Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
@@ -0,0 +1,220 @@
/*
* 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.hadoop.hbase.io.compress;

import edu.umd.cs.findbugs.annotations.Nullable;
import java.io.IOException;
import java.util.zip.CRC32;
import java.nio.ByteBuffer;
import java.util.zip.DataFormatException;
import java.util.zip.Inflater;
import org.apache.hadoop.hbase.nio.ByteBuff;
import org.apache.hadoop.hbase.nio.SingleByteBuff;
import org.apache.hadoop.io.compress.zlib.ZlibDecompressor;
import org.apache.yetus.audience.InterfaceAudience;

/**
* Glue for ByteBuffDecompressor on top of Hadoop's native
* {@link ZlibDecompressor.ZlibDirectDecompressor}.
*/
@InterfaceAudience.Private
public class GzipByteBuffDecompressor implements ByteBuffDecompressor {

private static final int GZIP_HEADER_LENGTH = 10;
private static final int GZIP_TRAILER_LENGTH = 8;

@Nullable

@SaadASTheDev SaadASTheDev Aug 7, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These are the Singleton objects reused across decompress calls, Allocated once per buffer, to avoid GC overhead

private final ZlibDecompressor.ZlibDirectDecompressor decompressor;

private final Inflater inflater = new Inflater(true);

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For consistency, I suggest you use the top-level ZlibDecompressor to handle on-heap

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will do, wasn't familiar with the top level decompressor but that simplifies things, Thanks!


private final CRC32 crc32 = new CRC32();

private boolean allowByteBuffDecompression;

GzipByteBuffDecompressor(boolean nativeZlibLoaded) {
decompressor = nativeZlibLoaded
? new ZlibDecompressor.ZlibDirectDecompressor(ZlibDecompressor.CompressionHeader.GZIP_FORMAT,
0)
: null;
allowByteBuffDecompression = true;
}

@Override

@SaadASTheDev SaadASTheDev Aug 7, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The flow for wether we can decompress is
1: verify config via context to ensure we havent disabled this feature before we try to decompress via bytebuff
2. we only support singleByteBuff decompression for now
3. Since Zlib is a JNI we need to ensure that its properly instantiated if and only if the buffers we are trying to decompress are direct, so if both are direct we ensure the decompressor is loaded
4. then we return if both I/O are not direct, if true it uses the inflater pathway

public boolean canDecompress(ByteBuff output, ByteBuff input) {
if (!allowByteBuffDecompression) {
return false;
}
if (!(output instanceof SingleByteBuff) || !(input instanceof SingleByteBuff)) {
return false;
}
boolean inputDirect = input.nioByteBuffers()[0].isDirect();
boolean outputDirect = output.nioByteBuffers()[0].isDirect();
if (inputDirect && outputDirect) {
return decompressor != null;
}
return !inputDirect && !outputDirect;
}

@Override
public int decompress(ByteBuff output, ByteBuff input, int inputLen) throws IOException {
if (!(output instanceof SingleByteBuff) || !(input instanceof SingleByteBuff)) {
throw new IllegalStateException(
"At least one buffer is not a SingleByteBuff, this is not supported");
}
if (inputLen < GZIP_HEADER_LENGTH + GZIP_TRAILER_LENGTH) {
throw new IOException("Input of length " + inputLen + " is too short to be a gzip member");
}

ByteBuffer nioInput = input.nioByteBuffers()[0];
ByteBuffer nioOutput = output.nioByteBuffers()[0];
boolean inputDirect = nioInput.isDirect();
boolean outputDirect = nioOutput.isDirect();

if (inputDirect && outputDirect) {
if (decompressor == null) {
throw new IllegalStateException(
"GzipByteBuffDecompressor#decompress() was called with direct buffers but Hadoop's "
+ "native zlib library is not loaded, this should never happen since "
+ "canDecompress() would have returned false");
}
return decompressOffHeap(nioInput, nioOutput, inputLen);
}
return decompressOnHeap(nioInput, nioOutput, inputLen);
}

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Respective off heap decompress which uses the zlib library and verifies the trailer on its own

private int decompressOffHeap(ByteBuffer nioInput, ByteBuffer nioOutput, int inputLen)
throws IOException {
int inputStart = nioInput.position();
int outputStart = nioOutput.position();

