GZIPInputStream的bug，还未到文件尾，read方法直接返回-1

关于GZIPInputStream的bug，在jdk的最新版本上竟然还没解决这个问题。用到gzip的需要注意了：

http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4691425

问题描述：

在使用GZIPInputStream对gizp文件进行读取的时候，使用read方法，方法返回-1，表示该文件读取真正的结束，但是实际却并不是这样。


这些文件在windowsxp用win-rar解压或者linux上zcat aaa.gz命令进行解压后，是可以获得文件的全部内容的。但是在使用GZIPInputStream却不行。

这个bug是java在读取gzip文件的时候，还没有读取完毕，过早的返回-1，造成有部分文件没有读取完毕。

不过好在已经有人解决这个问题了，下面是解决的方法：

// License: GPL. Copyright 2007-2008 by Brett Henderson and other contributors.
package org.openstreetmap.osmosis.core.util;

import java.io.InputStream;
import java.io.PushbackInputStream;
import java.io.IOException;
import java.util.zip.GZIPInputStream;

/**
* This class was copied directly from the workaround class provided in
* http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4691425.
*/
public class MultiMemberGZIPInputStream extends GZIPInputStream {

/**
* Creates a new instance.
*
* @param in
* The input stream.
* @param size
* The buffer size.
* @throws IOException
* if an IO exception occurs.
*/
public MultiMemberGZIPInputStream(InputStream in, int size)
throws IOException {
// Wrap the stream in a PushbackInputStream...
super(new PushbackInputStream(in, size), size);
this.size = size;
}

/**
* Creates a new instance.
*
* @param in
* The input stream.
* @throws IOException
* if an IO exception occurs.
*/
public MultiMemberGZIPInputStream(InputStream in) throws IOException {
// Wrap the stream in a PushbackInputStream...
super(new PushbackInputStream(in, 1024));
this.size = -1;
}

private MultiMemberGZIPInputStream(MultiMemberGZIPInputStream parent)
throws IOException {
super(parent.in);
this.size = -1;
if (parent.parent == null) {
this.parent = parent;
} else {
this.parent = parent.parent;
}
this.parent.child = this;
}

private MultiMemberGZIPInputStream(MultiMemberGZIPInputStream parent,
int size) throws IOException {
super(parent.in, size);
this.size = size;
if (parent.parent == null) {
this.parent = parent;
} else {
this.parent = parent.parent;
}
this.parent.child = this;
}

private MultiMemberGZIPInputStream parent;
private MultiMemberGZIPInputStream child;
private int size;
private boolean eos;

/**
* {@inheritDoc}
*/
@Override
public int read(byte[] inputBuffer, int inputBufferOffset,
int inputBufferLen) throws IOException {

if (eos) {
return -1;
}
if (this.child != null) {
return this.child.read(inputBuffer, inputBufferOffset,
inputBufferLen);
}

int charsRead = super.read(inputBuffer, inputBufferOffset,
inputBufferLen);
if (charsRead == -1) {
// Push any remaining buffered data back onto the stream
// If the stream is then not empty, use it to construct
// a new instance of this class and delegate this and any
// future calls to it...
int n = inf.getRemaining() - 8;
if (n > 0) {
// More than 8 bytes remaining in deflater
// First 8 are gzip trailer. Add the rest to
// any un-read data...
((PushbackInputStream) this.in).unread(buf, len - n, n);
} else {
// Nothing in the buffer. We need to know whether or not
// there is unread data available in the underlying stream
// since the base class will not handle an empty file.
// Read a byte to see if there is data and if so,
// push it back onto the stream...
byte[] b = new byte[1];
int ret = in.read(b, 0, 1);
if (ret == -1) {
eos = true;
return -1;
} else {
((PushbackInputStream) this.in).unread(b, 0, 1);
}
}

MultiMemberGZIPInputStream tmpChild;
if (this.size == -1) {
tmpChild = new MultiMemberGZIPInputStream(this);
} else {
tmpChild = new MultiMemberGZIPInputStream(this, this.size);
}
return tmpChild.read(inputBuffer, inputBufferOffset, inputBufferLen);
} else {
return charsRead;
}
}

}

重写了GZIPInputStream这个类。然后你在使用的时候直接引入MultiMemberGZIPInputStream这个类，并调用即可。 例如： BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(new MultiMemberGZIPInputStream(new FileInputStream(gzFile)),

转载自： http://cin-ie.iteye.com/blog/859822