Java8環境下使用restTemplate單/多線程下載大文件和小文件

0. 準備工作

下面使用的restTemplate, 都是使用整合了HttpClient連接池的restTemplate, 關於整合可以看我的上一篇文章 ,當然直接使用原生的也是可以的

我這裏還使用了VisualVm Launcher的idea插件,來查看運行時的內存夠和線程

1. 簡單的下載文件

這裏使用的是restTemplate調用getForEntity, 獲取到字節數組, 再將字節數組通過java8的Files工具類的write方法, 直接寫到目標文件.

這裏需要注意的點是:

1. 會將文件的字節數組全部放入內存中, 及其消耗資源
2. 注意目標文件夾不存在,需要手動創建文件夾的問題, 注意生成目標路徑的時候,斜槓\的處理問題

代碼如下:

import lombok.extern.slf4j.Slf4j;
import org.springframework.http.HttpMethod;
import org.springframework.http.MediaType;
import org.springframework.http.ResponseEntity;
import org.springframework.stereotype.Component;
import org.springframework.util.CollectionUtils;
import org.springframework.web.client.RequestCallback;
import org.springframework.web.client.ResponseExtractor;
import org.springframework.web.client.RestTemplate;
import org.springframework.web.util.UriComponentsBuilder;

import javax.annotation.Resource;
import java.io.IOException;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.time.Instant;
import java.time.temporal.ChronoUnit;
import java.util.Arrays;
import java.util.Map;
import java.util.Objects;

/**
 * @Author: zgd
 * @Date: 2019/3/29 10:49
 * @Description:
 */
@Component
@Slf4j
public class WebFileUtils {

  /**
   * 使用自定義的httpclient的restTemplate
   */
  @Resource(name = "httpClientTemplate")
  private RestTemplate httpClientTemplate;


  /**
   * 下載小文件,採用字節數組的方式,直接將所有返回都放入內存中,容易引發內存溢出
   *
   * @param url
   * @param targetDir
   */
  public void downloadLittleFileToPath(String url, String targetDir) {
    downloadLittleFileToPath(url, targetDir, null);
  }

  /**
   * 下載小文件,直接將所有返回都放入內存中,容易引發內存溢出
   *
   * @param url
   * @param targetDir
   */
  public void downloadLittleFileToPath(String url, String targetDir, Map<String, String> params) {
    Instant now = Instant.now();
    String completeUrl = addGetQueryParam(url, params);
    ResponseEntity<byte[]> rsp = httpClientTemplate.getForEntity(completeUrl, byte[].class);
    log.info("[下載文件] [狀態碼] code:{}", rsp.getStatusCode());
    try {
      String path = getAndCreateDownloadDir(url, targetDir);
      Files.write(Paths.get(path), Objects.requireNonNull(rsp.getBody(), "未獲取到下載文件"));
    } catch (IOException e) {
      log.error("[下載文件] 寫入失敗:", e);
    }
    log.info("[下載文件] 完成,耗時:{}", ChronoUnit.MILLIS.between(now, Instant.now()));
  }
    
    
    
  /**
   * 拼接get請求參數
   *
   * @param url
   * @param params
   * @return
   */
  private String addGetQueryParam(String url, Map<String, String> params) {
    UriComponentsBuilder uriComponentsBuilder = UriComponentsBuilder.fromHttpUrl(url);
    if (!CollectionUtils.isEmpty(params)) {
      for (Map.Entry<String, ?> varEntry : params.entrySet()) {
        uriComponentsBuilder.queryParam(varEntry.getKey(), varEntry.getValue());
      }
    }
    return uriComponentsBuilder.build().encode().toString();
  }


  /**
   * 創建或獲取下載文件夾的路徑
   *
   * @param url
   * @param targetDir
   * @return
   */
  public String getAndCreateDownloadDir(String url, String targetDir) throws IOException {
    String filename = url.substring(url.lastIndexOf("/") + 1);
    int i = 0;
    if ((i = url.indexOf("?")) != -1) {
      filename = filename.substring(0, i);
    }
    if (!Files.exists(Paths.get(targetDir))) {
      Files.createDirectories(Paths.get(targetDir));
    }
    return targetDir.endsWith("/") ? targetDir + filename : targetDir + "/" + filename;
  }


}

這裏找到一個搜狗瀏覽器的下載地址, 運行代碼,並啓動.