ByteBuffer gzipMember = nioInput.duplicate();
gzipMember.limit(inputStart + inputLen);

decompressor.reset();
while (!decompressor.finished()) {
int outputRemainingBefore = nioOutput.remaining();
try {
decompressor.decompress(gzipMember, nioOutput);
} catch (IOException e) {
throw new IOException("Invalid gzip stream: " + e.getMessage(), e);
}
if (nioOutput.remaining() == outputRemainingBefore && !decompressor.finished()) {
if (!nioOutput.hasRemaining()) {
throw new IOException("Output buffer is too small for the decompressed gzip stream");
}
throw new IOException("Unexpected end of gzip stream");
}
}
Comment on lines +110 to +123

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there a reason this needs to loop? Why would the decompressor need multiple attempts?

@SaadASTheDev SaadASTheDev Aug 7, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yea It doesnt need to loop. I was under the impression the zlib decompressor only follows a streaming, chunk-based pattern, where a block gets decompressed in chunks. Thats only true if we don't pre size the output buffer, going to fix this, Thanks


if (gzipMember.hasRemaining()) {
throw new IOException("Unexpected trailing bytes after decompressing gzip stream");
}

nioInput.position(inputStart + inputLen);
return nioOutput.position() - outputStart;
}

// ZlibDirectDecompressor requires direct buffers — heap ByteBuffers have no stable native
// address, so we fall back to Java's Inflater for the heap case.
private int decompressOnHeap(ByteBuffer nioInput, ByteBuffer nioOutput, int inputLen)
throws IOException {
if (!nioInput.hasArray() || !nioOutput.hasArray()) {
throw new IllegalStateException(
"decompressOnHeap() requires heap ByteBuffers with backing arrays");
}
int inputStart = nioInput.position();
int outputStart = nioOutput.position();

inflater.reset();
inflater.setInput(nioInput.array(), nioInput.arrayOffset() + inputStart + GZIP_HEADER_LENGTH,
inputLen - GZIP_HEADER_LENGTH - GZIP_TRAILER_LENGTH);
int totalDecompressed = 0;
while (!inflater.finished()) {
int remaining = nioOutput.remaining() - totalDecompressed;
if (remaining == 0) {
throw new IOException("Output buffer is too small for the decompressed gzip stream");
}
int n;
try {
n = inflater.inflate(nioOutput.array(),
nioOutput.arrayOffset() + outputStart + totalDecompressed, remaining);
} catch (DataFormatException e) {
throw new IOException("Invalid gzip stream: " + e.getMessage(), e);
}
if (n == 0 && !inflater.finished()) {
if (inflater.needsInput()) {
throw new IOException("Unexpected end of gzip stream");
}
throw new IOException("Unexpected state in gzip stream");
}
totalDecompressed += n;
}
verifyGzipTrailer(nioInput.array(), nioInput.arrayOffset() + inputStart, inputLen,
nioOutput.array(), nioOutput.arrayOffset() + outputStart, totalDecompressed);
nioOutput.position(outputStart + totalDecompressed);
nioInput.position(inputStart + inputLen);
return totalDecompressed;
}

// Inflater runs in nowrap (raw DEFLATE) mode and is unaware of the gzip envelope, so it never
// checks the trailer. ZlibDirectDecompressor handles this automatically via GZIP_FORMAT, but
// for heap buffers we must verify the CRC32 and ISIZE fields ourselves.
private void verifyGzipTrailer(byte[] inputData, int inputDataOffset, int inputLen,
byte[] outputData, int outputDataOffset, int decompressedLen) throws IOException {
long expectedCrc = readLittleEndianUInt32(inputData, inputDataOffset + inputLen - 8);
long expectedSize = readLittleEndianUInt32(inputData, inputDataOffset + inputLen - 4);
crc32.reset();
crc32.update(outputData, outputDataOffset, decompressedLen);
if (crc32.getValue() != expectedCrc) {
throw new IOException("Gzip CRC32 mismatch");
}
if ((decompressedLen & 0xFFFFFFFFL) != expectedSize) {
throw new IOException("Gzip size mismatch");
}
}

private static long readLittleEndianUInt32(byte[] data, int offset) {
return (data[offset] & 0xFFL) | ((data[offset + 1] & 0xFFL) << 8)
| ((data[offset + 2] & 0xFFL) << 16) | ((data[offset + 3] & 0xFFL) << 24);
}

