TL;DR

I noticed a pattern in my shell scripts lately.

Well, a few actually, but here’s one.

I already wrote about a possible way to implement a POSIX shell “modulino”, like this:

#!/bin/sh

module_function() {
   printf 'module function here\n'
}

main() {
   module_function
   printf 'main here\n'
}

! grep -- '6f8fc63c06-whatevah-db049fa26dd5953b5771e4' "$0" >/dev/null 2>&1 \
   || main "$@"

This allows writing scripts that double down as libraries of shell functions. The next step is to give access to all functions as if they were sub-commands:

#!/bin/sh

help() {
   printf >&2 'this is some help\n'
}

module_function() {
   printf 'module function here\n'
}

main() {
   [ $# -gt 0 ] || set -- help
   "$@"
}

! grep -- '6f8fc63c06-whatevah-db049fa26dd5953b5771e4' "$0" >/dev/null 2>&1 \
   || main "$@"

The main function is called only if the script is run directly (i.e. it is not sourced), and is passed the whole argument list. If this list is empty, it sets it to a default value (in this case, help); in any case, it executes the list as a command.

This leads to the following behaviour (suppose the script is named modulino.sh):

$ ./modulino.sh        # calls the default function, i.e. help
this is some help

$ ./modulino.sh help   # calls help explicitly 
this is some help

$ ./modulino.sh module_function  # calls other function
module function here

Neat, uh?