 
package com.zgd.springboot.demo.http.test;

import com.alibaba.fastjson.JSON;
import com.google.common.collect.Maps;
import com.google.common.util.concurrent.ThreadFactoryBuilder;
import com.zgd.springboot.demo.http.HttpApplication;
import com.zgd.springboot.demo.http.IO.utils.DownloadTool;
import com.zgd.springboot.demo.http.IO.utils.QiniuUtil;
import com.zgd.springboot.demo.http.IO.utils.WebFileUtils;
import com.zgd.springboot.demo.http.service.IHttpService;
import lombok.extern.slf4j.Slf4j;
import org.junit.Test;
import org.junit.runner.RunWith;
import org.springframework.boot.test.context.SpringBootTest;
import org.springframework.stereotype.Component;
import org.springframework.test.context.TestPropertySource;
import org.springframework.test.context.junit4.SpringRunner;

import javax.annotation.Resource;
import java.time.Instant;
import java.time.temporal.ChronoUnit;
import java.util.HashMap;
import java.util.concurrent.*;

/**
 * @Author: zgd
 * @Date: 2019/3/25 15:56
 * @Description:
 */
@Component
@RunWith(SpringRunner.class)
@SpringBootTest(classes = HttpApplication.class)
@TestPropertySource("classpath:application.yml")
@Slf4j
public class SpringTest {

  @Resource
  private WebFileUtils webFileUtils;

  @Test
  public void testDownloadQiniu(){
    String path = "D:/down/file/";
    String url = "http://cdn4.mydown.com/5c9df131/6dcdc2f2ff1aba454f90d8581eab1820/newsoft/sogou_explorer_fast_8.5.7.29587_7685.exe";
    webFileUtils.downloadLittleFileToPath(url,path);
  }
    
}

可以看到使用內存從一開始的100多M,後面飆升到300多M. 總耗時是8533ms.

爲了更好的展示這個方法對內存的佔用,下載一個500M左右的Idea看看

String url = "https://download.jetbrains.8686c.com/idea/ideaIU-2019.1.exe";

可以看到佔用內存一度達到900多M, 這才下載500M的軟件,如果我們需要服務器下載幾G的文件,內存肯定是不夠用的.

至於下載時間,速度是300k/s左右,實在沒耐心等500M的下載了

2. 單線程大文件下載

既然上面的方法只能下載小文件,那麼大文件怎麼辦呢? 我們使用流的方式來解決. 在上面的類里加l兩個方法. 這次使用Files的copy方法來處理流.

 /**
   * 下載大文件,使用流接收
   *
   * @param url
   * @param targetDir
   */
  public void downloadBigFileToPath(String url, String targetDir){
    downloadBigFileToPath(url,targetDir,null);
  }
  /**
   * 下載大文件,使用流接收
   *
   * @param url
   * @param targetDir
   */
  public void downloadBigFileToPath(String url, String targetDir, Map<String, String> params) {
    Instant now = Instant.now();
    String completeUrl = addGetQueryParam(url, params);
    try {
      String path = getAndCreateDownloadDir(url, targetDir);
      //定義請求頭的接收類型
      RequestCallback requestCallback = request -> request.getHeaders()
              .setAccept(Arrays.asList(MediaType.APPLICATION_OCTET_STREAM, MediaType.ALL));
      // getForObject會將所有返回直接放到內存中,使用流來替代這個操作
      ResponseExtractor<Void> responseExtractor = response -> {
        // Here I write the response to a file but do what you like
        Files.copy(response.getBody(), Paths.get(path));
        return null;
      };
      httpClientTemplate.execute(completeUrl, HttpMethod.GET, requestCallback, responseExtractor);
    } catch (IOException e) {
      log.error("[下載文件] 寫入失敗:", e);
    }
    log.info("[下載文件] 完成,耗時:{}", ChronoUnit.MILLIS.between(now, Instant.now()));
  }

先試試那個50M的搜狗瀏覽器:

看到內存基本保持在100M左右,總耗時:5514ms


再試試那個500M的Idea:內存基本穩定在150M以內,下載速度也是300kb/s左右

可以看得出, 使用流的方式還是可以很好的保證內存資源不會崩掉的

3. 多線程下載

上面雖然把大文件的問題解決了 ,但是下載速度300k/s實在是太慢了.雖然小文件還是可以達到5s左右下載完50M,但是大文件還是需要更快的下載速度(下載速度也取決於當前的運營商網速和資源)

