Managing Files and Directories

This chapter presents commands to do things that are typically handled by a file manager (explorer) in GUI. For example, viewing contents of a directory, navigating to other directories, cut/copy/paste files, renaming and so on. Some of the commands used for these purposes are provided by the shell itself.

As a good practice, make it a habit to go through the documentation of the commands you encounter. Getting used to looking up documentation from the command line will come in handy whenever you are stuck. You can also learn and experiment with options you haven't used yet.

info The example_files directory has the scripts used in this chapter.

Builtin and External commands

From bash manual: What is a shell?

Shells also provide a small set of built-in commands (builtins) implementing functionality impossible or inconvenient to obtain via separate utilities. For example, cd, break, continue, and exec cannot be implemented outside of the shell because they directly manipulate the shell itself. The history, getopts, kill, or pwd builtins, among others, could be implemented in separate utilities, but they are more convenient to use as builtin commands.

Most of the commands needed for everyday use are external commands, i.e. not part of the shell. Some builtins, pwd for example, might also be available as external command on your system (and these might have differences in features too). In such cases the builtin version will be executed by default, which you can override by using the path of the external version.

You can use the type command to distinguish between builtin and external commands. The type command is a shell builtin, and provides other features too (which will be discussed later). You can use the -a option to get all details about the given command.

$ type -a cd
cd is a shell builtin

$ type -a ls
ls is /bin/ls

$ type -a pwd
pwd is a shell builtin
pwd is /bin/pwd

info To look up documentation, use the help command for builtins and man for external commands (or info for complete documentation, where applicable). Use help help and man man for their own documentation.

info Typing just help will give the list of builtins, along with the command's syntax.

pwd

pwd is a shell builtin command to get the current working directory. This helps to orient yourself where you are in the filesystem. The absolute path printed is often handy to copy-paste elsewhere, in a script for example. Some users prefer their terminal emulators and/or shell prompt to always display the current working directory.

$ pwd
/home/learnbyexample

cd

cd is another shell builtin. This helps to change the current working directory. Here's an example of changing the current working directory using an absolute path:

$ pwd
/home/learnbyexample

# specifying / at end of the path is optional
$ cd /etc
$ pwd
/etc

You can use - as an argument to go back to the previous working directory. Continuing from the previous example:

$ cd -
/home/learnbyexample

info Most commands will treat strings starting with - as a command option. You can use -- to tell commands that all the following arguments should not be treated as options even if they start with -. For example, if you have a folder named -oops in the current working directory, you can use cd -- -oops to switch to that directory.

Relative paths are well, relative to the current working directory:

  • . refers to the current directory
  • .. refers to the directory one hierarchy above
  • ../.. refers to the directory two hierarchies above and so on
  • cd ./- will help you to switch to a directory named - in the current location
    • you cannot use cd - since that'll take you to the previous working directory
$ pwd
/home/learnbyexample

# go one hierarchy above
$ cd ..
$ pwd
/home

# change to 'learnbyexample' present in the current directory
# './' is optional in this case
$ cd ./learnbyexample
$ pwd
/home/learnbyexample

# go two hierarchies above
$ cd ../..
$ pwd
/

You can switch to the home directory using cd or cd ~ or cd ~/ from anywhere in the filesystem. This is determined by the value stored in the HOME shell variable.

$ pwd
/
$ echo "$HOME"
/home/learnbyexample

$ cd
$ pwd
/home/learnbyexample

Further Reading

clear

You can use this command to clear the terminal screen. By default, the clear command will move the prompt to the top of the terminal as well as try to remove the contents of the scrollback buffer. You can use the -x option if you want to retain the scrollback buffer contents.

info The Ctrl+l shortcut will also move the prompt line to the top of the terminal. It will retain any text you've typed on the prompt line and scrollback buffer contents won't be cleared.

ls

When you use a file explorer GUI application, you'll automatically see the directory contents. And such GUI apps typically have features to show file size, differentiate between files and folders and so on. ls is the equivalent command line tool with a plethora of options and functionality related to viewing the contents of directories.

info As mentioned earlier, the example_files directory has the scripts used in this chapter. You can source the ls.sh script to follow along the examples shown in this section.

