tar / bz2 сжимает файл, удаляя несжатый оригинал

tar can't do that, but you can achieve what you want with:

find dir1 -depth -print0 | xargs -0 tar --create --no-recursion --remove-file --file - | bzip2 > dir1.tar.bz2

where:

• find dir1 -depth -print0

  lists all files and directories in dir1, listing the directory contents before the directory itself (-depth). The use of -print0 (and -0 in xargs below) is the key to supporting directory and file names with embedded spaces.

• xargs -0 tar --create --no-recursion --remove-file --file -

  creates a tar archive and adds every file or directory to it. The tar archive is sent to standard output with option --file -.

• bzip2 > dir1.tar.bz2

  compresses the tar archive from standard input to a file called dir1.tar.bz2.

The amount of free disk space needed is the size of the largest compressed file in dir1 because tar, when processing a file, waits until archiving is complete before deleting it. Since tar is piped to bzip2, for a short moment, before tar removes it, every file resides in two places: uncompressed in the filesystem and compressed inside dir1.tar.bz2.

I was curious to see how disk space was used so I made this experiment on my Ubuntu VM:

1. Create a 1 GB filesystem:

   $ dd if=/dev/zero of=/tmp/1gb bs=1M count=1024 $ losetup /dev/loop0 /tmp/1gb $ mkfs.ext3 /dev/loop0 $ sudo mount /dev/loop0 /tmp/mnt $ df -h Filesystem Size Used Avail Use% Mounted on /dev/loop0 1008M 34M 924M 4% /tmp/mnt 

2. Fill the filesystem with 900 1 megabyte-files:

   $ chown jaume /tmp/mnt $ mkdir /tmp/mnt/dir1 $ for (( i=0; i<900; i++ )); do dd if=/dev/urandom of=/tmp/mnt/dir1/file$i bs=1M count=1; done $ chown -R jaume /tmp/mnt $ df -h Filesystem Size Used Avail Use% Mounted on /dev/loop0 1008M 937M 20M 98% /tmp/mnt 

   The filesystem is now 98% full.

3. Make a copy of dir1 for later verification:

   $ cp -a /tmp/mnt/dir1 /tmp/dir1-check 

4. Compress dir1:

   $ ls /tmp/mnt dir1 lost+found $ find /tmp/mnt/dir1 -depth -print0 | xargs -0 tar --create --no-recursion --remove-file --file - | bzip2 > /tmp/mnt/dir1.tar.bz2 $ 

   Note that the commands ran without any 'no space left on device' errors.

   dir1 was removed, only dir1.tar.bz2 exists:

   $ ls /tmp/mnt dir1.tar.bz2 lost+found 

5. Expand dir1.tar.bz2 and compare to /tmp/dir1-check:

   $ tar --extract --file dir1.tar.bz2 --bzip2 --directory /tmp $ diff -s /tmp/dir1 /tmp/dir1-check (...) Files /tmp/dir1/file97 and /tmp/dir1-check/file97 are identical Files /tmp/dir1/file98 and /tmp/dir1-check/file98 are identical Files /tmp/dir1/file99 and /tmp/dir1-check/file99 are identical $ 

   Copy of dir1 and uncompressed dir1.tar.bz2 are identical!

This can be generalized in a script:

1. Create a file called tarrm (or any other name of your liking) with these contents:

   #!/bin/bash # This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. # This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. # You should have received a copy of the GNU General Public License along with this program. If not, see <http://www.gnu.org/licenses/>. # dir is first argument dir="$1" # check dir exists if [ ! -d "$dir" ]; then echo "$(basename $0): error: '$dir' doesn't exist" 1>&2 exit 1 fi # check if tar file exists if [ -f "$.tar" -o -f "$.tar.bz2" ]; then echo "$(basename $0): error: '$dir.tar' or '$.tar.bz2' already exist" 1>&2 exit 1 fi # --keep is second argument if [ "X$2" == "X--keep" ]; then # keep mode removefile="" echo " Tarring '$dir'" else removefile="--remove-file" echo " Tarring and **deleting** '$dir'" fi # normalize directory name (for example, /home/jaume//// is a legal directory name, but will break $.tar.bz2 - it needs to be converted to /home/jaume) dir=$(dirname "$dir")/$(basename "$dir") # create compressed tar archive and delete files after adding them to it find "$dir" -depth -print0 | xargs -0 tar --create --no-recursion $removefile --file - | bzip2 > "$.tar.bz2" # return status of last executed command if [ $? -ne 0 ]; then echo "$(basename $0): error while creating '$.tar.bz2'" 1>&2 fi 

2. Make it executable:

   chmod a+x tarrm

The script does some basic error checking: dir1 must exist, dir1.tar.bz2 and dir1.tar shouldn't exist and has a keep mode. It also supports directory and file names with embedded spaces.

I've tested the script but can't guarantee it is flawless, so first use it in keep mode:

./tarrm dir1 --keep

This invocation will add dir1 to dir1.tar.bz2 but won't delete the directory.

When you trust the script use it like this:

./tarrm dir1

The script will inform you that dir1 will be deleted in the process of tarring it:

Tarring and **deleting** 'dir1'

For example:

$ ls -lF total 4 drwxrwxr-x 3 jaume jaume 4096 2013-10-11 11:00 dir 1/ $ find "dir 1" dir 1 dir 1/subdir 1 dir 1/subdir 1/file 1 dir 1/file 1 $ /tmp/tarrm dir\ 1/ Tarring and **deleting** 'dir 1/' $ echo $? 0 $ ls -lF total 4 -rw-rw-r-- 1 jaume jaume 181 2013-10-11 11:00 dir 1.tar.bz2 $ tar --list --file dir\ 1.tar.bz2 dir 1/subdir 1/file 1 dir 1/subdir 1/ dir 1/file 1 dir 1/