Snappy Java API簡介

Snappy是Google開源的壓縮/解壓縮庫。和其餘壓縮庫相比，snappy的壓縮率並非最高的，兼容性也並不是最好的。相反，它的誕生旨在以極高的壓縮/解壓縮速率提供合理的壓縮率。Snappy官指出：在64位單核core-i7處理器下，snappy的壓縮率可以達到250MB/S，而解壓縮速度則能達到500MB/S。目前不少軟件使用（或支持）snappy做爲壓縮庫，如MongoDB,Cassandra,Hadoop,Lucene…本文簡單介紹snappy Java Api的使用。

Snappy提供了不少高級API以方便對byte[],Sting,double[]等進行壓縮和解壓縮，示例代碼以下：
 

import org.xerial.snappy.Snappy;
String input = "Hello snappy-java! Snappy-java is a JNI-based wrapper of Snappy, a fast compresser/decompresser.";
{ 
    byte[] compressed = Snappy.compress(input.getBytes("UTF-8"));
    byte[] uncompressed = Snappy.uncompress(compressed);
    String result = new String(uncompressed, "UTF-8");
    System.out.println(result);
}

{
    byte[] compressed = Snappy.compress(input);
    System.out.println(Snappy.uncompressString(compressed));
}

{
    double [] arr = new double[]{123.456,234.567,345.678};
    byte[] compressed = Snappy.compress(arr);
    double [] unarr = Snappy.uncompressDoubleArray(compressed);
    System.out.println(Arrays.toString(unarr));
}

同時Snappy仍然保留着低級API，用於最小化內存的拷貝，例如

/*
*inputAddr:待壓縮數據的內存地址
*inputSize:待壓縮數據的byte size
*destAddr:壓縮結果的保存地址
*return：返回壓縮後數據的大小
*/
public static long rawCompress(long inputAddr,long inputSize, long destAddr)throws java.io.IOException;
/*
*inputAddr:待解壓縮數據的內存地址
*inputSize:待解壓縮數據的byte size
*destAddr:解壓縮結果的保存地址
*return：返回解壓縮後數據的大小
*/
public static long rawUncompress(long inputAddr,long inputSize,long destAddr)throws java.io.IOException;

基於輸入/輸出流的Snappy解壓縮 
SnappyOutputStream和SnappyInputStream分別用於流數據的壓縮/解壓縮。此外，從snappy v1.1.0開始提供了Framing-format(幀格式)輸入輸出流的壓縮/解壓縮的方法：SnappyFramedOutputStream和SnappyFramedInputStream。須要注意的是以SnappyOutputStream壓縮的數據不能以SnappyFramedInputStream方法解開，反之亦然。下面以SnappyOutputStream和SnappyInputStream爲例介紹。 
SnappyOutputStream繼承自java.io OutputStream,重寫了經常使用的的write,close,flush:
 

File file = new File("..."); //待壓縮文件
File out = new File("./", file.getName() + ".snappy"); //壓縮結果文件

byte[] buffer = new byte[1024 * 1024 * 8];
FileInputStream  fi = null;
FileOutputStream fo = null;
SnappyOutputStream sout = null;
try
{
    fi = new FileInputStream(file); 
    fo = new FileOutputStream(out);
    sout = new SnappyOutputStream(fo);
    while(true)
    {
        int count = fi.read(buffer, 0, buffer.length);
        if(count == -1) { break; }
        sout.write(buffer, 0, count);
    }
    sout.flush();
}
catch(Throwable ex)
{
    ex.printStackTrace();
}
finally 
{
    if(sout != null) {try { sout.close();} catch (Exception e) {}}
    if(fi != null) { try { fi.close(); } catch(Exception x) {} }
    if(fo != null) { try { fo.close(); } catch(Exception x) {} }
}

SnappyInputStream 繼承自java.io InputStream,重寫了經常使用的read,close方法等

File file = new File("xxx"); //待解壓文件
File out = new File("xxx");  //解壓後文件

byte[] buffer = new byte[1024 * 1024 * 8];
FileInputStream  fi = null;
FileOutputStream fo = null;
SnappyInputStream sin = null;
try
{
    fo = new FileOutputStream(out);
    fi = new FileInputStream(file.getPath());
    sin = new SnappyInputStream(fi);        

    while(true)
    {
        int count = sin.read(buffer, 0, buffer.length);
        if(count == -1) { break; }
        fo.write(buffer, 0, count);
    }
    fo.flush();
}
catch(Throwable ex)
{
    ex.printStackTrace();
}
finally 
{
    if(sin != null) { try { sin.close(); } catch(Exception x) {} }
    if(fi != null) { try { fi.close(); } catch(Exception x) {} }
    if(fo != null) { try { fo.close(); } catch(Exception x) {} }
}