bash script use function call in string substitution

The presented way:

status=$(~/bin/status)
echo "${status##* }"

is the good and best way to do it. Use it, it is great.

Notes: I typically use $tmp variable name for such things. Prefer lower case variables for script temporary variables, use upper case for exported variables. Use quotes to prevent filename expansion. Check your script with shellcheck.

You’re correct, Bash’s parameter expansion only works on variables, so you can’t apply ${…##* } directly to a command substitution. ${$(command)##pattern} is not valid Bash syntax.

If you want to avoid a variable, you could use one of these alternatives:

~/bin/status | awk '{print $NF}'

or

set -- $(~/bin/status)
echo "${!#}"

Both extract the last field without needing to store the full output in a separate variable.

One simple way:

read _ _ var < <(~/bin/status)
echo $var

Process Substitution >(command ...) or <(...) is replaced by a temporary filename. Writing or reading that file causes bytes to get piped to the command inside. Often used in combination with file redirection: cmd1 2> >(cmd2).

See http://mywiki.wooledge.org/ProcessSubstitution http://mywiki.wooledge.org/BashFAQ/024

printf has all the same tools as the C version.

printf "%.0s%.0s%s\n" $(~/bin/status)

This prints the first two strings received as zero-length.

$: tst(){ local x=( running halted ); echo program is ${x[RANDOM%2]}; }
program is running
$: tst
program is halted

$:  printf "%.0s%.0s%s\n" $(tst)
running

You could also just use a filter pipe, since $() runs another subshell anyway.

$: tst | awk '{print $3}'
running

$: tst | cut -d' ' -f 3
halted

$: tst | sed 's/^program is //'
halted

$: tst | grep -Eo 'running|halted'
running

And if you want it "pretty" you could add a little local syntactic sugar.

$: myStat() { tst | cut -d' ' -f 3; }
$: myStat
running

$: alias other='tst | cut -d" " -f 3'
$: other
running

No variables.

is there a way to do this without storing the output of the status-script in a variable?

Probably not without being computationally more inefficient by piping the output into an additional process.

Also depends on your definitions of "variable" and "storing". (Are the positional parameters $@ variables? Or the standard input buffer to a program? Or fields or input records in awk? Or the pattern and hold spaces in sed? etc).

To add to the other answers:

: $(~/bin/status); echo "$_"

Bash defines _ as:

($_, an underscore.) This has a number of meanings depending on context. At shell startup, $_ set to the pathname used to invoke the shell or shell script being executed as passed in the environment or argument list. Subsequently, it expands to the last argument to the previous simple command executed in the foreground, after expansion. It is also set to the full pathname used to invoke each command executed and placed in the environment exported to that command. When checking mail, $_ expands to the name of the mail file.

(Emphasis mine.)

Just pipe it through awk, like this:

~/bin/status | awk '{print $NF}'

Test with echo:

$ echo 'program is running' | awk '{print $NF}'
running