Linux Cent

Bash Error Output Redirect

Each open file gets assigned a file descriptor. The file descriptors for STDIN is 0, for STDOUT is 1 nad STDERR is 2.

If you want to redirect just STDERR (standard error output) to file, ju do:

cmd_name 2> /file

If you want o redirect STDOUT to other files, and STDERR to other files, just do:

cmd_name >/stdout_file 2>/stderr_file

If you want to merge both (STDERR, STDOUT) into one file, you can do:

cmd_name >/file_4_both 2>&1

For the same effect, you can also use this syntax:

cmd_name &> /file_4_both

Even more, it is very useful to use the tee command. By definition, tee read from standard input and write to standard output and files, in same time.

In next example, we merge STDOUT and STDERR of “find /” command together. Then, we pass it to STDIN of tee command. Tee command is executed with -a parameter, that means append to an existing file (if exists):

find / 2>&1 | tee -a /home/tee.output
ll /home

Output: (lines omitted) -rw-r–r–. 1 root root 2.2M Jun 10 13:16 tee.output

The child process inherits open file descriptors. If you want to prevent file descriptors from being inherited, close it. For example:

<&-

This close stddin descriptor.

What is Bash Script

Metaphorically speaking bash script is like a ‘to-do list’. After you read the first entry you start realizing it. After you finished first entry, you continue with the second entry and so on.

A bash script is a text file that contains a mixture of commands.

Bash script can contain also functions, loops, conditional constructs. Scripts are commonly used for administration task like change file permission, creating disk backups. Using bash scripts is often faster than using the graphical user interface.

It’s important to mention that there is no difference between putting series of 10 commands into a script file and executing that script or you entering commands one by one to the command-line interface. In both situations, the result will be exactly the same thing.

After you create your script it is good practice to add the extension ‘.sh’ to the filename, for example ‘myFirstScript.sh’.

Bash Error Output

Bash provides I/O redirection. There are 3 standard files: STDIN (standard input) with descriptor 0, STDOUT (standard output) with descriptor 1, and STDERR (standard error) with descriptor 2.

If you want to redirect your messages to STDERR, you can use >&2 symbol. This symbol is abbreviation of 1>&2 symbol which means, that everything in STDOUT will go to STDERR.

So, if you want to put the message “Cannot delete directory” in STDERR, you can do it this way:

echo Cannot delete directory >&2

Even more, you might to create your own function for error messages:

recho() { echo "$*" >&2 ; }
recho "Cannot delete directory" > /dev/null

Output: Cannot delete the directory

On the first line, we define recho function (error echo). This function will print all its arguments ($*) to STDERR (>&2).

On the second line, we try to use the function. To prove, that output will be written to STDERR, we will redirect STDIN to nowhere (/dev/null). So, if you see some output, it should be in STDERR.

Bash How to Trim String

You can use a bash parameter expansion, sed, cut, and tr to trim a string.

Let’s use bash parameter expansion to remove all whitespace characters from the variable foo:

foo="Hello world."
echo "${foo//[["space:]]/}"

Output: Helloworld.

“${foo// /}” removes all space characters, “$foo/ /” removes the first space character.

To remove only space characters before and after string use sed:

foo=" Hello world. "
echo "${foo}" | sed -e 's/^[[:space:]]*//'

Output: Hello world.

Easy to use and remember the way how to remove whitespaces before and afterword:

echo " text. text2 " | xargs

xargs remove all whitespaces before “text.” and let one space between text. and text2.

If you want to delete only the last character from the variable:

foo="hello"
echo "${foo::-1}"

Output: hell

Bash Hello World Script

This bash example creates an archive from /home directory to /backup/ directory as one tar.gz file. Let’s create a file backup.sh. It will consist of two lines:

#!/bin/bash
tar -czf /var/home-backup.tar.gz /home/

First line is a hashpling. Basically, it says who execute script. In this example, we choose /bin/bash.

The second line is tar command. It tarballs and compress the whole directory (/home) to one file.

I recommend you give also the third line (empty line). Why? If you execute this script in unusual UNIXes (ec. SCO UNIX), UNIX coudn’t execute last line, becasuse last symbol of file – EOF (end of file) is different from symbol EOLN (end of line). Symbol EOLN (enter) or semicolon (;) executes command.

Would you to extend this script to some output? Here is an example.

#!/bin/bash
echo -n Creating backup of home directory to /backup...
tar -czf /var/home-backup.tar.gz /home/ >/dev/null 2>&1
echo done.

On the second line, echo with n parameter doesn’t give a new line.

On the third line, output from tar command is redirected to trash (/dev/null).

Last line, just echoes done.

If you don’t know what to put on the first line (hashpling) type:

echo $SHELL

Output: /bin/bash

You will get path to your shell, which can use in the hashspling.