Java实现多线程下载，支持断点续传

完整代码：https://github.com/iyuanyb/Downloader

多线程下载及断点续传的实现是使用 HTTP/1.1 引入的 Range 请求参数，可以访问Web资源的指定区间的内容。虽然实现了多线程及断点续传，但还有很多不完善的地方。

包含四个类：

• Downloader: 主类，负责分配任务给各个子线程，及检测进度
• DownloadFile: 表示要下载的哪个文件，为了能写输入到文件的指定位置，使用 RandomAccessFile 类操作文件，多个线程写同一个文件需要保证线程安全，这里直接调用 getChannel 方法，获取一个文件通道，FileChannel是线程安全的。
• DownloadTask: 实际执行下载的线程，获取 [lowerBound, upperBound] 区间的数据，当下载过程中出现异常时要通知其他线程（使用 AtomicBoolean），结束下载
• Logger: 实时记录下载进度，以便续传时知道从哪开始。感觉这里做的比较差，为了能实时写出日志及方便地使用Properties类的load/store方法格式化输入输出，每次都是打开后再关闭。

演示：

随便找一个文件下载：

强行结束程序并重新运行：

 

 日志文件：

断点续传的关键是记录各个线程的下载进度，这里细节比较多，花了很久。只需要记录每个线程请求的Range区间极客，每次成功写数据到文件时，就更新一次下载区间。下面是下载完成后的日志内容。

代码:

Downloader.java

package downloader;

import java.io.*;
import java.net.*;
import java.nio.file.Files;
import java.nio.file.Path;
import java.util.concurrent.atomic.AtomicBoolean;

public class Downloader {
    private static final int DEFAULT_THREAD_COUNT = 4;  // 默认线程数量
    private AtomicBoolean canceled; // 取消状态，如果有一个子线程出现异常，则取消整个下载任务
    private DownloadFile file; // 下载的文件对象
    private String storageLocation;
    private final int threadCount; // 线程数量
    private long fileSize; // 文件大小
    private final String url;
    private long beginTime; // 开始时间
    private Logger logger;

    public Downloader(String url) {
        this(url, DEFAULT_THREAD_COUNT);
    }

    public Downloader(String url, int threadCount) {
        this.url = url;
        this.threadCount = threadCount;
        this.canceled = new AtomicBoolean(false);
        this.storageLocation = url.substring(url.lastIndexOf('/')+1);
        this.logger = new Logger(storageLocation + ".log", url, threadCount);
    }

    public void start() {
        boolean reStart = Files.exists(Path.of(storageLocation + ".log"));
        if (reStart) {
            logger = new Logger(storageLocation + ".log");
            System.out.printf("* 继续上次下载进度[已下载：%.2fMB]：%s
", logger.getWroteSize() / 1014.0 / 1024, url);
        } else {
            System.out.println("* 开始下载：" + url);
        }
        if (-1 == (this.fileSize = getFileSize()))
            return;
        System.out.printf("* 文件大小：%.2fMB
", fileSize / 1024.0 / 1024);

        this.beginTime = System.currentTimeMillis();
        try {
            this.file = new DownloadFile(storageLocation, fileSize, logger);
            if (reStart) {
                file.setWroteSize(logger.getWroteSize());
            }
            // 分配线程下载
            dispatcher(reStart);
            // 循环打印进度
            printDownloadProgress();
        } catch (IOException e) {
            System.err.println("x 创建文件失败[" + e.getMessage() + "]");
        }
    }

    /**
     * 分配器，决定每个线程下载哪个区间的数据
     */
    private void dispatcher(boolean reStart) {
        long blockSize = fileSize / threadCount; // 每个线程要下载的数据量
        long lowerBound = 0, upperBound = 0;
        long[][] bounds = null;
        int threadID = 0;
        if (reStart) {
           bounds = logger.getBounds();
        }
        for (int i = 0; i < threadCount; i++) {
            if (reStart) {
                threadID = (int)(bounds[i][0]);
                lowerBound = bounds[i][1];
                upperBound = bounds[i][2];
            } else {
                threadID = i;
                lowerBound = i * blockSize;
                // fileSize-1 !!!!! fu.ck，找了一下午的错
                upperBound = (i == threadCount - 1) ? fileSize-1 : lowerBound + blockSize;
            }
            new DownloadTask(url, lowerBound, upperBound, file, canceled, threadID).start();
        }
    }

