OSS實現多文件多線程的斷點下載(java)
本文地址:http://blog.csdn.net/zrk1000/article/details/46286767
開放存儲服務(Open Storage Service,OSS),是阿里雲對外提供的海量、安全和高可靠的雲存儲服務,目前越來越多的開發者將應用數據存放至OSS,對使用OSS實現文件的斷點續傳功能使用的也比較多,在這兒分享下自己使用OSS的實例。
所謂斷點下載,就是要從文件已經下載的地方開始繼續下載,對於斷點續傳這樣有狀態功能的實現,關鍵點在於如何在客戶端完成狀態維護。此篇主要介紹多文件的多線程的斷點下載。
爲了限定同時下載文件的個數和每個文件同時下載的線程數,使用了線程池嵌套線程池來完成,外層線程池downloadMainPool用來限定併發下載的文件數,內層線程池pool執行單個文件的多線程下載。
downloadMainPool 代碼片段:
public static ExecutorService downloadMainPool = null;
static{
downloadMainPool = Executors.newFixedThreadPool(Constant.CONCURRENT_FILE_NUMBER,new ThreadFactory() {
public Thread newThread(Runnable r) {
Thread s = Executors.defaultThreadFactory().newThread(r);
s.setDaemon(true);
return s;
}
});
}
在static中實例化爲固定大小的線程池,由於默認的ThreadFactory創建的線程爲非守護狀態,爲了避免java程序不能退出的問題,保證在文件下載完後當前java程序結束在jvm中的運行,需要重寫ThreadFactory,使其創建的線程爲守護線程。
Constant類中自定義了程序中需要使用到的變量,可在類中直接定義或讀取配置文件,Constant.CONCURRENT_FILE_NUMBER定義了併發文件數。
downloadMainPool中的每個線程負責一個文件,每個線程下載文件時創建子線程池,由子線程池分塊下載文件;要做到斷點續傳需要記錄每個子線程池中的每個線程下載位置,這裏使用定時序列化子線程池線程對象的方式,定時將包含了下載位置的線程序列化到文件,在再次下載同一文件時反序列化,直接丟到子線程池下載即可。下面是downloadMainPool中的線程OSSDownloadFile代碼清單
OSSDownloadFile 代碼:
package cloudStorage.oss;
import java.io.File;
import java.util.concurrent.Callable;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.Future;
import java.util.concurrent.ThreadFactory;
import java.util.concurrent.TimeUnit;
import org.apache.log4j.Logger;
import cloudStorage.basis.Constant;
import cloudStorage.basis.Global;
import cloudStorage.basis.OSSClientFactory;
import cloudStorage.oss.download.DownloadPartObj;
import cloudStorage.oss.download.DownloadPartThread;
import cloudStorage.util.ObjectSerializableUtil;
import com.aliyun.oss.OSSClient;
import com.aliyun.oss.OSSException;
import com.aliyun.oss.model.ObjectMetadata;
/**
* @Description: oss多線程分段下載文件
* @author: zrk
* @time: 2015年4月1日 上午10:37:35
*/
public class OSSDownloadFile implements Callable<Integer>{
public static final Logger LOGGER = Logger.getLogger(OSSDownloadFile.class);
//外層線程池
public static ExecutorService downloadMainPool = null;
//內層線程池
private ExecutorService pool ;
static{
downloadMainPool = Executors.newFixedThreadPool(Constant.CONCURRENT_FILE_NUMBER,new ThreadFactory() {
public Thread newThread(Runnable r) {
Thread s = Executors.defaultThreadFactory().newThread(r);
s.setDaemon(true);
return s;
}
});
}
private String localFilePath;//本地文件路徑
private String bucketName; //bucketName
private String key;//雲端存儲路徑
public OSSDownloadFile() {
super();
}
public OSSDownloadFile(String localFilePath,String bucketName,String key) {
//初始化子線程池
pool = Executors.newFixedThreadPool(Constant.SINGLE_FILE_CONCURRENT_THREADS);
this.localFilePath = localFilePath;
this.bucketName = bucketName;
this.key = key;
}
//執行當前線程
public Integer downloadFile() {
Integer r = Global.ERROR;
//向downloadMainPool中submit當前線程
Future<Integer> result = downloadMainPool.submit(this);
try {
r=result.get();
} catch (InterruptedException | ExecutionException e) {
e.printStackTrace();
} finally{
return r;
}
}
/**
*
* @param localFilePath 需要存放的文件路徑
* @param bucketName bucketName
* @param key 存儲key -在oss的存儲路徑
