Apache Commons Compress™
The Apache Commons Compress library defines an API for
working with ar, cpio, Unix dump, tar, zip, gzip, XZ,
Pack200, bzip2, 7z, arj, lzma, snappy, DEFLATE, lz4,
Brotli, Zstandard, DEFLATE64 and Z files.
The code in this component has many origins:
- The bzip2, tar and zip support came from Avalon's
Excalibur, but originally from Ant, as far as life in
Apache goes. The tar package is originally Tim Endres'
public domain package. The bzip2 package is based on
the work done by Keiron Liddle as well as Julian Seward's
libbzip2.
It has migrated
via:
Ant -> Avalon-Excalibur -> Commons-IO
-> Commons-Compress.
- The cpio package has been contributed by Michael Kuss
and
the jRPM
project.
Status
The current release is 1.20 and requires Java 7.
Below we highlight some new features, for a full list
of changes see the Changes
Report.
What's new in 1.20?
- SevenZFile now supports random
access.
- The zip package now supports split archives.
- The tar package now supports reading sparse entries.
What's new in 1.19?
- ParallelScatterZipCreator now writes
entries in the same order they have been added to the
archive.
- ZipArchiveInputStream and
ZipFile are more forgiving when parsing
extra fields by default now.
- TarArchiveInputStream has a new lenient
mode that may allow it to read certain broken
archives.
What's new in 1.18?
- The CPIO package now properly handles file names
using a mult-byte encoding.
- ZipArchiveInputStream can now deal with APK files
containing an APK signing block.
- It is no possible to specifiy various parameters
for Zstd output.
What's new in 1.17?
- A new InputStreamStatistics interface
is implemented by many streams and may be used to
provide feedback or detect abnormally high compression
ratios that may indicate a ZIP bomb during
decompression.
What's new in 1.16.1?
- Fixed the OSGi manifest that was broken in Compress 1.16.
What's new in 1.16?
- Support for Zstandard compression.
- Read-only support for DEFLATE64 compression as
stand-alone CompressorInputStream and as method used in
ZIP and 7z archives.
What's new in 1.15?
- Added Automatic-Module-Name so the module
name will be org.apache.commons.compress when the jar is used
as an automatic module in Java9+.
What's new in 1.14?
- Added support for writing the Snappy format
- Added support for the LZ4 compression format
- Added read-only support for Brotli decompression by
using the Google Brotli
decoder.
What's new in 1.13?
- The 7z package as well as
ZipArchiveOutputStream and
ZipFile can now use
SeekableByteChannel when random acces is
needed. This allows archives to be read from inputs and
written to outputs that are seekable but are not
represented by Files.
- It is now possible to add Compressor- and
ArchiverStream implementations using the JDK's
ServiceLoader mechanism. Please see
Extending Commons Compress.
- Added support for writing the legacy LZMA format as
compressor stream and inside 7z archives - this requires
XZ for Java 1.6.
What's new in 1.12?
- Added support for the Snappy dialect used in iWork archives.
- SevenZFile throws an
IllegalStateException for empty entries.
- BZip2CompressorOutputStream no longer tries to
finish the output stream in finalize. This is a breaking
change for code that relied on the finalizer.
- Various fixes and improvements for tar, cpio
and zip.
What's new in 1.11?
- Added read-only support for BZIP2 compression used
inside of ZIP archives.
- Speed improvements in SevenZFile
- Various fixes and improvements for tar, ar, snappy
and zip.
What's new in 1.10?
- The old
org.apache.commons.compress.compressors.z._internal_
now is
org.apache.commons.compress.compressors.lzw
and the code is now an official part of Commons
Compress' API.
- Added support for parallel ZIP compression.
- Added support for raw transfer of entries from one ZIP file to another without uncompress/compress.
- Performance improvements for creating ZIP files with lots of small entries.
- Added auto-detection for LZMA.
What's new in 1.9?
- support for raw DEFLATE streams
What's new in 1.8.1?
Compress 1.8.1 is a bug fix release with fixes for the
tar, ar and snappy formats as well as the
IOUtils class. In addition
CompressorStreamFactory can now autodetect
the .Z compress format.
What's new in 1.8?
- Bug fixes to the tar, zip and 7z packages
- Access to metadata when reading gzip streams
- Finer grained control over the methods used when
creating 7z archives and support for some the most
important filter methods used in 7z archives.
What's new in 1.7?
- Read-only support for the Snappy compression.
- Read-only support for the traditional Unix compress
format used for .Z files.
What's new in 1.6?
- Support for the 7z format.
- Read-only support for uncompressed ARJ
archives.
- Read-only support for the "stand-alone" LZMA
format.
Documentation
The compress component is split into compressors and
archivers. While compressors
(un)compress streams that usually store a single
entry, archivers deal with archives that contain
structured content represented
by ArchiveEntry instances which in turn
usually correspond to single files or directories.
Currently the bzip2, Pack200, XZ, gzip, lzma, brotli,
Zstandard and Z formats are
supported as compressors where gzip support is mostly provided by
the java.util.zip package and Pack200 support
by the java.util.jar package of the Java
class library. XZ and lzma support is provided by the public
domain XZ for
Java library. Brotli support is provided by the MIT
licensed Google
Brotli decoder. Zstandard support is provided by the BSD
licensed Zstd-jni.
As of Commons Compress 1.20 support for the DEFLATE64, Z and Brotli
formats is read-only.
The ar, arj, cpio, dump, tar, 7z and zip formats are supported as
archivers where the zip
implementation provides capabilities that go beyond the
features found in java.util.zip. As of Commons Compress
1.20 support for the dump and arj formats is
read-only - 7z can read most compressed and encrypted
archives but only write unencrypted ones. LZMA(2) support
in 7z requires XZ for
Java as well.
The compress component provides abstract base classes for
compressors and archivers together with factories that can
be used to choose implementations by algorithm name. In
the case of input streams the factories can also be used
to guess the format and provide the matching
implementation.
|
