using regex to skip ahead all characters until a specific sequence of letters is found using negative lookahead

Question

I'm alright with basic regular expressions, but I get a bit lost around pos/neg look aheads/behinds.

I'm trying to pull the id # from this:

[keyword stuff=otherstuff id=123 morestuff=stuff]

There could be unlimited amounts of "stuff" before or after. I've been using The Regex Coach to help debug what I've tried, but I'm not moving forward anymore...

So far I have this:

\[keyword (?:id=([0-9]+))?[^\]]*\]

Which takes care of any extra attributes after the id, but I can't figure out how to ignore everything between keyword and id. I know I can't go [^id]* I believe I need to use a negative lookahead like this (?!id)* but I guess since it's zero-width, it doesn't move forward from there. This doesn't work either:

\[keyword[A-z0-9 =]*(?!id)(?:id=([0-9]+))?[^\]]*\]

I've been looking all over for examples, but haven't found any. Or perhaps I have, but they went so far over my head I didn't even realize what they were.

Help! Thanks.

EDIT: It has to match [keyword stuff=otherstuff] as well, where id= doesn't exist at all, so I have to have a 1 or 0 on the id # group. There are also other [otherkeywords id=32] which I do not want to match. The document needs to match multiple [keyword id=3] throughout the documents using preg_match_all.

Answer 1

No lookahead/behind required:

/\[keyword(?:[^\]]*?\bid=([0-9]+))?[^\]]*?\]/

Added the ending '[^]]*]' to check for a real tag end, could be unnecessary.

Edit: added the \b to id as otherwise it could match [keyword you-dont-want-this-guid=123123-132123-123 id=123]

$ php -r 'preg_match_all("/\[keyword(?:[^\]]*?\bid=([0-9]+))?[^\]]*?\]/","[keyword stuff=otherstuff morestuff=stuff]",$matches);var_dump($matches);'
array(2) {
  [0]=>
  array(1) {
    [0]=>
    string(42) "[keyword stuff=otherstuff morestuff=stuff]"
  }
  [1]=>
  array(1) {
    [0]=>
    string(0) ""
  }
}
$ php -r 'var_dump(preg_match_all("/\[keyword(?:[^\]]*?\bid=([0-9]+))?[^\]]*?\]/","[keyword stuff=otherstuff id=123 morestuff=stuff]",$matches),$matches);'
int(1)
array(2) {
  [0]=>
  array(1) {
    [0]=>
    string(49) "[keyword stuff=otherstuff id=123 morestuff=stuff]"
  }
  [1]=>
  array(1) {
    [0]=>
    string(3) "123"
  }
}

Answer 2

You do not need look ahead / behind.

Since the question is tagged PHP, use preg_match_all() and store the match in $matches.

Here's how:

<?php

  // Store the string. I single quote, in case there are backslashes I
  // didn't see.
$string = 'blah blah[keyword stuff=otherstuff id=123 morestuff=stuff]
           blah blah[otherkeyword stuff=otherstuff id=555 morestuff=stuff]
           blah blah[keyword stuff=otherstuff id=444 morestuff=stuff]';

  // The pattern is '[keyword' followed by not ']' a space and id
  // The space before id is important, so you don't catch 'guid', etc.
  // If '[keyword'  is always at the beginning of a line, you can use
  // '^\[keyword'
$pattern = '/\[keyword[^\]]* id=([0-9]+)/';

  // Find every single $pattern in $string and store it in $matches
preg_match_all($pattern, $string, $matches);

  // The only tricky part you have to know is that each entire match is stored in
  // $matches[0][x], and the part of the match in the parentheses, which is what
  // you want is stored in $matches[1][x]. The brackets are optional, since it's
  // only one line.
foreach($matches[1] as $value)
{     
    echo $value . "<br/>";
}
?>

Output:

123
444   

( 555 is skipped, as it should be)

PS

You can also use \b instead of a literal space if there could be a tab instead. \b represents a word boundary... in this case the beginning of a word.

$pattern = '/\[keyword[^\]]*\bid=([0-9]+)/';

Answer 3

I think this is what you're getting at:

\[keyword(?:\s+(?!id\b)[A-Za-z]+=[^\]\s]+)*(?:\s+id=([0-9]+))?[^\]]*\]

(I'm assuming attribute names can only contain ASCII letters, while the values can contain any non-whitespace character except ].)

(?:\s+(?!id\b)[A-Za-z]+=[^\]\s]+)* matches any number of attribute=value pairs (and the whitespace preceding them), as long as the attribute name isn't id. The \b (word boundary) is there just in case there are attribute names that start with id, like idiocy. There's no need to put a \b in front of the attribute name this time, because you know any name it matches will be preceded by whitespace. But, as you've learned, the lookahead approach is overkill in this case.

Now, about this:

[A-z0-9 =]

That A-z is either a typo or an error. If you're expecting it to match all uppercase and lowercase letters, well, it does. But it also matches

'[', ']', '^', '_', '`` and '\'

...because their code points lie between those of the uppercase letters and the lowercase letters. ASCII letters, that is.