# first, cd into the 'scripts' directory
$ cd cli-computing/example_files/scripts

$ ls
cp.sh  file.sh  globs.sh  ls.sh  rm.sh    tar.sh
du.sh  find.sh  grep.sh   mv.sh  stat.sh  touch.sh

# 'ls.sh' script will create a directory named 'ls_examples'
# and automatically change to that directory as well
$ source ls.sh
$ pwd
/home/learnbyexample/cli-computing/example_files/scripts/ls_examples

Here are some examples:

# by default, the current directory contents are displayed
$ ls
backups     hello_world.py  ip.txt    report.log  todos
errors.log  hi              projects  scripts

# example with a single path argument
$ ls /sys
block  class  devices   fs          kernel  power
bus    dev    firmware  hypervisor  module

# multiple paths example
# directory listing will be preceded by its name
$ ls projects backups ip.txt
ip.txt

backups:
bookmarks.html  dot_files

projects:
calculator  tictactoe

You can use the -1 option (1 as in numeric one, not the alphabet l which does something else) to list the contents in a single column:

$ ls -1 backups
bookmarks.html
dot_files

The -F option appends a character to each filename indicating the file type (if it is other than a regular file):

  • / directory
  • * executable file
  • @ symbolic link
  • | FIFO
  • = socket
  • > door
$ ls -F
backups/    hello_world.py*  ip.txt     report.log  todos/
errors.log  hi*              projects/  scripts@

If you just need to distinguish between files and directories, you can use the -p option:

$ ls -p
backups/    hello_world.py  ip.txt     report.log  todos/
errors.log  hi              projects/  scripts

You can also use the --color option to visually distinguish file types:

example with --color option

The -l option displays the contents using a long listing format. You'll get details like file permissions, ownership, size, timestamp and so on. The first character of the first column distinguishes file types as d for directories, - for regular files, l for symbolic links, etc. Under each directory listing, the first line will display the total size of the entries (in terms of KB).

$ ls -l hi
-rwxrwxr-x 1 learnbyexample learnbyexample 21 Dec  5  2019 hi

# you can add -G option to avoid the group column
$ ls -lG
total 7516
drwxrwxr-x 3 learnbyexample    4096 Feb  4 09:23 backups
-rw-rw-r-- 1 learnbyexample   12345 Jan  1 03:30 errors.log
-rwxrwxr-x 1 learnbyexample      42 Feb 29  2020 hello_world.py
-rwxrwxr-x 1 learnbyexample      21 Dec  5  2019 hi
-rw-rw-r-- 1 learnbyexample      10 Jul 21  2017 ip.txt
drwxrwxr-x 4 learnbyexample    4096 Mar  5 11:21 projects
-rw-rw-r-- 1 learnbyexample 7654321 Jan  1 01:01 report.log
lrwxrwxrwx 1 learnbyexample      13 May  7 15:17 scripts -> ../../scripts
drwxrwxr-x 2 learnbyexample    4096 Apr  6 13:19 todos

Use the -h option to show file sizes in human readable format (default is byte count).

$ ls -lG report.log
-rw-rw-r-- 1 learnbyexample 7654321 Jan  1 01:01 report.log

$ ls -lhG report.log
-rw-rw-r-- 1 learnbyexample 7.3M Jan  1 01:01 report.log

You can use the -s option instead of long listing if you only need allocated file sizes and names:

$ ls -1sh errors.log report.log
 16K errors.log
7.4M report.log

There are several options for changing the order of output listing:

  • -t sorts by timestamp
  • -S sorts by file size (not suitable for directories)
  • -v version sorting (suitable for filenames with numbers in them)
  • -X sorts by file extension (i.e. characters after the last . in the filename)
  • -r reverse the listing order
$ ls -lGhtr
total 7.4M
-rw-rw-r-- 1 learnbyexample   10 Jul 21  2017 ip.txt
-rwxrwxr-x 1 learnbyexample   21 Dec  5  2019 hi
-rwxrwxr-x 1 learnbyexample   42 Feb 29  2020 hello_world.py
-rw-rw-r-- 1 learnbyexample 7.3M Jan  1 01:01 report.log
-rw-rw-r-- 1 learnbyexample  13K Jan  1 03:30 errors.log
drwxrwxr-x 3 learnbyexample 4.0K Feb  4 09:23 backups
drwxrwxr-x 4 learnbyexample 4.0K Mar  5 11:21 projects
drwxrwxr-x 2 learnbyexample 4.0K Apr  6 13:19 todos
lrwxrwxrwx 1 learnbyexample   13 May  7 15:17 scripts -> ../../scripts

Filenames starting with . are considered as hidden files and these are NOT shown by default. You can use the -a option to view them. The -A option is similar, but doesn't show the special . and .. entries.

