Compression & Archiving

Getting Started

On Linux, archiving and compression are two separate steps: tar bundles many files into one archive, while compressors like gzip, xz, and zstd shrink a single data stream. That is why the typical format is a compressed tar archive such as .tar.gz or .tar.zst. The zip and 7z formats do both jobs in one tool.

Tar Archives

The first option tells tar what to do: c creates, x extracts, t lists. Add f for the archive file name and v to print each file as it is processed.

Option	Description
c	Create an archive
x	Extract an archive
t	List archive contents
f	Archive file name (must come right before it)
v	Verbose: show files being processed
z	gzip compression (.tar.gz)
j	bzip2 compression (.tar.bz2)
J	xz compression (.tar.xz)
--zstd	zstd compression (.tar.zst)
-C dir	Change to directory before extracting
--exclude=pattern	Skip files matching a pattern

$ tar czf archive.tar.gz [files]

$ tar cJf archive.tar.xz [files]

$ tar --zstd -cf archive.tar.zst [files]

$ tar czf backup.tar.gz --exclude="*.log" [directory]

When extracting, modern GNU tar detects the compression automatically, so plain xf works for every format.

$ tar xf archive.tar.gz

$ tar xf archive.tar.zst -C [directory]

List the contents before extracting an archive from an untrusted source, and extract single files by naming them.

$ tar tf archive.tar.gz

$ tar xf archive.tar.gz path/inside/archive.txt

A well-behaved archive contains a single top-level directory. If tar tf shows loose files instead, extract into a fresh directory with -C to avoid littering your current one.

Choosing a Compressor

All compressors trade speed against ratio. As a rule of thumb: zstd is the modern default, gzip is the universal lowest common denominator, xz squeezes hardest when time does not matter.

Tool	Description
gzip	Fast, supported everywhere, moderate ratio
bzip2	Better ratio than gzip but slow, mostly legacy
xz	Highest ratio, slow to compress, common for releases
zstd	Very fast with ratios near xz, best general choice
lz4	Extremely fast, lower ratio, good for live pipes

Compressing Single Files

The classic compressors share one interface: they compress a file in place, replace it, and append their extension. -k keeps the original, -d decompresses, and -1 to -9 trade speed for ratio.

$ gzip [file]

$ gzip -k -9 [file]

$ bzip2 [file]

$ xz [file]

$ zstd [file]

$ zstd -19 [file]

Each has a matching decompressor.

$ gunzip [file].gz

$ bunzip2 [file].bz2

$ unxz [file].xz

$ unzstd [file].zst

These tools compress single files only. To compress a directory, tar it first or use zip.On multi-core machines, xz -T0 and zstd -T0 use all cores, and pigz is a parallel drop-in for gzip.

$ xz -T0 [file]

$ pigz [file]

Working with Compressed Files

Read, search, and compare compressed text files without unpacking them.

$ zcat [file].gz

$ zless [file].gz

$ zgrep "pattern" [file].gz

$ zdiff [file1].gz [file2].gz

The other formats have their own cat tools.

$ bzcat [file].bz2

$ xzcat [file].xz

$ zstdcat [file].zst

Zip Archives

zip is the standard interchange format with Windows and macOS. Use -r to include directories recursively and -e to encrypt with a password.

$ zip -r archive.zip [directory]

$ zip -e -r secret.zip [directory]

Extract with unzip, into a specific directory with -d, or just inspect the contents first.

$ unzip archive.zip

$ unzip archive.zip -d [directory]

$ unzip -l archive.zip

$ zipinfo archive.zip

Zip does not preserve full Unix ownership and permissions. For system backups, stick to tar.

7-Zip Archives

7z offers very high compression ratios and strong AES-256 encryption. a adds to an archive, x extracts with full paths, l lists.

$ 7z a archive.7z [files]

$ 7z a -p archive.7z [files]

$ 7z x archive.7z

$ 7z l archive.7z

Use x to extract, not e: the e command flattens all files into the current directory, discarding their paths.

Rar Archives

Rar is a proprietary format; unrar extracts it.

$ unrar x archive.rar

$ unrar l archive.rar

Universal Extractors

Tools that detect the format for you, handy when you do not want to remember per-format flags.

$ atool -x archive.tar.gz

$ unp archive.rar

$ dtrx archive.zip

$ ouch decompress archive.tar.zst

cpio & ar

cpio reads file lists from stdin; it is the format behind initramfs images and RPM packages.

$ find . | cpio -o > archive.cpio

$ cpio -id < archive.cpio

ar creates the archives behind static libraries (.a) and Debian packages (.deb).

$ ar rcs libfoo.a [object-files]

$ ar t package.deb