* @return
*/
@Override
public Integer call(){
//OSSClient 使用單例
OSSClient client = OSSClientFactory.getInstance();
ObjectMetadata objectMetadata = null;
//判斷文件在雲端是否存在
try {
objectMetadata = client.getObjectMetadata(bucketName, key);
} catch (OSSException e) {
LOGGER.info("==請檢查bucketName或key");
return Global.ERROR;
}
long fileLength = objectMetadata.getContentLength();
//自定義的每個下載分塊大小
Integer partSize = Constant.DOWNLOAD_PART_SIZE;
//需要下載的文件分塊數
int partCount=calPartCount(fileLength, partSize);
//子線程池的線程對象封裝類(用於序列化的)
DownloadPartObj downloadPartObj = null;
boolean isSerializationFile = false;
//序列化的文件路徑(與下載文件同路徑使用.dw.temp後綴)
String serializationFilePath = localFilePath+".dw.temp";
//若存在反序列化對象
if(new File(serializationFilePath).exists()){
downloadPartObj = (DownloadPartObj)ObjectSerializableUtil.load(serializationFilePath);
isSerializationFile = true;
}
//序列化文件不存在,分配分塊給子線程池線程對象
if(downloadPartObj==null||!isSerializationFile){
downloadPartObj = new DownloadPartObj();
for (int i = 0; i < partCount; i++) {
final long startPos = partSize * i;
final long endPos = partSize * i +( partSize < (fileLength - startPos) ? partSize : (fileLength - startPos)) - 1;
//DownloadPartThread是執行每個分塊下載任務的線程
downloadPartObj.getDownloadPartThreads().add(new DownloadPartThread(startPos, endPos, localFilePath, bucketName, key,Constant.ACCESS_ID,Constant.ACCESS_KEY));
}
}
try {
int i = 0;
//download方法提交分塊下載線程至子線程池下砸,while循環用於下載失敗重複下載,Constant.RETRY定義重複下載次數
while (download(downloadPartObj,serializationFilePath).isResult()==false) {
if(++i == Constant.RETRY)break;
LOGGER.info(Thread.currentThread().getName()+"重試第"+i+"次");
}
} catch (Exception e) {
LOGGER.info("=="+e.getMessage());
return Global.THREAD_ERROR;
}
if(!downloadPartObj.isResult()){
return Global.NETWORK_ERROR;
}
return Global.SUCCESS;
}
/**
* 多線程下載單個文件
* @param partThreadObj
* @param serializationFilePath
* @return
*/
private DownloadPartObj download(DownloadPartObj partThreadObj,String serializationFilePath){
try {
partThreadObj.setResult(true);
//向子線程池中submit單個文件所有分塊下載線程
for (int i=0 ;i<partThreadObj.getDownloadPartThreads().size();i++) {
if (partThreadObj.getDownloadPartThreads().get(i).geteTag() == null)
pool.submit(partThreadObj.getDownloadPartThreads().get(i));
}
//shutdown子線程池,池內所下載任務執行結束後停止當前線程池
pool.shutdown();
//循環檢查線程池,同時在此序列化partThreadObj
while (!pool.isTerminated()) {
ObjectSerializableUtil.save(partThreadObj,serializationFilePath);
pool.awaitTermination(Constant.SERIALIZATION_TIME, TimeUnit.SECONDS);
}
//判斷下載結果
for (DownloadPartThread downloadPartThread: partThreadObj.getDownloadPartThreads()) {
if (downloadPartThread.geteTag() == null){
partThreadObj.setResult(false);
}
}
//下載成功 刪除序列化文件
if (partThreadObj.isResult()==true)
ObjectSerializableUtil.delSerlzFile(serializationFilePath);
} catch (Exception e) {
LOGGER.info("=="+e.getMessage());
}
return partThreadObj;
}
/**
* 獲取分塊數
* @param fileLength
* @param partSize
* @return
*/
private static int calPartCount(long fileLength,long partSize) {
int partCount = (int) (fileLength / partSize);
if (fileLength % partSize != 0){
partCount++;
}
return partCount;