主要就是先調用一次HEAD方法去獲取到文件大小, 我這裏默認開啓了10個線程,然後每個線程分配好下載的數據量,在請求頭中設置Range屬性,分別去下載屬於它那一部分的數據,然後最後合併成一個文件

直接上代碼吧:

package com.zgd.springboot.demo.http.IO.utils;

import com.google.common.collect.Lists;
import com.google.common.util.concurrent.ThreadFactoryBuilder;
import lombok.extern.slf4j.Slf4j;
import org.springframework.http.*;
import org.springframework.stereotype.Component;
import org.springframework.util.Assert;
import org.springframework.web.client.RequestCallback;
import org.springframework.web.client.ResponseExtractor;
import org.springframework.web.client.RestTemplate;

import javax.annotation.Resource;
import java.io.File;
import java.io.IOException;
import java.io.RandomAccessFile;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.Objects;
import java.util.concurrent.*;

/**
 * Created by yangzheng03 on 2018/1/16.
 * https://www.dubby.cn/
 */
@Component
@Slf4j
public class DownloadTool {

  /**
   * 使用自定義的httpclient的restTemplate
   */
  @Resource(name = "httpClientTemplate")
  private RestTemplate httpClientTemplate;

  @Resource
  private WebFileUtils webFileUtils;

  /**
   * 線程最小值
   */
  private static final int MIN_POOL_SIZE = 10;
  /**
   * 線程最大值
   */
  private static final int MAX_POOL_SIZE = 100;
  /**
   * 等待隊列大小
   */
  private static final int WAIT_QUEUE_SIZE = 1000;
  /**
   * 線程池
   */
  private static ExecutorService threadPool;

  private static final int ONE_KB_SIZE = 1024;
  /**
   * 大於20M的文件視爲大文件,採用流下載
   */
  private static final int BIG_FILE_SIZE = 20 * 1024 * 1024;
  private static String prefix = String.valueOf(System.currentTimeMillis());


  public void downloadByMultithread(String url, String targetPath, Integer threadNum) {

    long startTimestamp = System.currentTimeMillis();
    //開啓線程
    threadNum = threadNum == null ? MIN_POOL_SIZE : threadNum;
    Assert.isTrue(threadNum > 0, "線程數不能爲負數");
    ThreadFactory factory = new ThreadFactoryBuilder().setNameFormat("http-demo-%d").build();
    threadPool = new ThreadPoolExecutor(
            threadNum, MAX_POOL_SIZE, 0, TimeUnit.MINUTES,
            new LinkedBlockingDeque<>(WAIT_QUEUE_SIZE), factory);
    boolean isBigFile;

    //調用head方法,只獲取頭信息,拿到文件大小
    long contentLength = httpClientTemplate.headForHeaders(url).getContentLength();
    isBigFile = contentLength >= BIG_FILE_SIZE;

    if (contentLength > 1024 * ONE_KB_SIZE) {
      log.info("[多線程下載] Content-Length\t" + (contentLength / 1024 / 1024) + "MB");
    } else if (contentLength > ONE_KB_SIZE) {
      log.info("[多線程下載] Content-Length\t" + (contentLength / 1024) + "KB");
    } else {
      log.info("[多線程下載] Content-Length\t" + (contentLength) + "B");
    }
    long tempLength = contentLength / threadNum;
    long start, end = -1;

    ArrayList<CompletableFuture<DownloadTemp>> futures = Lists.newArrayListWithCapacity(threadNum);
    String fileFullPath;
    RandomAccessFile resultFile;
    try {
      fileFullPath = webFileUtils.getAndCreateDownloadDir(url, targetPath);
      //創建目標文件
      resultFile = new RandomAccessFile(fileFullPath, "rw");

      log.info("[多線程下載] Download started, url:{}\tfileFullPath:{}", url, fileFullPath);
      for (int i = 0; i < threadNum; ++i) {
        start = end + 1;
        end = end + tempLength;
        if (i == threadNum - 1) {
          end = contentLength;
        }
        log.info("[多線程下載] start:{}\tend:{}",start, end);

        DownloadThread thread = new DownloadThread(httpClientTemplate, i, start, end, url, fileFullPath, isBigFile);
        CompletableFuture<DownloadTemp> future = CompletableFuture.supplyAsync(thread::call, threadPool);
        futures.add(future);
      }

    } catch (Exception e) {
      log.error("[多線程下載] 下載出錯", e);
      return;
    }finally {
      threadPool.shutdown();
    }

