Enough developers out there deliver great day-to-day work without having to touch bash whatsoever. Yet, every now and then we all encounter a piece of bash script, be it from a colleague or copied from a Stack Overflow post. Yeah, the script works, but do we understand what is happening? I wanted to know myself so I started small. Let's dive into redirection.
Redirection
A program that runs in the command line always has three important ways to connect with it; the input, output, and possible error output of the program. We call these types of connections data- or file streams.
Examples:
- The input of a program can be a keyboard or the output of another command, file, or program.
- Examples of the output or error output could be another file or the terminal.
These three file streams have specific names and even values. These values are called 'file descriptors'. File descriptors are nothing more than numbers that are given to specific files. These three, have the numbers 0 to 2:
| Name | Symbolic name | File descriptor | Sign | Default |
|---|---|---|---|---|
| Standard input | stdin | 0 | < | Keyboard |
| Standard output | stdout | 1 | > | Terminal |
| Standard error | stderr | 2 | 2> | Terminal |
We can change the sources of these streams. This is what we call redirection.
Example
Let's say we have a list of animals in a file called animals.txt.
puppy
kangaroo
kitten
snake
wombat
tiger
bird
What if we want to sort this list and save the outcome to a new file?
$ sort < animals.txt > sortedAnimals.txt
We call the sort command and give it animals.txt as input by using the < symbol. Then we set the sortedAnimals.txt file as the output by using the > symbol.
When we open the sortedAnimals.txt file we will find:
bird
kangaroo
kitten
puppy
snake
tiger
wombat
Now, what if we wanted to sort a file that does not exist? For example:
$ sort < ghostFile.txt > sortedGhostFile.txt
-bash: ghostFile.txt: No such file or directory
We get an error thrown into the terminal. Of course, this is good because we want to get feedback on why something is not working. But what if we didn’t want the error to be thrown into the terminal, but written to a file? We could redirect the standard error (stderr) to a file as well:
$ sort 2> sortedGhostFileError.txt < ghostFile.txt
We first say where the stderr of sort needs to direct to with the 2> sign. Then we say that ghostFile.txt needs to be input for the sort command by using the < sign. If you were to check, you would see that the file sortedGhostFileError.txt has been created and it contains the same error we saw printed into the terminal before.
You can also write the standard output and standard error to the same file. There are two ways to do this.
$ sort > ghostFileOutput.txt 2>&1 < ghostFile.txt
We first define that ghostFileOutput.txt is the output, and then use the 2>&1 sign to redirect the standard error (2) to standard output (1).
You can also use the shorthand:
$ sort &> ghostFileOutput.txt < ghostFile.txt
Pipelines
Our next step in understanding redirection is by understanding ‘Pipelines’. The pipe operator | lets us chain different commands together. This can be very useful. It lets us write small programs or commands, and chain them to let them perform more complex tasks. With the pipe symbol, you can use the output of one command as the input of the next.
Let's say that we want to use the same list.txt we had before. We want to sort the list and grab the animal that is first in the list after sorting. We could do:
$ sort animals.txt | head -1
This will first sort the list.txt file, use the outcome of that as the input of the head -1 command, which will grab only the first line of the input it gets. As we have not specified what output the last command should have, it just prints bird to the terminal. As it is the first animal in the sorted list.
We could also save that animal as a separate file by combining our pipe symbol with one of the redirection symbols:
$ sort animals.txt | head -1 > first-animal.txt
This will save the outcome to a file called first-animal.txt.
Redirection summary
So, a file can have input, output, and error output. We can explicitly say where these three need to point at. This is possible by using different signs or operators. Including <, >, 2>, and the pipe operator |. If you want to read more about this try typing man bash in your terminal and search for /REDIRECTION. Other than that, try it out yourself by fiddling around in your terminal. Have fun!
Do you know an interesting fact about redirection? Make sure to share it in the comments below!
Top comments (1)
Rather than only showing the information direction, maybe also show sed examples?
I would be interested :)
edit: I always like to see bash / zsh stuff, good job :)