So, we've already created two password generators. Since three is the magic number, and this is a really good example to demonstrate chaining pipes, we'll create one more (the last one, promise):

reader@ubuntu:~/scripts/chapter_12$ vim piped-passwords.sh
reader@ubuntu:~/scripts/chapter_12$ cat piped-passwords.sh 
#!/bin/bash

#####################################
# Author: Sebastiaan Tammer
# Version: v1.0.0
# Date: 2018-11-10
# Description: Generate a password, using only pipes.
# Usage: ./piped-passwords.sh
#####################################

password=$(head /dev/urandom | tr -dc 'a-zA-Z0-9' | head -c20)

echo "Your random password is: ${password}"

First, we grab the first 10 lines from /dev/urandom (the default behavior for head). We send this to tr, which trims it to the character sets we want (because it also outputs non-readable characters). Then, when we have a character set that we can use, we grab the first 20 characters from that using head again.

If you run just head /dev/urandom | tr -dc 'a-zA-Z0-9' a few times, you'll see that the length differs; this is because of the randomness of the newline byte. By grabbing 10 lines from /dev/urandom, the chance of not having enough readable characters to create a 20-character password is very small.

(Challenge to the reader: create a script with a loop that does this long enough to encounter this situation!)

This example illustrates a few things. Firstly, we can often achieve a lot of things we want to do with a few smart pipes. Secondly, it is not unusual to use the same command multiple times. We could have also chosen tail -c20 for the final command in the chain, by the way, but this has a nice symmetry to the whole command!

Finally, we have seen three different password generators that do, in reality, the same thing. As always, in Bash there are many ways to accomplish the same goal; it is up to you to decide which is most applicable. Readability and performance should be the two main factors in this decision, as far as we're concerned.