# . and .. point to the current and parent directories
$ ls -aF backups/dot_files/
./  ../  .bashrc  .inputrc  .vimrc

# -A will exclude . and ..
$ ls -A backups/dot_files/
.bashrc  .inputrc  .vimrc

The -R option recursively lists sub-directories as well:

$ ls -ARF
.:
backups/    hello_world.py*  .hidden  projects/   scripts@
errors.log  hi*              ip.txt   report.log  todos/

./backups:
bookmarks.html  dot_files/

./backups/dot_files:
.bashrc  .inputrc  .vimrc

./projects:
calculator/  tictactoe/

./projects/calculator:
calc.sh

./projects/tictactoe:
game.py

./todos:
books.txt  outing.txt

Often you'd want to list only specific files or directories based on some criteria, file extension for example. The shell provides a matching technique called globs or wildcards. Some simple examples are shown below (globs will be discussed in more detail later).

* is a placeholder for zero or more characters:

# '*.py *.log' will give filenames ending with '.py' or '.log'
$ echo *.py *.log
hello_world.py errors.log report.log

# glob expansion can be prevented by using quotes
$ echo '*.py' *.log
*.py errors.log report.log

# long list only files ending with '.log'
$ ls -lG *.log
-rw-rw-r-- 1 learnbyexample   12345 Jan  1 03:30 errors.log
-rw-rw-r-- 1 learnbyexample 7654321 Jan  1 01:01 report.log

[] helps you specify a set of characters to be matched once. For example, [ad] matches a or d once. [c-i] matches a range of characters from c to i.

# entries starting with 'c' to 'i'
$ echo [c-i]*
errors.log hello_world.py hi ip.txt

$ ls -1sh [c-i]*
 16K errors.log
4.0K hello_world.py
4.0K hi
4.0K ip.txt

info info As shown in the above examples, globs are expanded by the shell. Beginners often associate globs as something specific to the ls command, which is why I've delibrately used echo as well in the above examples.

You can use the -d option to not show directory contents:

$ echo b*
backups
# since backups is a directory, ls will show its contents
$ ls b*
bookmarks.html  dot_files
# -d will show the directory entry instead of its contents
$ ls -d b*
backups

# a handy way to get only the directory entries
$ echo */
backups/ projects/ scripts/ todos/
$ ls -1d */
backups/
projects/
scripts/
todos/

Further Reading

tree

The tree command displays the contents of a directory recursively, in a hierarchical manner. Here's a screenshot of using tree -a from the ls_examples sample directory seen in the previous section. The -a option is used to show the hidden files as well.

example for tree command

info You might have to install this command. sudo apt install tree can be used to get this command on Debian like distributions.

mkdir

The mkdir command helps you to create new directories. You can pass one or more paths along with the name of the directories you want to create. Quote the names if it can contain shell special characters like space, * and so on.

info Create a practice directory for this section:

$ mkdir practice_mkdir
$ cd practice_mkdir

Here's an example of creating multiple directories:

$ mkdir reports 'low power adders'

$ ls -1
low power adders
reports

The -p option will help you to create multiple directory hierarchies in one shot:

# error because 'a' and 'a/b' paths do not exist yet
$ mkdir a/b/c
mkdir: cannot create directory ‘a/b/c’: No such file or directory

# -p is handy in such cases
$ mkdir -p a/b/c

The -p option has another functionality too. It will not complain if the directory you are trying to create already exists. This is especially helpful in shell scripts.