    /**
     * 循环打印进度，直到下载完毕，或任务被取消
     */
    private void printDownloadProgress() {
        long downloadedSize = file.getWroteSize();
        int i = 0;
        long lastSize = 0; // 三秒前的下载量
        while (!canceled.get() && downloadedSize < fileSize) {
            if (i++ % 4 == 3) { // 每3秒打印一次
                System.out.printf("下载进度：%.2f%%, 已下载：%.2fMB，当前速度：%.2fMB/s
",
                        downloadedSize / (double)fileSize * 100 ,
                        downloadedSize / 1024.0 / 1024,
                        (downloadedSize - lastSize) / 1024.0 / 1024 / 3);
                lastSize = downloadedSize;
                i = 0;
            }
            try {
                Thread.sleep(1000);
            } catch (InterruptedException ignore) {}
            downloadedSize = file.getWroteSize();
        }
        file.close();
        if (canceled.get()) {
            try {
                Files.delete(Path.of(storageLocation));
            } catch (IOException ignore) {
            }
            System.err.println("x 下载失败，任务已取消");
        } else {
            System.out.println("* 下载成功，本次用时"+ (System.currentTimeMillis() - beginTime) / 1000 +"秒");
        }
    }

    /**
     * @return 要下载的文件的尺寸
     */
    private long getFileSize() {
        if (fileSize != 0) {
            return fileSize;
        }
        HttpURLConnection conn = null;
        try {
            conn = (HttpURLConnection)new URL(url).openConnection();
            conn.setConnectTimeout(3000);
            conn.setRequestMethod("HEAD");
            conn.connect();
            System.out.println("* 连接服务器成功");
        } catch (MalformedURLException e) {
            throw new RuntimeException("URL错误");
        } catch (IOException e) {
            System.err.println("x 连接服务器失败["+ e.getMessage() +"]");
            return -1;
        }
        return conn.getContentLengthLong();
    }

    public static void main(String[] args) throws IOException {
        new Downloader("http://js.xiazaicc.com//down2/ucliulanqi_downcc.zip").start();
    }
}

DownloadTask.java

package downloader;

import java.io.*;
import java.net.HttpURLConnection;
import java.net.URL;
import java.nio.ByteBuffer;
import java.nio.channels.Channels;
import java.nio.channels.ReadableByteChannel;
import java.util.concurrent.atomic.AtomicBoolean;

class DownloadTask extends Thread {
    private final String url;
    private long lowerBound; // 下载的文件区间
    private long upperBound;
    private AtomicBoolean canceled;
    private DownloadFile downloadFile;
    private int threadId;

    DownloadTask(String url, long lowerBound, long upperBound, DownloadFile downloadFile,
                        AtomicBoolean canceled, int threadID) {
        this.url = url;
        this.lowerBound = lowerBound;
        this.upperBound = upperBound;
        this.canceled = canceled;
        this.downloadFile = downloadFile;
        this.threadId = threadID;
    }

    @Override
    public void run() {
        ReadableByteChannel input = null;
        try {
            ByteBuffer buffer = ByteBuffer.allocate(1024 * 1024 * 2); // 2MB
            input = connect();
            System.out.println("* [线程" + threadId + "]连接成功，开始下载...");

            int len;
            while (!canceled.get() && lowerBound <= upperBound) {
                buffer.clear();
                len = input.read(buffer);
                downloadFile.write(lowerBound, buffer, threadId, upperBound);
                lowerBound += len;
            }
            if (!canceled.get()) {
                System.out.println("* [线程" + threadId + "]下载完成" + ": " + lowerBound + "-" + upperBound);
            }
        } catch (IOException e) {
            canceled.set(true);
            System.err.println("x [线程" + threadId + "]遇到错误[" + e.getMessage() + "]，结束下载");
        } finally {
            if (input != null) {
                try {
                    input.close();
                } catch (IOException e) {
                    e.printStackTrace();
                }
            }
        }
    }

