连接池

Administrator

2023-04-14 c5bdf339674f5762125b25104d7269b1b50e4d91

 连接池

 JavaSource/com/mandi/fendan/util/ThreadPoolUtil.java

¶Ô±ÈÐÂÎļþ
@@ -0,0 +1,121 @@
package com.mandi.fendan.util;

import javax.validation.constraints.NotNull;
import java.util.concurrent.LinkedBlockingQueue;
import java.util.concurrent.ThreadFactory;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicInteger;

/**
 * çº¿ç¨‹æ± çš„工具类
 * ç”¨äºŽè¿›è¡Œçº¿ç¨‹çš„管理，防止重复创建、杀死线程。
 *

 * å¤šçº¿ç¨‹è¿è¡ŒæœŸé—´ï¼Œå¦‚果系统不断的创建、杀死新线程，
 * ä¼šäº§ç”Ÿè¿‡åº¦æ¶ˆè€—系统资源，以及过度切换线程的问题，甚至可能导致系统资源的崩溃。
 * å› æ­¤éœ€è¦çº¿ç¨‹æ± ï¼Œå¯¹çº¿ç¨‹è¿›è¡Œç®¡ç†ã€‚
 */
public class ThreadPoolUtil {

    private String TAG = getClass().getName();
    private static volatile ThreadPoolUtil mInstance;
    //核心线程池的数量，同时能够执行的线程数量
    private int corePoolSize;
    //最大线程池数量，表示当缓冲队列满的时候能继续容纳的等待任务的数量
    private int maxPoolSize;
    //存活时间
    private long keepAliveTime = 1;
    private TimeUnit unit = TimeUnit.HOURS;
    private ThreadPoolExecutor executor;

    private ThreadPoolUtil() {
        //给corePoolSize赋值：当前设备可用处理器核心数*2 + 1,能够让cpu的效率得到最大程度执行（有研究论证的）
        corePoolSize = Runtime.getRuntime().availableProcessors() * 2 + 1;
        maxPoolSize = corePoolSize;
        executor = new ThreadPoolExecutor(
                //当某个核心任务执行完毕，会依次从缓冲队列中取出等待任务
                corePoolSize,
                // ç„¶åŽnew LinkedBlockingQueue(),然后maximumPoolSize,但是它的数量是包含了corePoolSize的
                maxPoolSize,
                //表示的是maximumPoolSize当中等待任务的存活时间
                keepAliveTime,
                unit,
                //缓冲队列，用于存放等待任务，Linked的先进先出
                new LinkedBlockingQueue(),
                new DefaultThreadFactory(Thread.NORM_PRIORITY, "thread-pool-"),
                new ThreadPoolExecutor.AbortPolicy()
        );
    }

    public static ThreadPoolUtil getInstance() {
        if (mInstance == null) {
            synchronized (ThreadPoolUtil.class) {
                if (mInstance == null) {
                    mInstance = new ThreadPoolUtil();
                }
            }
        }
        return mInstance;
    }

    /**
     * æ‰§è¡Œä»»åŠ¡
     *
     * @param runnable
     */
    public void execute(Runnable runnable) {
        if (executor == null) {
            executor = new ThreadPoolExecutor(
                    corePoolSize,
                    maxPoolSize,
                    keepAliveTime,
                    TimeUnit.SECONDS,
                    new LinkedBlockingQueue(),
                    new DefaultThreadFactory(Thread.NORM_PRIORITY, "thread-pool-"),
                    new ThreadPoolExecutor.AbortPolicy());
        }
        if (runnable != null) {
            executor.execute(runnable);
        }
    }

    /**
     * ç§»é™¤ä»»åŠ¡
     *
     * @param runnable
     */
    public void remove(Runnable runnable) {
        if (runnable != null) {
            executor.remove(runnable);
        }
    }

    private static class DefaultThreadFactory implements ThreadFactory {
        //线程池的计数
        private static final AtomicInteger poolNumber = new AtomicInteger(1);
        //线程的计数
        private final AtomicInteger threadNumber = new AtomicInteger(1);
        private final ThreadGroup group;
        private final String namePrefix;
        private final int threadPriority;

        DefaultThreadFactory(int threadPriority, String threadNamePrefix) {
            this.threadPriority = threadPriority;
            this.group = Thread.currentThread().getThreadGroup();
            this.namePrefix = threadNamePrefix + poolNumber.getAndIncrement() + "-thread-";
        }

        @Override
        public Thread newThread(@NotNull Runnable r) {
            Thread thread = new Thread(group, r, namePrefix + threadNumber.getAndIncrement(), 0);
            // è¿”回True该线程就是守护线程
            // å®ˆæŠ¤çº¿ç¨‹åº”该永远不去访问固有资源，如：数据库、文件等。因为它会在任何时候甚至在一个操作的中间发生中断。
            if (thread.isDaemon()) {
                thread.setDaemon(false);
            }
            thread.setPriority(threadPriority);
            return thread;
        }
    }
}