# 'reports' directory was already created in an earlier example
$ mkdir reports
mkdir: cannot create directory ‘reports’: File exists
# exit status will reflect that something went wrong
$ echo $?
1

# the -p option will override such errors
$ mkdir -p reports
$ echo $?
0

As seen in the examples above, you can check the exit status of the last executed command using the $? special variable. 0 means everything went well and higher numbers indicate some sort of failure has occurred (the details of which you can look up in the command's manual).

info warning Linux filenames can use any character other than / and the ASCII NUL character. Quote the arguments if it contains characters like space, *, etc to prevent shell expansion. Shell considers space as the argument separator, * is a wildcard character and so on. As a good practice, use only alphabets, numbers and underscores for filenames, unless you have some specific requirements.

info You can delete the practice directory if you wish:

$ cd ..
$ rm -r practice_mkdir

Further Reading

touch

You'll usually create files using a text editor or by redirecting output of a command to a file. For some cases, empty files are needed for testing purposes or to satisfy a particular build process. A real world use case is the empty .nojekyll file for GitHub Pages.

The touch command's main functionality is altering timestamps (which will be discussed in a later chapter). If a file doesn't exist, touch will create an empty file using the current timestamp. You can also pass more than one file argument if needed.

$ mkdir practice_touch
$ cd practice_touch

$ ls ip.txt
ls: cannot access 'ip.txt': No such file or directory

$ touch ip.txt

$ ls -s ip.txt
0 ip.txt

info You can also create an empty file using > ip.txt, but one difference is that the file gets overwritten if it already exists.

rm

The rm command will help you to delete files and directories. You can pass one or more paths as arguments:

# change to the 'scripts' directory and source the 'rm.sh' script
$ source rm.sh
$ ls -F
empty_dir/  hello.py  loops.py  projects/  read_only.txt  reports/

# delete files ending with .py
$ rm *.py
$ ls -F
empty_dir/  projects/  read_only.txt  reports/

You'll need to add -r option to recursively delete directory contents. You can use rm -d or the rmdir command to delete only empty directories.

# -r is needed to delete directory contents recursively
$ rm reports
rm: cannot remove 'reports': Is a directory
$ rm -r reports
$ ls -F
empty_dir/  projects/  read_only.txt

# delete empty directories, same as using the 'rmdir' command
$ rm -d empty_dir
# you'll get an error if the directory is not empty
$ rm -d projects
rm: cannot remove 'projects': Directory not empty

Typos like misplaced space, wrong glob, etc could wipe out files not intended for deletion. Apart from having backups and snapshots, you could also take some mitigating steps:

  • using -i option to interactively delete each file
    • you can also use -I option for lesser number of prompts
  • using echo as a dry run to see how the glob expands
  • using a trash command (see links below) instead of rm

Use y for confirmation and n to cancel deletion with -i or -I options. Here's an example of cancelling deletion:

$ rm -ri projects
rm: descend into directory 'projects'? n

$ ls -F
projects/  read_only.txt

And here's an example of providing confirmation at each step of the deletion process:

$ tree projects
projects
├── calculator
│   └── calc.sh
└── tictactoe
    └── game.py

2 directories, 2 files

$ rm -ri projects
rm: descend into directory 'projects'? y
rm: descend into directory 'projects/tictactoe'? y
rm: remove regular empty file 'projects/tictactoe/game.py'? y
rm: remove directory 'projects/tictactoe'? y
rm: descend into directory 'projects/calculator'? y
rm: remove regular empty file 'projects/calculator/calc.sh'? y
rm: remove directory 'projects/calculator'? y
rm: remove directory 'projects'? y

$ ls -F
read_only.txt

The -f option can be used to ignore complaints about non-existing files (somewhat similar to the mkdir -p feature). It also helps to remove write protected files (provided you have appropriate permissions to delete those files). This option is especially useful for recursive deletion of directories, .git/objects for example (see my list for links to git and related resources).

$ rm xyz.txt
rm: cannot remove 'xyz.txt': No such file or directory
$ echo $?
1
$ rm -f xyz.txt
$ echo $?
0

# example for removing write protected files
# you'll be asked for confirmation even without the -i/-I options
$ rm read_only.txt
rm: remove write-protected regular empty file 'read_only.txt'? n
# with -f, files will be deleted without asking for confirmation
$ rm -f read_only.txt

Further Reading

cp

You can use the cp command to make copies of files and directories. With default syntax, you have to specify the source first followed by the destination. To copy multiple items, the last argument as destination can only be a directory. You'll also need to use the -r option to copy directories (similar to rm -r seen earlier).

