Regex convert string to valid number, no decimal, sign optional

Question

I want to remove everything after and including the decimal, and everything non-numerical except the hyphen if it's the first character. So far I have this: /[^0-9^-]|[^\.]+$/. Notice how I block dashes from being removed with ^-, because somehow I want to only remove the dashes that aren't the first character (not the sign). Any help? Thanks.

I just want it to remove

• Any non 0-9 characters, except the the first character if it is a dash (negative sign)
• Everything after and including the decimal point

Ex.: 10js-_67.09090FD => 1067
-10a.h96 => -10

EDIT: Never mind, I was approaching this the wrong way, trying to match the characters that don't belong, and I realized I shouldn't be using a regex for this. Thanks for your answers though, I learned a bit about regex and maybe someone else with a similar problem will find this.

Answer 1

Try this:

Regex numbers = new Regex(@"^(-?\d*)[^0-9]*(\d*)\.", 
    RegexOptions.ECMAScript | RegexOptions.Multiline);
foreach (Match number in numbers.Matches("10js-_67.09090FD"))
{
    Console.WriteLine(
        Int32.Parse(
            number.Groups[1].Value + 
            number.Groups[2].Value));
}

Or this one:

Console.WriteLine(
    Int32.Parse(
        Regex.Replace(
            "10js-_67.09090FD", 
            @"^(-?\d*)[^0-9]*(\d*)\.([\s\S]*?)$", "$1$2", 
            RegexOptions.ECMAScript | RegexOptions.Multiline)));

Or this one:

var re = /^(-?\d*)[^0-9]*(\d*)\.([\s\S]*?)$/
alert(parseInt("10js-_67.09090FD".replace(re, "$1$2"),10));

Answer 2

That would be /^(-?[0-9]+)[^0-9\.]*([0-9]*).*$/\1\2/ (use for sed as you don't tell me what language ar you using).

/^(-?[0-9]+)[^0-9\.]*([0-9]*).*$/
// '^'          ==>l From the Start
// '(..)'       ==>l Group 1
//     '-?'     ==>l An optiona '-'
//     '[0-9]+' ==>l Some numbers
// '[^0-9\.]*'  ==>l Anything but numbers and dot
// '(..)'       ==>l Group 2 (So this is the number after the dot)
//     '[0-9]*' ==>l Some numbers
// '.*$'        ==>l The rest

Then Only print Group 1 and Group 2 (/\1\2/).

Tests:

$:~/Desktop$ echo "10js-_67.09090FD" | sed -r "s/^(-?[0-9]+)[^0-9\.]*([0-9]*).*$/\1\2/"
1067
$:~/Desktop$ echo "-10a.h96" | sed -r "s/^(-?[0-9]+)[^0-9\.]*([0-9]*).*$/\1\2/"
-10

Hope this helps