how to capture every single match of a repeated pattern?

Question

I have the following regexp:

/xxx ([a-z]+)(?:, ([a-z]+))* xxx/

I want to capture all colors in the following test string:

xxx red, blue, pink, purple xxx

(now only red and purple get captured)

open this url and see the matched groups: http://www.regex101.com/r/oZ2cH4

I have read the following http://www.regular-expressions.info/captureall.html but the trick didn't work

(or maybe I did it wrong)

how can I resolve this?

thank you in advance

Match against `/xxx ([a-z]+(?:, [a-z]+)*) xxx/`, take the first capturing group, then split on `, ` — nhahtdh, May 12 '13 at 05:01
This answer demonstrate the idea http://stackoverflow.com/a/15922245/1400768. The other way with regex alone is very monstrous and I wouldn't recommend it - it is only for educational purpose: http://stackoverflow.com/a/15418942/1400768 — nhahtdh, May 12 '13 at 05:05

score 3 · Accepted Answer · answered May 12 '13 at 09:28

You probably want to return a matching group based on a previous pattern matches:

$word = '[a-z]+';
$sep  = '[, ]+';

$words = $captures("~($word)(?:{$sep})?~");
$of    = $captures("~xxx ({$word}(?:{$sep}{$word})*) xxx~");

print_r($words($of($subject)));

Output:

Array
(
    [0] => red
    [1] => blue
    [2] => pink
    [3] => purple
)

Whereas $captures is a function that return a pre-configured preg_match_all call allowing to process not only a string as subject but anything foreach can operate on:

$captures = function ($pattern, $group = 1) {
    return function ($subject) use ($pattern, $group) {
        if (is_string($subject)) {
            $subject = (array)$subject;
        }
        $captures = [];
        foreach ($subject as $step) {
            preg_match_all($pattern, $step, $matches);
            $captures = array_merge($captures, $matches[$group]);
        }
        return $captures;
    };
};

By default and as used in the example above, it returns the first group (1), but this can be configured.

This allows to first match the outer pattern ($of) and then on each of those matches the inner pattern ($words). The example in full:

$subject = '/xxx red, blue, pink, purple xxx/';

$captures = function ($pattern, $group = 1) {
    return function ($subject) use ($pattern, $group) {
        if (is_string($subject)) {
            $subject = (array)$subject;
        }
        $captures = [];
        foreach ($subject as $step) {
            preg_match_all($pattern, $step, $matches);
            $captures = array_merge($captures, $matches[$group]);
        }
        return $captures;
    };
};

$word = '[a-z]+';
$sep  = '[, ]+';
$seq  = "";

$words = $captures("~($word)(?:{$sep})?~");
$of    = $captures("~xxx ({$word}(?:{$sep}{$word})*) xxx~");

print_r($words($of($subject)));

See the live-demo.

Hmm, interesting. You create a high-order function to process the array of strings/string. — nhahtdh, May 12 '13 at 09:54

score 0 · Answer 2 · edited May 12 '13 at 08:18

0

The tutorial "Repeating a Capturing Group vs. Capturing a Repeated Group" (by regular-expressions.info) describes how you would capture all of the content "red, blue, pink, purple" in a single capture. The pattern it would suggest is

/xxx ((?:[a-z]+(?:, )?)+) xxx/

but if this were really what you were trying to accomplish, you may as well use the simpler expression

/xxx ([a-z, ]*) xxx/

I suspect what you actually want is to capture each color individually. This might be best accomplished by capturing the entire list once, then parsing that captured content.

edited May 12 '13 at 08:18

hakre

193,403
52
435
836

answered May 12 '13 at 05:03

Clark

890
8
20

2

It is a bad idea to repeat the separator along with the tokens `/xxx ((?:[a-z]+(?:, )?)+) xxx/`. If you are not careful (like in this case), you will induce backtracking hell on invalid input: http://www.regex101.com/r/bS5mG3 – nhahtdh May 12 '13 at 05:11

how to capture every single match of a repeated pattern?

2 Answers2