# change to the 'scripts' directory and source the 'cp.sh' script
$ source cp.sh
$ ls -F
backups/  reference/

# recall that . is a relative path referring to the current directory
$ cp /usr/share/dict/words .
$ ls -F
backups/  reference/  words

# error because -r is needed to copy directories
# other file arguments (if present) will still be copied
$ cp /usr/share/dict .
cp: -r not specified; omitting directory '/usr/share/dict'
$ cp -r /usr/share/dict .
$ ls -F
backups/  dict/  reference/  words

info warning By default, cp will overwrite an existing file of the same name in the destination directory. You can use the -i option to interactively confirm or deny overwriting existing files. The -n option will prevent overwriting existing files without asking for confirmation.

$ echo 'hello' > ip.txt
$ ls -F
backups/  dict/  ip.txt  reference/  words
$ ls backups
ip.txt  reports
$ cat backups/ip.txt
apple banana cherry
# file will be overwritten without asking for confirmation!
$ cp ip.txt backups/
$ cat backups/ip.txt
hello

# use -i to interactively confirm or deny overwriting
$ echo 'good morning' > ip.txt
$ cp -i ip.txt backups/
cp: overwrite 'backups/ip.txt'? n
$ cat backups/ip.txt
hello

# use -n to prevent overwriting without needing confirmation
$ cp -n ip.txt backups/
$ cat backups/ip.txt
hello

If there's a folder in the destination path with the same name as a folder being copied, the contents will be merged. If there are files of identical names in such directories, the same rules discussed above will apply.

$ tree backups
backups
├── ip.txt
└── reports
    └── jan.log

1 directory, 2 files

$ mkdir reports
$ touch reports/dec.log
$ cp -r reports backups/
$ tree backups
backups
├── ip.txt
└── reports
    ├── dec.log
    └── jan.log

1 directory, 3 files

Often, you'd want to copy a file (or a directory) under a different name. In such cases, you can simply use a new name while specifying the destination.

# copy 'words' file from source as 'words_ref.txt' at destination
$ cp /usr/share/dict/words words_ref.txt

# copy 'words' file as 'words.txt' under the 'reference' directory
$ cp /usr/share/dict/words reference/words.txt

# copy 'dict' directory as 'word_lists'
$ cp -r /usr/share/dict word_lists

As mentioned earlier, to copy multiple files and directories, you'll have to specify the destination directory as the last argument. You can use the -t option if you want to specify the destination before the source paths (helpful with find command for example, will be discussed later).

$ cp -r ~/.bashrc /usr/share/dict backups/

$ ls -AF backups
.bashrc  dict/  ip.txt  reports/

Here are some more notable options:

  • -u copy files from source only if they are newer or don't exist in the destination
  • -b and --backup options will allow you to create backup copies of files already existing in the destination
  • --preserve option will help you to copy files along with source file attributes like ownership, timestamp, etc

Further Reading

mv

You can use the mv command to move one or more files and directories from one location to another. Unlike rm and cp, you do not need the -r option for directories.

Syntax for specifying the source and destination is same as seen earlier with cp. Here's an example of moving a directory into another directory:

# change to the 'scripts' directory and source the 'mv.sh' script
$ source mv.sh
$ ls -F
backups/  dot_files/  hello.py  ip.txt  loops.py  manuals/
$ ls -F backups
projects/

$ mv dot_files backups

$ ls -F
backups/  hello.py  ip.txt  loops.py  manuals/
$ ls -F backups
dot_files/  projects/

Here's an example for moving multiple files and directories to another directory:

$ mv *.py manuals backups

$ ls -F
backups/  ip.txt
$ ls -F backups
dot_files/  hello.py  loops.py  manuals/  projects/

When you are dealing with a single file or directory, you can also rename them:

# within the same directory
$ mv ip.txt report.txt
$ ls -F
backups/  report.txt

# between different directories
$ mv backups/dot_files rc_files
$ ls -F
backups/  rc_files/  report.txt
$ ls -F backups
hello.py  loops.py  manuals/  projects/

