How to handle strings that contain negative numbers using regex-CodePudding

I am using a simple regex [^0-9.] that doesn't match any numbers between 0 to 9 and a period in my given string and matches the rest, and I replace that with empty string and then convert my valid value (integer or double) in string format into an integer or double.

For example:

'123A.478' => '123.478'
'123.48' => '123.48'
'123AX' => '123'

I want to also handle negative valued strings as well i.e. if a string has '-123', I want to retain it. So, I need to convert the strings as follows:

'--123.46' => '-123.46',
'123A-.46' => '123.46',
'-123--.46' => '-123.46', 
'A-123-.46' => '-123.46'

I tried using a quantifier but I was unable to build a correct regex with my existing one.

Is there any way I can achieve this using regex?

CodePudding user response：

You can use

preg_replace('~^(-)|- ~', '$1', preg_replace('~[^\d.-] ~', '', $s))

See the PHP demo:

<?php

$strings = ['--123.46', '123A-.46', '-123--.46', 'A-123-.46'];
foreach ($strings as $s) {
    echo $s . " => " . preg_replace('~^(-)|- ~', '$1', preg_replace('~[^\d.-] ~', '', $s)) . PHP_EOL;
}

Output:

--123.46 => -123.46
123A-.46 => 123.46
-123--.46 => -123.46
A-123-.46 => -123.46

There are two steps:

All occurrences of the [^\d.-] pattern (any char other than a digit, dot or hyphen) are removed from the input string and then
The ^(-)|- pattern occurrences (the first char in the string that is a hyphen is placed into Capturing group 1 and all other hyphens are just matched) are replaced with $1, Group 1 value (so, all hyphens are removed but the first one in the string).

CodePudding user response：

You might use an alternation to remove the parts from the example strings, and in the replacement use an empty string.

^[^\d\s] (?=-)|[^\d\s.] (?!\d)

Explanation

^ Start of string
[^\d\s] Match 1 chars other than a digit or a whitespace char
(?=-) Assert a - directly to the right
| Or
[^\d\s.] Match 1 chars other than a digit, whitespace char or dot
(?!\d) Assert not a digit directly to the right

See a regex demo and a PHP demo.

Example code

$strings = [
    "123A.478",
    "123.48",
    "123AX",
    "--123.46",
    "123A-.46",
    "-123--.46",
    "A-123-.46"
];

$pattern = '/^[^\d\s] (?=-)|[^\d\s.] (?!\d)/';

foreach ($strings as $str) {
    echo preg_replace($pattern, "", $str) . PHP_EOL;
}

Output