Question

I’m trying to (in php)

1. Verify that the contents of a textarea are valid
2. Separate them into the tokens for processing

In order to be considered valid, it needs to be a string containing only numbers and letters that form a “code” that is anywhere from 3-6 in length, and the comma used to separate them. I’ve broken this down into something like this:

[A-Za-z0-9]{3,6},

I’m having trouble finishing it though. I want them to be able to separate with either a comma, or a space and comma space, comma space, etc. I only want there to be a comma if there is a following valid token.

For example, the input string:

abe 123, PlE43,54drt , r2344

Should be separated into the following tokens:

‘abe 123’ and ‘PlE43′ and ’54drt’ and ‘r2344’

How can I fix my regex to fit the conditions? (right now I’m having trouble making the comma optional, but if it is there I expect another valid token, as well as using comma and any combination of space before or after as a valid separator)

Answer 1

Assuming you want to collapse all spaces (i.e., both spaces inside a token and spaces between tokens/next to commas should be ignored), you can do it much more simply with some preprocessing.

$input = 'abe 123, PlE43,54drt , r2344';
$input = str_replace(' ', '', $input); // strip all spaces
$tokens = explode(',', $input);
foreach ($tokens as $token) {
    if(!preg_match('/^[A-Za-z0-9]{3,6}$/', $token)) {
        // error
    }
}

This code will also report an error if you have two consecutive commas, or if you end the input string with a comma, because that would generate an empty element in $tokens which does not validate by the 3-to-6 alphanumeric rule.

See it in action.

Update: to preserve the spaces inside tokens, a slight modification would be required:

$input = 'abe 123, PlE43,54drt , r2344';
$tokens = explode(',', $input);
foreach ($tokens as &$token) {
    $token = trim($token);
    if(!preg_match('/^[A-Za-z0-9]{3,6}$/', str_replace(' ', '', $token))) {
        // error
    }
}

Be careful with this though, as it thinks that

 a         b                       42

is one valid token.