Replacing all the numbers with their currency format in given line of text

Question

I want to write a Shell (AWK, Sed also fine) program to take as a input a SINGLE line of text.

Which will have arbitrarily spread integer strings in it. e.g

"12884 and 111933 are two numbers and 323232 is also a number"

I want the output to be

"12,884 and 1,11,933 are two numbers and 2,23,232 is also a number"

If this was PHP a simple preg_replace_callback would have served the purpose but I want this to be in shell. :/

Any pointers would of great help.

Answer 1

for line in $(echo "12884 and 111933 are two numbers and 323232 is also a number" 
    | tr ' ' '\n');
    do 
        if [ $line -eq $line 2> /dev/null ]; 
            then printf "%'d\n" $line;
        else 
            echo $line; 
        fi; 
    done | tr '\n' ' '

I understand this may be long and ugly but by now is the best I could do, I hope it will help.

Follows the explanation:

• fist I split the line on more lines so I can loop and recognize which strings are number and which are not
• then I test if the current string is a number
• if it is a number I parse with the usage of printf
• if it is not I simply echo it, leaving as it was
• finish the loop and put everything back on one line

Answer 2

It's not necessary to split the line using tr. You can make use of Bash's word splitting feature:

line="12884 and 111933 are two numbers and 323232 is also a number"
for word in $line
do
    if [[ $word = *[^0-9]* ]]
    then
        printf "%s " $word
    else
        printf "%'d " $word
    fi
done

I've also used globbing to test for a sequence of digits rather than relying on something that creates an error or not depending on whether it's an integer.