}
public String getLocalFilePath() {
return localFilePath;
}
public void setLocalFilePath(String localFilePath) {
this.localFilePath = localFilePath;
}
public String getBucketName() {
return bucketName;
}
public void setBucketName(String bucketName) {
this.bucketName = bucketName;
}
public String getKey() {
return key;
}
public void setKey(String key) {
this.key = key;
}
}
此處引用了多個類
cloudStorage.basis.Constant;//定義程序中使用的變量
cloudStorage.basis.Global;//定義了全局的靜態值,錯誤狀態值
cloudStorage.basis.OSSClientFactory;//OSSClient工廠
cloudStorage.oss.download.DownloadPartObj;//分塊下載線程類封裝
cloudStorage.oss.download.DownloadPartThread;//分塊下載線程
cloudStorage.util.ObjectSerializableUtil;//序列工具類調下載方法是阻塞式,需要給調用者返回下結果,所以使用Callable和Future返回int型狀態值,下面是子線程池pool中的分塊下載線程DownloadPartThread的代碼清單
DownloadPartThread 代碼:
package cloudStorage.oss.download;
import java.io.File;
import java.io.IOException;
import java.io.RandomAccessFile;
import java.io.Serializable;
import java.util.Date;
import java.util.concurrent.Callable;
import org.apache.log4j.Logger;
import cloudStorage.basis.OSSClientFactory;
import com.aliyun.oss.common.utils.IOUtils;
import com.aliyun.oss.model.GetObjectRequest;
import com.aliyun.oss.model.OSSObject;
/**
* @Description: 用於上傳每個part的線程類 可序列化 用於上傳的斷點續傳
* @author: zrk
* @time: 2015年4月1日 上午10:35:34
*/
public class DownloadPartThread implements Callable<DownloadPartThread>,Serializable {
private static final long serialVersionUID = 1L;
public static final Logger LOGGER = Logger.getLogger(DownloadPartThread.class);
// 當前線程的下載開始位置
private long startPos;
// 當前線程的下載結束位置
private long endPos;
// 保存文件路徑
private String localFilePath;
private String bucketName;
private String fileKey;
private String eTag;
private String accessId;
private String accessKey;
public DownloadPartThread(long startPos, long endPos, String localFilePath,
String bucketName, String fileKey, String accessId,
String accessKey) {
this.startPos = startPos;
this.endPos = endPos;
this.localFilePath = localFilePath;
this.bucketName = bucketName;
this.fileKey = fileKey;
this.accessId = accessId;
this.accessKey = accessKey;
}
@Override
public DownloadPartThread call() {
RandomAccessFile file = null;
OSSObject ossObject = null;
try {
File pFile = new File(localFilePath);
if(!pFile.getParentFile().exists())
pFile.getParentFile().mkdirs();
file = new RandomAccessFile(localFilePath, "rw");
//調用ossapi
GetObjectRequest getObjectRequest = new GetObjectRequest(bucketName, fileKey);
getObjectRequest.setRange(startPos, endPos);
ossObject = OSSClientFactory.getInstance().getObject(getObjectRequest);
file.seek(startPos);
int bufSize = 1024;
byte[] buffer = new byte[bufSize];
int bytesRead;
while ((bytesRead = ossObject.getObjectContent().read(buffer)) > -1) {
file.write(buffer, 0, bytesRead);
//更新開始位置,保證在出錯後重下載是從上次結束的地方開始下,而不是下載整個塊
startPos += bytesRead;
}
this.eTag = ossObject.getObjectMetadata().getETag();
} catch (Exception e) {
LOGGER.info("=="+e.getMessage());
} finally{
if(ossObject!=null)IOUtils.safeClose(ossObject.getObjectContent());
try {if(file!=null)file.close();} catch (IOException e) {e.printStackTrace();}
