Data is entered into the computer via stdin (usually the keyboard), and the resulting output goes to stdout (usually the shell). These pathways are called streams. However, it's possible to alter these input and output locations, causing the computer to get information from somewhere other than stdin or send the results somewhere other than stdout. This functionality is referred to as redirection.
In this article, you'll learn five redirect operators, including one for stderr. I've provided examples of each and presented the material in a way that you can duplicate on your own Linux system.
Regular output > operator
The output redirector is probably the most recognized of the operators. The standard output (stdout) is usually to the terminal window. For example, when you type the
date command, the resulting time and date output is displayed on the screen.
[damon@localhost ~]$ date Tue Dec 29 04:07:37 PM MST 2020 [damon@localhost ~]$
It is possible, however, to redirect this output from stdout to somewhere else. In this case, I'm going to redirect the results to a file named
specifications.txt. I'll confirm it worked by using the
cat command to view the file contents.
[damon@localhost ~]$ date > specifications.txt [damon@localhost ~]$ cat specifications.txt Tue Dec 29 04:08:44 PM MST 2020 [damon@localhost ~]$
The problem with the
> redirector is that it overwrites any existing data in the file. At this stage, you now have
date information in the
specifications.txt file, right? If you type
hostname > specifications.txt, the output will be sent to the text file, but it will overwrite the existing time and date information from the earlier steps.
[damon@localhost ~]$ hostname > specifications.txt [damon@localhost ~]$ cat specifications.txt localhost.localdomain [damon@localhost ~]$
It is easy to get in trouble with the
> redirector by accidentally overwriting existing information.
[ Readers also enjoyed: 10 basic Linux commands you need to know ]
Regular output append >> operator
>> operator adds the output to the existing content instead of overwriting it. This allows you to redirect the output from multiple commands to a single file. For example, I could redirect the output of
date by using the
> operator and then redirect
uname -r to the
specifications.txt file by using
[damon@localhost ~]$ date > specifications.txt [damon@localhost ~]$ hostname >> specifications.txt [damon@localhost ~]$ uname -r >> specifications.txt [damon@localhost ~]$ cat specifications.txt Tue Dec 29 04:11:51 PM MST 2020 localhost.localdomain 5.9.16-200.fc33.x86_64 [damon@localhost ~]$
>> redirector even works on an empty file. That means that you could conceivably
ignore the regular
> redirector to alleviate the potential to overwrite data, and always rely on the
>> redirector instead. It's not a bad habit to get into.
Regular input < operator
The input redirector pulls data in a stream from a given source. Usually, programs receive their input from the keyboard. However, data can be pulled in from another source, such as a file.
It's time to build an example by using the
sort command. First, create a text file named
mylist.txt that contains the following lines:
cat dog horse cow
Observe that the animals are not listed in alphabetical order.
What if you need to list them in order? You can pull the contents of the file into the
sort command by using the
[damon@localhost ~]$ sort < mylist.txt cat cow dog horse [damon@localhost ~]$
You could even get a little fancier and redirect the sorted results to a new file:
[damon@localhost ~]$ sort < mylist.txt > alphabetical-file.txt [damon@localhost ~]$ cat alphabetical-file.txt cat cow dog horse [damon@localhost ~]$
Regular error 2> operator
The stdout displays expected results. If errors appear, they are managed differently. Errors are labeled as file descriptor 2 (standard output is file descriptor 1). When a program or script does not generate the expected results, it throws an error. The error is usually sent to the stdout, but it can be redirected elsewhere. The stderr operator is
2> (for file descriptor 2).
Here is a simple example, using the misspelled
[damon@localhost ~]$ png bash: png: command not found... [damon@localhost ~]$
Here is the same misspelled command with the error output redirected to
[damon@localhost ~]$ png 2> /dev/null [damon@localhost ~]$
The resulting error message is redirected to
/dev/null instead of the stdout, so no result or error message is displayed on the screen.
/dev/null, or the bit bucket, is used as a garbage can for the command line. Unwanted output can be redirected to this location to simply make it disappear. For example, perhaps you're writing a script, and you want to test some of its functionality, but you know it will throw errors that you don't care about at this stage of development. You can run the script and tell it to redirect errors to
/dev/null for convenience.
Pipe | operator
Ken Hess already has a solid article on using the pipe
| operator, so I'm only going to show a very quick demonstration here.
The pipe takes the output of the first command and makes it the input of the second command. You might want to see a list of all directories and files in the
/etc directory. You know that's going to be a long list and that most of the output will scroll off the top of the screen. The
less command will break the output into pages, and you can then scroll upward or downward through the pages to display the results. The syntax is to issue the
ls command to list the contents of
/etc, and then use pipe to send that list into
less so that it can be broken into pages.
[damon@localhost ~]$ ls /etc | less
Ken's article has many more great examples. Personally, I find myself using
command | less and
command | grep string the most often.
[ Download now: A sysadmin's guide to Bash scripting. ]
Redirect operators are very handy, and I hope this brief summary has provided you with some tricks for manipulating input and output. The key is to remember that the
> operator will overwrite existing data.