printf field width doesn't support multibyte characters?

Question

I want printf to recognize multi-byte characters when calculating the field width so that columns line up properly... I can't find an answer to this problem and was wondering if anyone here had any suggestions, or maybe a function/script that takes care of this problem.

Here's a quick and dirty example:

printf "## %5s %5s %5s ##\n## %5s %5s %5s ##\n" '' '*' '' '' "•" ''
>##           *       ##
>##         •       ##

Obviously, I want the result:

>##           *       ##
>##           •       ##

Any way to achieve this?

Answer 1

The best I can think of is:

function formatwidth
{
  local STR=$1; shift
  local WIDTH=$1; shift
  local BYTEWIDTH=$( echo -n "$STR" | wc -c )
  local CHARWIDTH=$( echo -n "$STR" | wc -m )
  echo $(( $WIDTH + $BYTEWIDTH - $CHARWIDTH ))
}

printf "## %5s %*s %5s ##\n## %5s %*s %5s ##\n" \
    '' $( formatwidth "*" 5 ) '*' '' \
    '' $( formatwidth "•" 5 ) "•" ''

You use the * width specifier to take the width as an argument, and calculate the width you need by adding the number of additional bytes in multibyte characters.

Note that in GNU wc, -c returns bytes, and -m returns (possibly multibyte) characters.

Answer 2

I will probably use GNU awk:

awk 'BEGIN{ printf "## %5s %5s %5s ##\n## %5s %5s %5s ##\n", "", "*", "", "", "•", "" }'
##           *       ##
##           •       ##

You can even write shell wrapper function called printf on top of awk to keep same interface:

tr2awk() { 
    FMT="$1"
    echo -n "gawk 'BEGIN{ printf \"$FMT\""
    shift
    for ARG in "$@"
        do echo -n ", \"$ARG\""
    done
    echo " }'"
}

and then override printf with simple function:

printf() { eval `tr2awk "$@"`; }

Test it:

# buggy printf binary test:
/usr/bin/printf "## %5s %5s %5s ##\n## %5s %5s %5s ##\n" '' '*' '' '' "•" ''
##           *       ##
##         •       ##
# buggy printf shell builin test:
builtin printf "## %5s %5s %5s ##\n## %5s %5s %5s ##\n" '' '*' '' '' "•" ''
##           *       ##
##         •       ##

# fixed printf function test:
printf "## %5s %5s %5s ##\n## %5s %5s %5s ##\n" '' '*' '' '' "•" ''
##           *       ##
##           •       ##