一、创建一个大文件

下载文件时往往会创建一个指定大小的空文件

package com.lazy.nio;

import java.io.IOException;

import java.nio.ByteBuffer;

import java.nio.channels.FileChannel;

import java.nio.file.Files;

import java.nio.file.Path;

import java.nio.file.Paths;

import java.nio.file.StandardOpenOption;

/**

* 创建大文件

* @author Lazy Gene

*

*/

public class FileCreator {

public static void main(String[] args) {

FileCreator creator = new FileCreator();

creator.createBigEmptyFile();

}

void createBigEmptyFile(){

Path filePath = Paths.get("from/test.tmp");

// 这段代码实际上可以在FIleChannel 调用open方式时指定OpenOption 为Create_NEW

try {

if (!Files.exists(filePath)) {

Files.createFile(filePath);

}

} catch (IOException e1) {

e1.printStackTrace();

}

//写一个字节

ByteBuffer buffer = ByteBuffer.allocate(1);

try (FileChannel fileChannel = FileChannel.open(filePath, StandardOpenOption.CREATE, StandardOpenOption.WRITE)) {

fileChannel.position(2L << 32 - 1); //移动位置, 生成一个4G的空文件

fileChannel.write(buffer);

} catch (IOException e) {

e.printStackTrace();

}

}

}

二、文件转移

NIO 提供transferTo tansferFrom, 和传统的文件访问方式相比减少了数据从内核到用户空间的复制，数据直接在内核移动，在Linux系统中使用sendfile系统调用

这里分别通过FileChannel.transferFrom 和Files.copy以及普通的io调用实现文件的复制

package com.lazy.nio;

import java.io.FileInputStream;

import java.io.FileNotFoundException;

import java.io.FileOutputStream;

import java.io.IOException;

import java.io.InputStream;

import java.io.OutputStream;

import java.nio.channels.FileChannel;

import java.nio.file.Files;

import java.nio.file.Path;

import java.nio.file.Paths;

import java.nio.file.StandardOpenOption;

/**

* 文件转移

*

* @author Lazy Gene(ljgeng@iflytek.com)

*

*/

public class FileTransfer {

Path fromPath = Paths.get("from/test.tmp");

Path toPath = Paths.get("to/test.tmp");

public static void main(String[] args) throws IOException {

FileTransfer transfer = new FileTransfer();

long start = System.currentTimeMillis();

transfer.transferFrom();

long end1 = System.currentTimeMillis();

System.out.println("transferFrom: "+ (end1 - start));

Files.deleteIfExists(transfer.toPath);

transfer.copy();

long end2 = System.currentTimeMillis();

System.out.println("Files copy: "+(end2 - end1));

Files.deleteIfExists(transfer.toPath);

transfer.ioCopy();

long end3 = System.currentTimeMillis();

System.out.println("original io: "+(end3 - end2));

}

void transferFrom() {

try (FileChannel channelFrom = FileChannel.open(fromPath, StandardOpenOption.READ);

FileChannel channelTo = FileChannel.open(toPath, StandardOpenOption.CREATE_NEW,

StandardOpenOption.WRITE);) {

channelTo.transferFrom(channelFrom, 0L, channelFrom.size());

} catch (IOException e) {

e.printStackTrace();

}

}

void copy() {

try {

Files.copy(fromPath, toPath);

} catch (IOException e) {

e.printStackTrace();

}

}

void ioCopy() {

try (InputStream is = new FileInputStream(fromPath.toFile());

OutputStream os = new FileOutputStream(toPath.toFile());) {

byte[] buffer = new byte[4096];

int byteread = 0;

while ((byteread = is.read(buffer)) != -1) {

os.write(buffer, 0, byteread);

}

} catch (FileNotFoundException e) {

e.printStackTrace();

} catch (IOException e1) {

e1.printStackTrace();

}

}

}

执行结果

transferFrom: 32726

Files copy: 33002

original io: 112975

将ioCopy方法中每次读取字节数调大十倍和百倍后结果

original io: 54743 (十倍)

original io: 93110 (百倍)

三、FileChannel.map 的使用

FileChannel.map 将文件按照一定大小块映射到内存区域，程序方式内存区域操作文件，适合大文件只读操作，如大文件的md5校验。

package com.lazy.nio;

import java.io.File;

import java.io.FileInputStream;

import java.io.IOException;

import java.io.RandomAccessFile;

import java.nio.channels.FileChannel;

import java.nio.file.Path;

import java.nio.file.Paths;

import java.security.MessageDigest;

import java.security.NoSuchAlgorithmException;

/**

* 使用介绍

* @author Lazy Gene

*

*/

public class FileChannelMap {

public static void main(String[] args) throws NoSuchAlgorithmException, IOException {

// 计算test.tmp 的md5值，将文件分成多块计算

Path path = Paths.get("from/test.tmp");

File file = path.toFile();

RandomAccessFile accessFile = new RandomAccessFile(file, "r");

MessageDigest MD5 = MessageDigest.getInstance("MD5");

long start = System.currentTimeMillis();

long eachSize = 1024 * 1024L;

long length = file.length();

int count = 1 + (int) (length / eachSize); //分块数量

long remaining = length; // 剩下的大小

for (int i=0;i

MD5.update(accessFile.getChannel().map(FileChannel.MapMode.READ_ONLY, i * eachSize, Math.min(remaining, eachSize)));

remaining -= eachSize;

}

accessFile.close();

long end = System.currentTimeMillis();

System.out.println("用时："+(end - start));

System.out.println("md5: "+ new String(Hex.encodeHex(MD5.digest())));

//方法二

MessageDigest new_MD5 = MessageDigest.getInstance("MD5");

FileInputStream in=new FileInputStream(file);

byte[] buffer=new byte[65536];

int rv=0;

while((rv=in.read(buffer))>0) {

new_MD5.update(buffer,0,rv);

}

long end1 = System.currentTimeMillis();

in.close();

System.out.println("用时："+(end1 - end));

System.out.println("io md5: "+ new String(Hex.encodeHex(new_MD5.digest())));

}

}

对于4g文件运行结果：

用时：15172

md5: f18c798ff5d450dfe4d3acdc12b621ff

用时：15811

io md5: f18c798ff5d450dfe4d3acdc12b621ff

差别不大呀，将文件调到16g，运行结果

用时：65046

md5: 0eb76b1bf69255feec7bdf4a3b5e2805

用时：62697

io md5: 0eb76b1bf69255feec7bdf4a3b5e2805

这里只所以差别不大，估计是应用map操作和操作系统的底层io实现相关，下次换成linux看看。