为具体的业务逻辑抽象方法,该方法在子类中实现

 

IConcurrentThreadPool.java

package com.gdut.thread.multiThread;

import java.util.List;
import java.util.concurrent.ExecutionException;

public interface IConcurrentThreadPool {

    /**
     * 初始化线程池
     */
    void initConcurrentThreadPool();
    
    /**
     * 提交单个任务
     * @param <V>
     * @param task
     * @return
     * @throws InterruptedException
     * @throws ExecutionException
     */
    <V> V submit(CallableTemplate<V> task) throws InterruptedException, ExecutionException;
    
    /**
     * 提交多个任务
     * @param <V>
     * @param tasks
     * @return
     * @throws InterruptedException 
     * @throws ExecutionException 
     */
    <V> List<V> invokeAll(List<? extends CallableTemplate<V>> tasks) throws InterruptedException, ExecutionException;
}

  IConcurrentThreadPool是多线程接口类,声名了三个方法,initConcurrentThreadPool:初始化线程池,submit:提交单个任务的线程,并有返回值,invokeAll:提交多个任务的线程,并有返回值

 

ConcurrentThreadPool.java

package com.gdut.thread.multiThread;

import java.util.ArrayList;
import java.util.List;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.Future;
import java.util.concurrent.LinkedBlockingDeque;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;

public class ConcurrentThreadPool implements IConcurrentThreadPool{
    
    private ThreadPoolExecutor threadPoolExecutor;
    // 核心线程数
    private int corePoolSize = 10;
    // 最大线程数
    private int maximumPoolSize = 20;
    // 超时时间30秒
    private long keepAliveTime = 30;

    @Override
    public void initConcurrentThreadPool() {
        threadPoolExecutor = new ThreadPoolExecutor(corePoolSize,
                                                    maximumPoolSize, 
                                                    keepAliveTime, 
                                                    TimeUnit.SECONDS,
                                                    new LinkedBlockingDeque<Runnable>()
                                                    );
    }

    @Override
    public <V> V submit(CallableTemplate<V> task) throws InterruptedException, ExecutionException {
        Future<V> result = threadPoolExecutor.submit(task);
        return result.get();
    }

    @Override
    public <V> List<V> invokeAll(List<? extends CallableTemplate<V>> tasks) throws InterruptedException, ExecutionException {
        List<Future<V>> tasksResult = threadPoolExecutor.invokeAll(tasks);
        List<V> resultList = new ArrayList<V>();
        
        for(Future<V> future : tasksResult) {
            resultList.add(future.get());
        }
        return resultList;
    }

}

  ConcurrentThreadPool是创建线程池的实现类,用到了ThreadPoolExecutor线程池类及这个类的invokeAll方法和submit方法,这两个方法的返回值,都可以通过Future类的get方法获得

 

ICallableTaskFrameWork.java

package com.gdut.thread.multiThread;

import java.util.List;
import java.util.concurrent.ExecutionException;

public interface ICallableTaskFrameWork {
    <V> List<V> submitsAll(List<? extends CallableTemplate<V>> tasks)
            throws InterruptedException, ExecutionException;
}

  ICallableTaskFrameWork是定义的线程任务框架接口,所有的多线程调用,都通过该接口发起

 

CallableTaskFrameWork.java

package com.gdut.thread.multiThread;

import java.util.List;
import java.util.concurrent.ExecutionException;

public class CallableTaskFrameWork implements ICallableTaskFrameWork{

    private IConcurrentThreadPool concurrentThreadPool = new ConcurrentThreadPool();
    
    @Override
    public <V> List<V> submitsAll(List<? extends CallableTemplate<V>> tasks)
            throws InterruptedException, ExecutionException {
        concurrentThreadPool.initConcurrentThreadPool();
        return concurrentThreadPool.invokeAll(tasks);
    }

}

  CallableTaskFrameWork是ICallableTaskFrameWork 的实现类,在submitsAll实现方法中,通过调用线程池对象IConcurrentThreadPool接口的invokeAll方法来发起多线程的调用,这里注意一个,在submitAll实现方法中,我手动的调用了初始化线程池的方法concurrentThreadPool.initConcurrentThreadPool(),在真实的项目上,应该在应用启动的时候就调用该方法来初始化线程池

测试类代码 
SendMessageService.java,假设这是一个发送邮件信息的服务类

package com.gdut.thread.multiThread;

public class SendMessageService {
    public void sendMessage(String email,String content){
        System.out.println("发送邮件。。。");
    }
}

 

SendMessageHander.java,多线程发送邮件的处理类

package com.gdut.thread.multiThread;

import java.util.HashMap;
import java.util.Map;

public class SendMessageHander extends CallableTemplate<Map<String, String>>{
    
    private String email;
    private String content;
    public SendMessageHander(String email,String content) {
        this.email = email;
        this.content = content;
    }

    @Override
    public Map<String, String> process() {
        SendMessageService sendMessageService = new SendMessageService();
        sendMessageService.sendMessage(email, content);
        Map<String, String> map = new HashMap<String, String>();
        map.put(email, content);
        return map;
    }

}

  这个类继承了上面的CallableTemplate,我们要的返回值是Map,因此泛型类型是Map,在类中还重写了process方法,在方法中调用发送邮件的业务逻辑接口SendMessageService.sendMessage,并将返回结果组装成Map返回,这里我就简单处理了,将邮件地址及内容放在Map中直接返回了;另外还要注意这个类有个有参构造器,通过构建器可以接收需要传递进来的参数

 

SendMessageTest.java,测试类

package com.gdut.thread.multiThread;

import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.concurrent.ExecutionException;

public class SendMessageTest {

    public static void main(String[] args) throws InterruptedException, ExecutionException {
        ICallableTaskFrameWork callableTaskFrameWork = new CallableTaskFrameWork();
        
        List<CallableTemplate<Map<String, String>>> tasks = new ArrayList<CallableTemplate<Map<String, String>>>();
        
        SendMessageHander sendMessageHander = null;
        
        // 将需要发送邮件的邮件地址及内容组装好,放在一个集合中
        for (int i = 0; i < 1000; i++) {
            sendMessageHander = new SendMessageHander("email" + i, "content" + i);
            tasks.add(sendMessageHander);
        }
        
        //通过多线程一次性发起邮件,并拿到返回结果集
        List<Map<String, String>> results = callableTaskFrameWork.submitsAll(tasks);
        
        // 解析返回结果集
        for (Map<String, String> map : results) {
            for (Entry<String, String> entry : map.entrySet()) {
                System.out.println(entry.getKey() + "\t" + entry.getValue());
            }
        }
    }

}

运行结果 

             

 

 

附录:还可以看这边文章: java并发异步编程 原来十个接口的活现在只需要一个接口就搞定!

 

posted @ 2019-09-12 21:59  myseries  阅读(2015)  评论(0编辑  收藏  举报