/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
package org.apache.commons.compress.compressors.xz;
import java.io.IOException;
import java.io.InputStream;
import org.tukaani.xz.XZ;
import org.tukaani.xz.SingleXZInputStream;
import org.tukaani.xz.XZInputStream;
import org.apache.commons.compress.MemoryLimitException;
import org.apache.commons.compress.compressors.CompressorInputStream;
import org.apache.commons.compress.utils.CountingInputStream;
import org.apache.commons.compress.utils.IOUtils;
import org.apache.commons.compress.utils.InputStreamStatistics;
XZ decompressor.
Since: 1.4
/**
* XZ decompressor.
* @since 1.4
*/
public class XZCompressorInputStream extends CompressorInputStream
implements InputStreamStatistics {
private final CountingInputStream countingStream;
private final InputStream in;
Checks if the signature matches what is expected for a .xz file.
Params: - signature – the bytes to check
- length – the number of bytes to check
Returns: true if signature matches the .xz magic bytes, false otherwise
/**
* Checks if the signature matches what is expected for a .xz file.
*
* @param signature the bytes to check
* @param length the number of bytes to check
* @return true if signature matches the .xz magic bytes, false otherwise
*/
public static boolean matches(final byte[] signature, final int length) {
if (length < XZ.HEADER_MAGIC.length) {
return false;
}
for (int i = 0; i < XZ.HEADER_MAGIC.length; ++i) {
if (signature[i] != XZ.HEADER_MAGIC[i]) {
return false;
}
}
return true;
}
Creates a new input stream that decompresses XZ-compressed data
from the specified input stream. This doesn't support
concatenated .xz files.
Params: - inputStream – where to read the compressed data
Throws: - IOException – if the input is not in the .xz format,
the input is corrupt or truncated, the .xz
headers specify options that are not supported
by this implementation, or the underlying
inputStream
throws an exception
/**
* Creates a new input stream that decompresses XZ-compressed data
* from the specified input stream. This doesn't support
* concatenated .xz files.
*
* @param inputStream where to read the compressed data
*
* @throws IOException if the input is not in the .xz format,
* the input is corrupt or truncated, the .xz
* headers specify options that are not supported
* by this implementation, or the underlying
* <code>inputStream</code> throws an exception
*/
public XZCompressorInputStream(final InputStream inputStream)
throws IOException {
this(inputStream, false);
}
Creates a new input stream that decompresses XZ-compressed data
from the specified input stream.
Params: - inputStream – where to read the compressed data
- decompressConcatenated –
if true, decompress until the end of the
input; if false, stop after the first .xz
stream and leave the input position to point
to the next byte after the .xz stream
Throws: - IOException – if the input is not in the .xz format,
the input is corrupt or truncated, the .xz
headers specify options that are not supported
by this implementation, or the underlying
inputStream
throws an exception
/**
* Creates a new input stream that decompresses XZ-compressed data
* from the specified input stream.
*
* @param inputStream where to read the compressed data
* @param decompressConcatenated
* if true, decompress until the end of the
* input; if false, stop after the first .xz
* stream and leave the input position to point
* to the next byte after the .xz stream
*
* @throws IOException if the input is not in the .xz format,
* the input is corrupt or truncated, the .xz
* headers specify options that are not supported
* by this implementation, or the underlying
* <code>inputStream</code> throws an exception
*/
public XZCompressorInputStream(final InputStream inputStream,
final boolean decompressConcatenated)
throws IOException {
this(inputStream, decompressConcatenated, -1);
}
Creates a new input stream that decompresses XZ-compressed data
from the specified input stream.
Params: - inputStream – where to read the compressed data
- decompressConcatenated –
if true, decompress until the end of the
input; if false, stop after the first .xz
stream and leave the input position to point
to the next byte after the .xz stream
- memoryLimitInKb – memory limit used when reading blocks. If the estimated memory limit is exceeded on
read()
, a MemoryLimitException
is thrown.
Throws: - IOException – if the input is not in the .xz format,
the input is corrupt or truncated, the .xz
headers specify options that are not supported
by this implementation,
or the underlying
inputStream
throws an exception
Since: 1.14
/**
* Creates a new input stream that decompresses XZ-compressed data
* from the specified input stream.
*
* @param inputStream where to read the compressed data
* @param decompressConcatenated
* if true, decompress until the end of the
* input; if false, stop after the first .xz
* stream and leave the input position to point
* to the next byte after the .xz stream
* @param memoryLimitInKb memory limit used when reading blocks. If
* the estimated memory limit is exceeded on {@link #read()},
* a {@link MemoryLimitException} is thrown.
*
* @throws IOException if the input is not in the .xz format,
* the input is corrupt or truncated, the .xz
* headers specify options that are not supported
* by this implementation,
* or the underlying <code>inputStream</code> throws an exception
*
* @since 1.14
*/
public XZCompressorInputStream(InputStream inputStream,
boolean decompressConcatenated, final int memoryLimitInKb)
throws IOException {
countingStream = new CountingInputStream(inputStream);
if (decompressConcatenated) {
in = new XZInputStream(countingStream, memoryLimitInKb);
} else {
in = new SingleXZInputStream(countingStream, memoryLimitInKb);
}
}
@Override
public int read() throws IOException {
try {
final int ret = in.read();
count(ret == -1 ? -1 : 1);
return ret;
} catch (org.tukaani.xz.MemoryLimitException e) {
throw new MemoryLimitException(e.getMemoryNeeded(), e.getMemoryLimit(), e);
}
}
@Override
public int read(final byte[] buf, final int off, final int len) throws IOException {
if (len == 0) {
return 0;
}
try {
final int ret = in.read(buf, off, len);
count(ret);
return ret;
} catch (org.tukaani.xz.MemoryLimitException e) {
//convert to commons-compress MemoryLimtException
throw new MemoryLimitException(e.getMemoryNeeded(), e.getMemoryLimit(), e);
}
}
@Override
public long skip(final long n) throws IOException {
try {
return IOUtils.skip(in, n);
} catch (org.tukaani.xz.MemoryLimitException e) {
//convert to commons-compress MemoryLimtException
throw new MemoryLimitException(e.getMemoryNeeded(), e.getMemoryLimit(), e);
}
}
@Override
public int available() throws IOException {
return in.available();
}
@Override
public void close() throws IOException {
in.close();
}
Since: 1.17
/**
* @since 1.17
*/
@Override
public long getCompressedCount() {
return countingStream.getBytesRead();
}
}