    /**
     * 连接WEB服务器，并返回一个数据通道
     * @return 返回通道
     * @throws IOException 网络连接错误
     */
    private ReadableByteChannel connect() throws IOException {
        HttpURLConnection conn = (HttpURLConnection)new URL(url).openConnection();
        conn.setConnectTimeout(3000);
        conn.setRequestMethod("GET");
        conn.setRequestProperty("Range", "bytes=" + lowerBound + "-" + upperBound);
//        System.out.println("thread_"+ threadId +": " + lowerBound + "-" + upperBound);
        conn.connect();

        int statusCode = conn.getResponseCode();
        if (HttpURLConnection.HTTP_PARTIAL != statusCode) {
            conn.disconnect();
            throw new IOException("状态码错误：" + statusCode);
        }

        return Channels.newChannel(conn.getInputStream());
    }
}

DownloadFile.java

package downloader;

import java.io.IOException;
import java.io.RandomAccessFile;
import java.nio.ByteBuffer;
import java.nio.channels.FileChannel;
import java.util.concurrent.atomic.AtomicLong;

class DownloadFile {
    private final RandomAccessFile file;
    private final FileChannel channel; // 线程安全类
    private AtomicLong wroteSize; // 已写入的长度
    private Logger logger;

    DownloadFile(String fileName, long fileSize, Logger logger) throws IOException {
        this.wroteSize = new AtomicLong(0);
        this.logger = logger;
        this.file = new RandomAccessFile(fileName, "rw");
        file.setLength(fileSize);
        channel = file.getChannel();
    }

    /**
     * 写数据
     * @param offset 写偏移量
     * @param buffer 数据
     * @throws IOException 写数据出现异常
     */
    void write(long offset, ByteBuffer buffer, int threadID, long upperBound) throws IOException {
        buffer.flip();
        int length = buffer.limit();
        while (buffer.hasRemaining()) {
            channel.write(buffer, offset);
        }
        wroteSize.addAndGet(length);
        logger.updateLog(threadID, length, offset + length, upperBound); // 更新日志
    }

    /**
     * @return 已经下载的数据量，为了知道何时结束整个任务，以及统计信息
     */
    long getWroteSize() {
        return wroteSize.get();
    }

    // 继续下载时调用
    void setWroteSize(long wroteSize) {
        this.wroteSize.set(wroteSize);
    }

    void close() {
        try {
            file.close();
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
}

Logger.java

package downloader;

import java.io.*;
import java.util.Properties;

class Logger {
    private String logFileName; // 下载的文件的名字
    private Properties log;

     /**
      * 重新开始下载时，使用该构造函数
      * @param logFileName
      */
    Logger(String logFileName) {
        this.logFileName = logFileName;
        log = new Properties();
        FileInputStream fin = null;
        try {
            log.load(new FileInputStream(logFileName));
        } catch (IOException ignore) {
        } finally {
            try {
                fin.close();
            } catch (Exception ignore) {}
        }
    }

    Logger(String logFileName, String url, int threadCount) {
        this.logFileName = logFileName;
        this.log = new Properties();
        log.put("url", url);
        log.put("wroteSize", "0");
        log.put("threadCount", String.valueOf(threadCount));
        for (int i = 0; i < threadCount; i++) {
            log.put("thread_" + i, "0-0");
        }
    }


    synchronized void updateLog(int threadID, long length, long lowerBound, long upperBound) {
        log.put("thread_"+threadID, lowerBound + "-" + upperBound);
        log.put("wroteSize", String.valueOf(length + Long.parseLong(log.getProperty("wroteSize"))));

        FileOutputStream file = null;
        try {
            file = new FileOutputStream(logFileName); // 每次写时都清空文件
            log.store(file, null);
        } catch (IOException e) {
            e.printStackTrace();
        } finally {
            if (file != null) {
                try {
                    file.close();
                } catch (IOException e) {
                    e.printStackTrace();
                }
            }
        }
    }

    /**
     * 获取区间信息
     *      ret[i][0] = threadID, ret[i][1] = lowerBoundID, ret[i][2] = upperBoundID
     * @return
     */
    long[][] getBounds() {
        long[][] bounds = new long[Integer.parseInt(log.get("threadCount").toString())][3];
        int[] index = {0};
        log.forEach((k, v) -> {
            String key = k.toString();
            if (key.startsWith("thread_")) {
                String[] interval = v.toString().split("-");
                bounds[index[0]][0] = Long.parseLong(key.substring(key.indexOf("_") + 1));
                bounds[index[0]][1] = Long.parseLong(interval[0]);
                bounds[index[0]++][2] = Long.parseLong(interval[1]);
            }
        });
       return bounds;
    }
    long getWroteSize() {
        return Long.parseLong(log.getProperty("wroteSize"));
    }
}