Question

I want to try to filter all operators and operands including floating points
from a commandline input with regexp. the operators work fine and the floating point works when i use it with a seperate matcher but i want to solve this in one regexp term
my code till now does

    Matcher numbers = Pattern.compile("[0-9]*\\.?[0-9]+").matcher(expr);
    Matcher m = Pattern.compile("[\\+|\\*|/|\\-|\\^|\\!|_|([0-9]*\\.?[0-9]+)]").matcher(expr);

    while(numbers.find()) {
            System.out.print(" " + numbers.group() + " num \n"); 
    }

    while (m.find()) {
       System.out.print(" " + m.group() + " -- \n");
    }

heres the output

3.0 num
2 num

3 —
. —
0 —
2 —

as you see the numbers matcher just finds floating points and normal numbers and that works fine. but if i use it in the second matcher devided by an or
i get 3 . 0 as separate matches.

Answer 1

IIRC Java regexes use parentheses for a grouping operator, not square brackets like you’ve got there. Have you tried

"(\\+|\\*|/|\\-|\\^|\\!|_|([0-9]*\\.?[0-9]+))"

instead? Because what you’ve got is just a large, multiply redundant character set: It should merely match any one of the characters between the first left square bracket and the final right square bracket. That seems to be the behavior you’re seeing.

Or did I completely misunderstand the intent of the second regex?