    //合併文件
    futures.forEach(f -> {
      try {
        f.thenAccept(o -> {
          try {
            log.info("[多線程下載] {} 開始合併,文件:{}", o.threadName, o.filename);
            RandomAccessFile tempFile = new RandomAccessFile(o.filename, "rw");
            tempFile.getChannel().transferTo(0, tempFile.length(), resultFile.getChannel());
            tempFile.close();
            File file = new File(o.filename);
            boolean b = file.delete();
            log.info("[多線程下載] {} 刪除臨時文件:{}\t結果:{}", o.threadName, o.filename, b);
          } catch (IOException e) {
            e.printStackTrace();
            log.error("[多線程下載] {} 合併出錯", o.threadName, e);
          }
        }).get();
      } catch (Exception e) {
        log.error("[多線程下載] 合併出錯", e);
      }finally {
        threadPool.shutdown();
      }
    });


    long completedTimestamp = System.currentTimeMillis();
    log.info("=======下載完成======,耗時{}",
            isBigFile ? (completedTimestamp - startTimestamp) / 1000 + "s" : (completedTimestamp - startTimestamp) + "ms");
  }

  public class DownloadThread implements Callable<DownloadTemp> {

    private int index;
    private String filePath;
    private long start, end;
    private String urlString;
    private RestTemplate httpClientTemplate;
    private boolean isBigFile;

    DownloadThread(RestTemplate restTemplate, int index, long start, long end, String url, String fileFullPath, boolean isBigFile) {
      this.httpClientTemplate = restTemplate;
      this.urlString = url;
      this.index = index;
      this.start = start;
      this.end = end;
      this.isBigFile = isBigFile;
      Assert.hasLength(fileFullPath, "文件下載路徑不能爲空");
      this.filePath = String.format("%s-%s-%d", fileFullPath, prefix, index);
    }

    @Override
    public DownloadTemp call() {
      //定義請求頭的接收類型
      try {
        if (isBigFile) {
          downloadBigFIle();
        } else {
          downloadLittleFIle();
        }
      } catch (Exception e) {
        log.error("[線程下載] 下載失敗:", e);
      }
      DownloadTemp downloadTemp = new DownloadTemp();
      downloadTemp.index = index;
      downloadTemp.filename = filePath;
      downloadTemp.threadName = Thread.currentThread().getName();
      log.info("[線程下載] \tcompleted.");
      return downloadTemp;
    }

    /**
     * 下載小文件
     *
     * @throws IOException
     */
    private void downloadLittleFIle() throws IOException {
      HttpHeaders headers = new HttpHeaders();
      headers.set(HttpHeaders.RANGE, "bytes=" + start + "-" + end);
      headers.setAccept(Collections.singletonList(MediaType.ALL));
      ResponseEntity<byte[]> rsp = httpClientTemplate.exchange(urlString, HttpMethod.GET, new HttpEntity<>(headers), byte[].class);
      log.info("[線程下載] 返回狀態碼:{}", rsp.getStatusCode());
      Files.write(Paths.get(filePath), Objects.requireNonNull(rsp.getBody(), "未獲取到下載文件"));
    }

    /**
     * 下載大文件
     *
     * @throws IOException
     */
    private void downloadBigFIle() {
      RequestCallback requestCallback = request -> {
        HttpHeaders headers = request.getHeaders();
        headers.set(HttpHeaders.RANGE, "bytes=" + start + "-" + end);
        headers.setAccept(Arrays.asList(MediaType.APPLICATION_OCTET_STREAM, MediaType.ALL));
      };
      // getForObject會將所有返回直接放到內存中,使用流來替代這個操作
      ResponseExtractor<Void> responseExtractor = response -> {
        // Here I write the response to a file but do what you like
        Files.copy(response.getBody(), Paths.get(filePath));
        log.info("[線程下載] 返回狀態碼:{}", response.getStatusCode());
        return null;
      };
      httpClientTemplate.execute(urlString, HttpMethod.GET, requestCallback, responseExtractor);
    }
  }

  private static class DownloadTemp {
    private int index;
    private String filename;
    private String threadName;
  }
}

下載那個50M的搜狗瀏覽器, 耗時5s, 因爲這個類裏面我判斷了20M以上爲大文件,採用上面的流的方式下載,所以這裏內存基本100到200M之間.



下載500M的Idea試試看,可以看到內存還是保持在200M以內,速度的話達到3M/s, 總共花了200多秒也就是4分鐘左右下完.