[Solved] Regular expression match against a string (blue:bar AND darkblue:foo) to darkblue

Question

These are some patterns that I have written that work (in order of least efficient to most efficient in terms of “step count”):

Step counts are based on this sample input:

(green:bar AND black:foo)
(blue:bar AND darkblue:foo)
(yellow:bar AND grey:foo)
(greengarden:bar AND red:foo)

/(?:red|blue|green)(*SKIP)(*FAIL)|[a-z]+(?=:)/ Demo (513 steps)
/\b(?!red:|blue:|green:)[a-z]+(?=:)/ Demo (372 steps)
/(?<=\(|AND )(?!red:|blue:|green:)[^:]*/ Demo (319 steps)
/(?<=\(|AND )(?:(?:red:|blue:|green:)(*SKIP)(*FAIL)|[^:]+)/ Demo (304 steps)
/(?:\(|AND )\K(?!red:|blue:|green:)[^:]*/ Demo (291 steps)
/(?:\(|AND )\K(?!red\b|blue\b|green\b)[^:]+/ Demo (291 steps)
/[( ]\K(?!red\b|blue\b|green\b)[a-z]+/ Demo (172 steps)

This final pattern is the best performer and takes full advantage of the strict format of your input data — pertaining to capitalization, opening parentheses, and the two spaces per line.

It finds the opening parenthesis or a space, then restarts the fullstring match using \K, disqualifies any substring that is wholly red, blue, or green and stops matching on the last lowercase letter.

In regex, speed gains are accomplished by using narrow character classes ([] & [^]), limiting alternatives (pipes |), “lookarounds”, and capture/non-capture groups.

My patterns deliberately avoid making capture groups because they only lead to output array bloat. All of your desired “color” substrings will be found in the fullstring [0] matches subarray that preg_match_all() returns.

Code: (Demo)

$string='(green:bar AND black:foo)
    (blue:bar AND darkblue:foo)
    (yellow:bar AND grey:foo)
    (greengarden:bar AND red:foo)';

var_export(preg_match_all('/[( ]\K(?!red\b|blue\b|green\b)[a-z]+/',$string,$out)?$out[0]:'fail');

Output:

array (
  0 => 'black',
  1 => 'darkblue',
  2 => 'yellow',
  3 => 'grey',
  4 => 'greengarden',
)

Accepted Answer