Here are some more notable options, some of which behave similar to those seen with the cp command:

  • -i interactively confirm or deny when the destination already has a file of the same name
  • -n always deny overwriting of files
  • -f always overwrite files
  • -t specify the destination elsewhere instead of final argument
  • -u move only if the files are newer or don't exist in the destination
  • -b and --backup options will allow you to create backup copies of files already existing in the destination
  • -v verbose option

Further Reading

rename

mv command is useful for single file renaming. rename helps when you need to change multiple filenames based on a pattern. There are different implementations of the rename command, with wildly different set of features. See askubuntu: What's the difference between the different "rename" commands? for details.

Perl implementation of the rename command will be discussed in this section. You'd need to know regular expressions to use this command. Basic explanations will be given here, more details can be found in the links mentioned at the end of this section. Here's an example to change the file extensions:

$ mkdir practice_rename
$ cd practice_rename
# create sample files
$ touch caves.jpeg waterfall.JPEG flower.JPG

# substitution command syntax is s/search/replace/flags
# \. matches . character literally
# e? matches e optionally (? is a quantifier to match 0 or 1 times)
# $ anchors the match to the end of the input
# i flag matches the input case-insensitively
$ rename 's/\.jpe?g$/.jpg/i' *

$ ls
caves.jpg  flower.jpg  waterfall.jpg
$ rm *.jpg

As a good practice, use the -n option to see how the files will be renamed before actually renaming the files.

$ touch 1.png 3.png 25.png 100.png
$ ls
100.png  1.png  25.png  3.png

# use -n option for sanity check
# note that 100.png isn't part of the output, since it isn't affected
# \d matches a digit character
# \d+ matches 1 or more digits (+ is a quantifier to match 1 or more times)
# e flag treats the replacement string as Perl code
# $& is a backreference to the matched portion
$ rename -n 's/\d+/sprintf "%03d", $&/e' *.png
rename(1.png, 001.png)
rename(25.png, 025.png)
rename(3.png, 003.png)

# remove -n option after sanity check to actually rename the files
$ rename 's/\d+/sprintf "%03d", $&/e' *.png
$ ls
001.png  003.png  025.png  100.png

If the new filename already exists, you'll get an error, which you can override with -f option if you wish. If you are passing filenames with path components in them, you can use the -d option to affect only the filename portion. Otherwise, the logic you are using might affect directory names as well.

$ mkdir projects
$ touch projects/toc.sh projects/reports.py

# aim is to uppercase the non-extension part of the filename
# [^.]+ matches 1 or more non '.' characters
# \U changes the following characters to uppercase
# $& is a backreference to the matched portion
$ rename -n -d 's/[^.]+/\U$&/' projects/*
rename(projects/reports.py, projects/REPORTS.py)
rename(projects/toc.sh, projects/TOC.sh)

# without -d option, directory name will also be affected
$ rename -n 's/[^.]+/\U$&/' projects/*
rename(projects/reports.py, PROJECTS/REPORTS.py)
rename(projects/toc.sh, PROJECTS/TOC.sh)

Further Reading

ln

The ln command helps you create a link to another file or directory within the same or different path. There are two types of links — symbolic links and hard links. Symbolic links point to another file or directory:

  • if the original file is deleted or moved to another location, symbolic link will no longer work
  • if the symbolic link is moved to another location, it will still work if the link was done using absolute path (for relative path, it will depend on whether or not there's another file with same name in that location)
  • a symbolic link file has its own inode, permissions, timestamps, etc
  • some commands will work the same when original file or the symbolic file is given as the command line argument and some require additional options

Syntax is similar to the cp command. You have to specify the source first followed by the destination (which is optional if it is the current working directory).

$ mkdir practice_ln
$ cd practice_ln

# destination is optional for making a link in the current directory
$ ln -s /usr/share/dict/words

# you can also rename the link if needed
$ ln -s /usr/share/dict/words words.txt
$ ls -1sF
total 0
0 words@
0 words.txt@

Long listing with ls -l will show the path connected to links. You can also use the readlink command, which has features like resolving recursively to the canonical file.

