Snzip is one of command line tools using snappy. This supports several file formats; framing-format, old framing-format, hadoop-snappy format, raw format and obsolete three formats used by snzip, snappy-java and snappy-in-java before official framing-format was defined. The default format is framing-format.
The default format was changed to framing-format in 1.0.0.
Set --with-default-format=snzip
as a configure option to use obsolete snzip
format as the default format as before.
Download snzip-1.0.5.tar.gz from https://github.com/kubo/snzip/releases, uncompress and untar it, and run configure.
tar xvfz snzip-1.0.5.tar.gz
cd snzip-1.0.5
./configure
make
make install
If you didn't install snappy under /usr
or /usr/local
, you need to specify
the location by --with-snappy
as follows.
# install snzip
tar xvfz snzip-1.0.5.tar.gz
cd snzip-1.0.5
./configure --with-snappy=/xxx/yyy/
make
make install
When both dynamic and static snappy libraries are available, the former
is used by default. The compiled snzip
depends on libsnappy.so
.
When --with-static-snappy
is passed as a configure option, the latter
is used. The compiled snzip
includes snappy library.
Note: --with-static-snappy
isn't available on some platforms.
You can use --with-default-format
to change the default compression format.
./configure --with-default-format=snzip
We don't provide rpm packages. You need to download snzip-1.0.5.tar.gz from https://github.com/kubo/snzip/releases, create a rpm package as follows and install it.
# The rpm package will be created under $HOME/rpmbuild/RPMS.
rpmbuild -tb snzip-1.0.5.tar.gz
To use source code in the github repository.
git clone git://github.com/kubo/snzip.git
cd snzip
./autogen.sh
./configure
make
make install
Download snzip-1.0.5-win32.zip
or snzip-1.0.5-win64.zip
from
https://github.com/kubo/snzip/releases and copy snzip.exe
and snunzip.exe
to a directory in the PATH environment variable.
snzip file.tar
Compressed file name is file.tar.sz
and the original file is deleted.
The file attributes such as timestamp, mode and permissions are not changed
as possible as it can.
The compressed file's format is framing-format. You need to add an option -t snappy-java
or
-t snappy-in-java
to use other formats.
snzip -t snappy-java file.tar
or
snzip -t snappy-in-java file.tar
snzip -c file.tar > file.tar.sz
or
cat file.tar | snzip > file.tar.sz
You need to add an option -t [format-name]
to use formats except framing-format.
tar cf - files-to-be-archived | snzip > archive.tar.sz
snzip -d file.tar.sz
or
snunzip file.tar.sz
Uncompressed file name is file.tar
and the original file is deleted.
The file attributes such as timestamp, mode and permissions are not changed
as possible as it can.
If the program name includes un
such as snunzip
, it acts as -d
is set.
The file format is automatically determined from the file header. However it doesn't work for some file formats such as raw and Apple iWork .iwa.
snzip -dc file.tar.sz > file.tar
snunzip -c file.tar.sz > file.tar
snzcat file.tar.sz > file.tar
cat file.tar.sz | snzcat > file.tar
If the program name includes cat
such as snzcat, it acts as -dc
is set.
snzip -dc archive.tar.sz | tar xf -
Raw format is native format of snappy. Unlike other formats, there are a few limitations: (1) The total data length before compression must be known on compression. (2) Automatic file format detection doesn't work on uncompression. (3) The raw format support is enabled only when snzip is compiled for snappy 1.1.3 or upper.
snzip -t raw file.tar
or
snzip -t raw < file.tar > file.tar.raw
In these examples, snzip uses a file descriptor, which directly opens
the file.tar
file, and gets the file length to be compressed.
However the following command doesn't work.
cat file.tar | snzip -t raw > file.tar.raw
It uses a pipe. snzip cannot get the total length before compression.
The total length must be specified by the -s
option in this case.
cat file.tar | snzip -t raw -s "size of file.tar" > file.tar.raw
snzip -t raw -d file.tar.sz
or
snunzip -t raw file.tar.sz
You need to set the -t raw
option to tell snzip the format of the
file to be uncompressed.
Hadoop-snappy format is one of the compression formats used in Hadoop. It uses its own framing format as follows:
- A compressed file consists of one or more blocks.
- A block consists of uncompressed length (big endian 4 byte integer) and one or more subblocks.
- A subblock consists of compressed length (big endian 4 byte integer) and raw compressed data.
snzip -t hadoop-snappy file_name
The default block size used by snzip
for hadoop-snappy format is 256k.
It is same with the default value of the io.compression.codec.snappy.buffersize
parameter. If the block size used by snzip
is larger than the parameter,
you would get an InternalError Could not decompress data. Buffer length is too small
while hadoop is reading a file compressed by snzip. You need to change the block
size by the -b
option as follows if you get the error.
# if io.compression.codec.snappy.buffersize is 32768
snzip -t hadoop-snappy -b 32768 file_name_to_be_compressed
snzip -d compressed_file.snappy
The file format is guessed by the first 8 bytes of the file.
Apple iWork .iwa format is a file format used by Apple iWork. The format was demystified here. Basically the .iwa format consists of a Protobuf stream compressed by Snappy.
Snzip uncompresses .iwa files to Protbuf streams and compresses Protobuf streams
to .iwa files. You need to set -t iwa
on compression and uncompression to
specify the file format.
Note: This is obsolete format. The default format was changed to framing-format.
The first three bytes are magic characters 'SNZ'.
The fourth byte is the file format version. It is 0x01.
The fifth byte is the order of the block size. The input data is divided into fixed-length blocks and each block is compressed by snappy. When it is 16 (default value), the block size is 16th power of 2; 64 kilobytes.
The rest is pairs of a compressed data length and a compressed data block
The compressed data length is encoded as snappy::Varint::Encode32()
does.
If the length is zero, it is the end of data.
Though the rest after the end of data is ignored for now, they may be continuously read as a next compressed file as gzip does.
Note that the uncompressed length of each compressed data block must be less than or equal to the block size specified by the fifth byte.
2-clause BSD-style license.