@Override
public void reinit(@Nullable Compression.HFileDecompressionContext newHFileDecompressionContext) {

@SaadASTheDev SaadASTheDev Aug 7, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We load in a new context at every new Hfile, it would be too expensive to do it at every block

if (newHFileDecompressionContext == null) {
return;
}
if (!(newHFileDecompressionContext instanceof GzipHFileDecompressionContext)) {
throw new IllegalArgumentException(
"GzipByteBuffDecompressor#reinit() was given an HFileDecompressionContext that was not "
+ "a GzipHFileDecompressionContext, this should never happen");
}
GzipHFileDecompressionContext gzipContext =
(GzipHFileDecompressionContext) newHFileDecompressionContext;
allowByteBuffDecompression = gzipContext.isAllowByteBuffDecompression();
}

@Override
public void close() {
inflater.end();
if (decompressor != null) {
decompressor.end();
}
}

}
Original file line number Diff line number Diff line change
@@ -0,0 +1,68 @@
/*
* 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.hadoop.hbase.io.compress;

import java.io.IOException;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.hbase.util.ClassSize;
import org.apache.yetus.audience.InterfaceAudience;

/**
* Holds HFile-level settings used by GzipByteBuffDecompressor. It's expensive to pull these from a
* Configuration object every time we decompress a block, so pull them upon opening an HFile, and
* reuse them in every block that gets decompressed.
*/
@InterfaceAudience.Private
public final class GzipHFileDecompressionContext extends Compression.HFileDecompressionContext {

public static final long FIXED_OVERHEAD =
ClassSize.estimateBase(GzipHFileDecompressionContext.class, false);

public static final String ALLOW_BYTE_BUFF_DECOMPRESSION_KEY =
"hbase.io.compress.gz.allowByteBuffDecompression";

private final boolean allowByteBuffDecompression;

private GzipHFileDecompressionContext(boolean allowByteBuffDecompression) {
this.allowByteBuffDecompression = allowByteBuffDecompression;
}

public boolean isAllowByteBuffDecompression() {
return allowByteBuffDecompression;
}

public static GzipHFileDecompressionContext fromConfiguration(Configuration conf) {
return new GzipHFileDecompressionContext(
conf.getBoolean(ALLOW_BYTE_BUFF_DECOMPRESSION_KEY, true));
}

@Override
public void close() throws IOException {
}

@Override
public long heapSize() {
return FIXED_OVERHEAD;
}

@Override
public String toString() {
return "GzipHFileDecompressionContext{allowByteBuffDecompression=" + allowByteBuffDecompression
+ '}';
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@
import java.io.OutputStream;
import java.util.Arrays;
import java.util.zip.GZIPOutputStream;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.hbase.util.JVM;
import org.apache.hadoop.io.compress.CompressionOutputStream;
import org.apache.hadoop.io.compress.CompressorStream;
Expand All @@ -35,9 +36,9 @@
* Fixes an inefficiency in Hadoop's Gzip codec, allowing to reuse compression streams.
*/
@InterfaceAudience.Private
public class ReusableStreamGzipCodec extends GzipCodec {
public class ReusableStreamGzipCodec extends GzipCodec implements ByteBuffDecompressionCodec {

private static final Logger LOG = LoggerFactory.getLogger(ReusableStreamGzipCodec.class);
private static final Logger LOG = LoggerFactory.getLogger(Compression.class);

/**
* A bridge that wraps around a DeflaterOutputStream to make it a CompressionOutputStream.
Expand Down Expand Up @@ -185,4 +186,20 @@ public CompressionOutputStream createOutputStream(OutputStream out) throws IOExc
return new ReusableGzipOutputStream(out);
}

@Override
public ByteBuffDecompressor createByteBuffDecompressor() {
return new GzipByteBuffDecompressor(ZlibFactory.isNativeZlibLoaded(getConf()));
}

@Override
public Class<? extends ByteBuffDecompressor> getByteBuffDecompressorType() {
return GzipByteBuffDecompressor.class;
}

@Override
public Compression.HFileDecompressionContext
getDecompressionContextFromConfiguration(Configuration conf) {
return GzipHFileDecompressionContext.fromConfiguration(conf);
}

}
Loading
Loading