return this;
}
}
}
每個DownloadPartThread下載線程執行結束都會將自己作爲返回值返回,當前文件是否下載完整有每個線程的返回值決定。
ObjectSerializableUtil 代碼 點擊查看
DownloadPartObj 代碼:
package cloudStorage.oss.download;
import java.io.Serializable;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
/**
* @Description: 單個文件的下載線程集合
* @author: zrk
* @time: 2015年5月5日 上午10:15:11
*/
public class DownloadPartObj implements Serializable{
private static final long serialVersionUID = 1L;
/**
* 下載線程集合
*/
List<DownloadPartThread> downloadPartThreads = Collections.synchronizedList(new ArrayList<DownloadPartThread>());
/**
* 下載結果
*/
boolean result = true;
public List<DownloadPartThread> getDownloadPartThreads() {
return downloadPartThreads;
}
public void setDownloadPartThreads(List<DownloadPartThread> downloadPartThreads) {
this.downloadPartThreads = downloadPartThreads;
}
public boolean isResult() {
return result;
}
public void setResult(boolean result) {
this.result = result;
}
}
所有下載當前文件的線程都會操作downloadPartThreads,所以downloadPartThreads使用集合Collections.synchronizedList將其轉換爲一個線程安全的類。DownloadPartObj封裝了downloadPartThreads和一個用於標識下載成功與失敗的boolean值,定時保存序列化文件就直接序列化DownloadPartObj,DownloadPartObj的線程安全是至關重要的。
OSSClientFactory 代碼:
package cloudStorage.basis;
import com.aliyun.oss.OSSClient;
public class OSSClientFactory {
private static OSSClient ossClient = null;
private OSSClientFactory() {
}
public static OSSClient getInstance() {
if (ossClient == null) {
// 可以使用ClientConfiguration對象設置代理服務器、最大重試次數等參數。
// ClientConfiguration config = new ClientConfiguration();
ossClient = new OSSClient(Constant.OSS_ENDPOINT,Constant.ACCESS_ID, Constant.ACCESS_KEY);
}
return ossClient;
}
}
Constant 代碼:
package cloudStorage.basis;
import cloudStorage.service.OSSConfigService;
/**
* @Description:
* @author: zrk
* @time: 2015年4月1日 下午5:22:28
*/
public class Constant {
public static String OSS_ENDPOINT = "http://oss.aliyuncs.com/";
public static String ACCESS_ID;
public static String ACCESS_KEY;
public static Integer DOWNLOAD_PART_SIZE ; // 每個下載Part的大小
public static Integer UPLOAD_PART_SIZE ; // 每個上傳Part的大小
public static int CONCURRENT_FILE_NUMBER ; // 併發文件數。
public static int SINGLE_FILE_CONCURRENT_THREADS ; // 單文件併發線程數。
public static int RETRY ;//失敗重試次數
public static int SERIALIZATION_TIME;//斷點保存時間間隔(秒)
//。。。
}
Constant中數值是加載外部配置文件,也可在這兒直接配置,個別參數值斷點上傳時使用,只看下載的話請忽略。
Global代碼:
package cloudStorage.basis;
/**
* @Description: TODO
* @author: zrk
* @time: 2015年4月1日 下午5:22:46
*/
public class Global {
public static final int SUCCESS = 1;
public static final int ERROR = 10;
public static final int FILE_NOT_FOUND_ERROR = 11;
public static final int THREAD_ERROR = 12;
public static final int NETWORK_ERROR = 13;
public static final int OSS_SUBMIT_ERROR = 14;
// META
public static final String X_OSS_META_MY_MD5 = "x-oss-meta-my-md5";
}
調用下載方式:
實例化OSSDownloadFile後調用downloadFile方法:
return new OSSDownloadFile(localFilePath,bucketName, key).downloadFile();
OSS SDK文件的分塊下載支持的很好,OSS官方的SDK裏面也提供了一個多線程下載功能的實現,所以在實現文件的分塊下載並沒有什麼難度。這塊兒多線程下載僅供大家參考,後面自己在使用過程中繼續優化。基於OSS-SDK的分塊斷點上傳請參考OSS實現多文件多線程的斷點上傳(java) 。
本文地址:http://blog.csdn.net/zrk1000/article/details/46286767