How do I set the output of a function as a variable value in a bash script?

Question

I have the following function to count the number of files in a directory, within my bash script.

file_count() {
  no_of_files=$(find "$1" -maxdepth 1 -type f -printf '.' | wc -c)
}

I'd like to use it repeatedly on different directories and save the count into a variable for each directory. Currently to do this, I use

file_count $somedir
files_in_somedir="$no_of_files"

I'm aware that I'm setting the no_of_files variable outside of the function each time, and would like to make it local to the function, not settign an intermediate variable in the main script. This is just in case there's some mistake meaning that the variable doesn't change between calls of the function (mistyping the function name maybe), and the old value of no_of _files is used.

If my function were:

file_count() {
  local no_of_files=$(find "$1" -maxdepth 1 -type f -printf '.' | wc -c)
}

How would I easily set these directory count variables?

Answer 1

Bash functions are not like functions on other programming languages, they are more like commands. This means they have no classical return value, but

• an exit/return code. This is an integer number in the range 0-255, where 0 means "success" and every other value represents an error. If you try to specify a number outside this range, it will be taken modulo 256 (add or subtract 255 from your number until it fits the range 0-255).

  This code is automatically set to the return code of the last statement that got executed inside the function, unless you manually set it using the return command, like this:

  return 1
  

• output streams. Each Bash function can write arbitrary strings to the output streams (STDOUT and STDERR), just like normal scripts. The output can either be directly from the commands you run in the function, or set manually by using echo.

  But instead of letting this output get displayed in the console, you can capture it when you run the function, e.g. using Bash's command substitution syntax, and store it in a variable in your main script:

  example_function() {
      # do something useful
      echo "return this message"
  }
  
  returned_value="$(example_function)"
  

So your code would have to look like this:

file_count() {
    find "$1" -maxdepth 1 -type f -printf '.' | wc -c
    # the line above already prints the desired value, so we don't need an echo
}

files_in_somedir="$(file_count "$somedir")"

See Return value in bash script (on Stack Overflow) for more info.

Answer 2

I'm a C++ programmer, so I like to mimic the "getter and setter" philosophy, where you use tiny little functions to get and set the value of a variable. I know there are some non-portable solutions out there like new versions of declare, but I like writing only portable code. The downside to using getters and setters is that you need to create a function (or two functions) for every value you want to manage. So... I made a "factory function" that handles creating getters/setters for you (below). Then, you just drop the equals sign for setting, and everything else looks pretty similar (see usage)... except now, you can pass "pointer-like things" into functions.

makeGetSet() {
. /dev/fd/0 <<END_FUNC
${1}Val()  { [ "\${1}" ] && ${1}="\${1}" || echo "\${${1}}"; }
END_FUNC
}

Usage:

Setup: makeGetSet myVariable
Set: myVariableVal newValue
Get: anotherVariable="`myVariableVal`"
Print: myVariableVal

It does not require any particular shell (100% POSIX) with special features like indirection or namerefs, or the declare utility. You just pass in your variable name to makeGetSet, and your getter/setter function has that same name with a "Val" at the end (e.g. myVariableVal). Tested with bash and dash.

For your specific question, you would use it like this:

makeGetSet files_in_somedir

file_count() {
  "${2}" $(find "$1" -maxdepth 1 -type f -printf '.' | wc -c)
}

file_count "/some/directory" files_in_somedirVal