# to know which file the link points to
$ ls -lG words
lrwxrwxrwx 1 learnbyexample 21 Jul  9 13:41 words -> /usr/share/dict/words
$ readlink words
/usr/share/dict/words

# the linked file may be another link
# use -f option to get the original file
$ readlink -f words
/usr/share/dict/english

Hard links can only point to another file. You cannot use them for directories and the usage is also restricted to within the same filesystem. The . and .. directories are exceptions, these special purpose hard links are automatically created. Here are some more details about hard links:

  • once a hard link is created, there is no distinction between the two files other than their paths. They have same inode, permissions, timestamps, etc
  • hard links will continue working even if all the other hard links are deleted
  • if a hard link is moved to another location, the links will still be in sync. Any change in one of them will be reflected in all the other links
$ touch apple.txt
$ ln apple.txt banana.txt

# the -i option gives inode
$ ls -1i apple.txt banana.txt
649140 banana.txt
649140 apple.txt

info You can use unlink or rm commands to delete links.

Further Reading

tar and gzip

tar is an archiving utility. Depending on the implementation, you can also use options to compress the archive.

Here's an example that creates a single archive file from multiple input files and directories:

# change to the 'scripts' directory and source the 'tar.sh' script
$ source tar.sh
$ ls -F
projects/  report.log  todos/

# -c option creates a new archive, any existing archive will be overwritten
# -f option allows to specify name of archive to be created
# rest of the arguments are the files/directories to be archived
$ tar -cf bkp.tar report.log projects

$ ls -F
bkp.tar  projects/  report.log  todos/
$ ls -sh bkp.tar
7.4M bkp.tar

Once you have an archive file, you can then compress it using tools like gzip, bzip2, xz, etc. In the below example, the command replaces the archive file with the compressed version and adds a .gz suffix to indicate that gzip was the technique used.

# the input '.tar' file will be overwritten with the compressed version
$ gzip bkp.tar

$ ls -F
bkp.tar.gz  projects/  report.log  todos/
$ ls -sh bkp.tar.gz
5.6M bkp.tar.gz

Use the -t option if you want to check the contents of the compressed file. This will work with the uncompressed .tar version as well.

$ tar -tf bkp.tar.gz 
report.log
projects/
projects/scripts/
projects/scripts/calc.sh
projects/errors.log

To uncompress .gz files, you can use gunzip or gzip -d. This will replace the compressed version with the uncompressed archive file:

# this '.gz' file will be overwritten with the uncompressed version
$ gunzip bkp.tar.gz

$ ls -F
bkp.tar  projects/  report.log  todos/
$ ls -sh bkp.tar
7.4M bkp.tar

To extract the files from an archive, use tar along with the -x option:

$ mkdir test_extract
$ mv bkp.tar test_extract
$ cd test_extract
$ ls
bkp.tar

$ tar -xf bkp.tar
$ tree
.
├── bkp.tar
├── projects
│   ├── errors.log
│   └── scripts
│       └── calc.sh
└── report.log

2 directories, 4 files

$ cd ..
$ rm -r test_extract

With GNU tar, you can compress/uncompress along with the tar command instead of having to use tools like gzip separately. For example, the -z option will use gzip, -j will use bzip2 and -J will use xz. Use the -a option if you want tar to automatically select the compression technique based on the extension provided.

$ ls -F
projects/  report.log  todos/

# -z option gives same compression as the gzip command
$ tar -zcf bkp.tar.gz report.log projects
$ ls -sh bkp.tar.gz
5.6M bkp.tar.gz

# extract original files from compressed file
$ mkdir test_extract
$ cd test_extract
$ tar -zxf ../bkp.tar.gz
$ tree
.
├── projects
│   ├── errors.log
│   └── scripts
│       └── calc.sh
└── report.log

2 directories, 3 files

$ cd ..
$ rm -r test_extract

tar has lots and lots of options for various needs. Some are listed below, see documentation for complete details.

  • -v verbose option
  • -r to append files to an existing archive
  • --exclude= specify files to be ignored from archiving

There are also commands starting with z to work with compressed files, for example:

  • zcat to display file contents of a compressed file
  • zless to display file contents of a compressed file one screenful at a time
  • zgrep to search compressed files

info If you need to work with .zip files, use zip and unzip commands.

Further Reading