snappy-java is a Java port of the snappy, a fast C++ compresser/decompresser developed by Google.

• Fast compression/decompression around 200~400MB/sec.
• Less memory usage. SnappyOutputStream uses only 32KB+ in default.
• JNI-based implementation to achieve comparable performance to the native C++ version.
  • Although snappy-java uses JNI, it can be used safely with multiple class loaders (e.g. Tomcat, etc.).
• Compression/decompression of Java primitive arrays (float[], double[], int[], short[], long[], etc.)
  • To improve the compression ratios of these arrays, you can use a fast data-rearrangement implementation (BitShuffle) before compression
• Portable across various operating systems; Snappy-java contains native libraries built for Window/Mac/Linux, etc. snappy-java loads one of these libraries according to your machine environment (It looks system properties, os.name and os.arch).
• Simple usage. Add the snappy-java-(version).jar file to your classpath. Then call compression/decompression methods in org.xerial.snappy.Snappy.
• Framing-format support (Since 1.1.0 version)
• OSGi support
• Apache License Version 2.0. Free for both commercial and non-commercial use.

• Snappy's main target is very high-speed compression/decompression with reasonable compression size. So the compression ratio of snappy-java is modest and about the same as LZF (ranging 20%-100% according to the dataset).

• Here are some benchmark results, comparing snappy-java and the other compressors LZO-java/LZF/QuickLZ/Gzip/Bzip2. Thanks Tatu Saloranta @cotowncoder for providing the benchmark suite.

• The benchmark result indicates snappy-java is the fastest compressor/decompressor in Java: https://ning.github.io/jvm-compressor-benchmark/results/canterbury-roundtrip-2011-07-28/index.html

• The decompression speed is twice as fast as the others: https://ning.github.io/jvm-compressor-benchmark/results/canterbury-uncompress-2011-07-28/index.html

• Release Notes

The current stable version is available from here:

• Release version: https://repo1.maven.org/maven2/org/xerial/snappy/snappy-java/
• Snapshot version (the latest beta version): https://oss.sonatype.org/content/repositories/snapshots/org/xerial/snappy/snappy-java/

Snappy-java is available from Maven's central repository. Add the following dependency to your pom.xml:

<dependency>
  <groupId>org.xerial.snappy</groupId>
  <artifactId>snappy-java</artifactId>
  <version>(version)</version>
  <type>jar</type>
  <scope>compile</scope>
</dependency>

implementation("org.xerial.snappy:snappy-java:(version)")

libraryDependencies += "org.xerial.snappy" % "snappy-java" % "(version)"

First, import org.xerial.snapy.Snappy in your Java code:

import org.xerial.snappy.Snappy;

Then use Snappy.compress(byte[]) and Snappy.uncompress(byte[]):

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);

In addition, high-level methods (Snappy.compress(String), Snappy.compress(float[] ..) etc. ) and low-level ones (e.g. Snappy.rawCompress(.. ), Snappy.rawUncompress(..), etc.), which minimize memory copies, can be used.

Stream-based compressor/decompressor SnappyOutputStream/SnappyInputStream are also available for reading/writing large data sets. SnappyFramedOutputStream/SnappyFramedInputStream can be used for the framing format.

• See also Javadoc API

The original Snappy format definition did not define a file format. It later added a "framing" format to define a file format, but by this point major software was already using an industry standard instead -- represented in this library by the SnappyOutputStream and SnappyInputStream methods.

For interoperability with other libraries, check that compatible formats are used. Note that not all libraries support all variants.

• SnappyOutputStream and SnappyInputStream use [magic header:16 bytes]([block size:int32][compressed data:byte array])* format. You can read the result of Snappy.compress with SnappyInputStream, but you cannot read the compressed data generated by SnappyOutputStream with Snappy.uncompress.
• SnappyHadoopCompatibleOutputStream does not emit a file header but write out the current block size as a preemble to each block

BitShuffle is an algorithm that reorders data bits (shuffle) for efficient compression (e.g., a sequence of integers, float values, etc.). To use BitShuffle routines, import org.xerial.snapy.BitShuffle:

import org.xerial.snappy.BitShuffle;

int[] data = new int[] {1, 3, 34, 43, 34};
byte[] shuffledByteArray = BitShuffle.shuffle(data);
byte[] compressed = Snappy.compress(shuffledByteArray);
byte[] uncompressed = Snappy.uncompress(compressed);
int[] result = BitShuffle.unshuffleIntArray(uncompress);

System.out.println(result);

Shuffling and unshuffling of primitive arrays (e.g., short[], long[], float[], double[], etc.) are supported. See Javadoc for the details.

If you have snappy-java-(VERSION).jar in the current directory, use -classpath option as follows:

$ javac -classpath ".;snappy-java-(VERSION).jar" Sample.java  # in Windows
or
$ javac -classpath ".:snappy-java-(VERSION).jar" Sample.java  # in Mac or Linux

Post bug reports or feature request to the Issue Tracker: https://github.com/xerial/snappy-java/issues

Public discussion forum is here: Xerial Public Discussion Group

snappy-java uses sbt (simple build tool for Scala) as a build tool. Here is a simple usage

$ ./sbt            # enter sbt console
> ~test            # run tests upon source code change
> ~testOnly        # run tests that matches a given name pattern  
> publishM2        # publish jar to $HOME/.m2/repository
> package          # create jar file
> findbugs         # Produce findbugs report in target/findbugs
> jacoco:cover     # Report the code coverage of tests to target/jacoco folder    

If you need to see detailed debug messages, launch sbt with -Dloglevel=debug option:

$ ./sbt -Dloglevel=debug

For the details of sbt usage, see my blog post: Building Java Projects with sbt

See the build instruction. Building from the source code is an option when your OS platform and CPU architecture is not supported. To build snappy-java, you need Git, JDK (1.6 or higher), g++ compiler (mingw in Windows) etc.

$ git clone https://github.com/xerial/snappy-java.git
$ cd snappy-java
$ make

When building on Solaris, use gmake:

$ gmake

A file target/snappy-java-$(version).jar is the product additionally containing the native library built for your platform.

GitHub action [https://github.com/xerial/snappy-java/blob/master/.github/workflows/release.yml] will publish a new relase to Maven Central (Sonatype) when a new tag vX.Y.Z is pushed.

Simply put the snappy-java's jar to WEB-INF/lib folder of your web application. Usual JNI-library specific problem no longer exists since snappy-java version 1.0.3 or higher can be loaded by multiple class loaders.

Prepare org-xerial-snappy.properties file (under the root path of your library) in Java's property file format. Here is a list of the available properties:

• org.xerial.snappy.lib.path (directory containing a snappyjava's native library)
• org.xerial.snappy.lib.name (library file name)
• org.xerial.snappy.tempdir (temporary directory to extract a native library bundled in snappy-java)
• org.xerial.snappy.use.systemlib (if this value is true, use system installed libsnappyjava.so looking the path specified by java.library.path)

Snappy-java is developed by Taro L. Saito. Twitter @taroleo