How to exclude string from preg_split when string already contains splittable character in PHP?

Question

I am using the preg_split function in PHP in order to create one array containing several different elements. However, I want to exclude a string which happens to contain one of the elements that I'm preg_splitting by.

$array['stuff'] = preg_split('/\[#]|\ &amp  |\ &amp |\&amp |\&amp|\ &amp|\ &gt  |\ &gt |\&gt |\&gt|\ &gt|\ &  |\ & |\& |\&|\ &|\ \/  |\ \/ |\\/ |\\/|\ \/|\ >  |\ > |\> |\>|\ >|\ ,  |\ , |\, |\,|\, |\ ::  |\ :: |\:: |\ ::|\::|\ ::|\ :  |\ : |\: |\:|\ :|\ -  |\ - |\- |\-|\ -/', $array['stuff'] ) ;

What I would like to do is to exclude a string such as 'foo-bar' from being matched for a split because it contains a dash. 'foo-bar' would need to be an exact match for my purposes.

Answer 1

The resulting regular expression would be very complicated specially if you have a lot exceptions like 'foo-bar'.

You should use a conditional subpattern with a lookbehind as condition and a lookahead as its yes-pattern:

$res = preg_split('/(?(?<=foo)\-(?!bar)|\-)/', 'aasdf-fafsdf-foo-bar-asdf' );
var_dump( $res );

result:

array(4) {
  [0]=>
  string(5) "aasdf"
  [1]=>
  string(6) "fafsdf"
  [2]=>
  string(7) "foo-bar"
  [3]=>
  string(4) "asdf"
}

---

Let me explain what is happening here. \- means

Match any dash character.

but what we want is

Match any dash character that is not part of foo-bar.

Since we can't implement that in regex as it is we change it a little:

Match any dash character that if preceded by foo is not followed by bar.

To implement the if part we use a conditional subpattern, this is the syntax:

(?(condition)yes-pattern|no-pattern)

Our "condition" would be "preceded by foo" to check for that we use a lookbehind:

(?<=foo)

If that is true we should look for "a dash that is not followed by bar" to do that we use a negative lookahead:

\-(?!bar)

And that becomes our "yes-pattern". Our "no-pattern" should be \- or "any dash". The complete regex would be:

(?(?<=foo)\-(?!bar)|\-)

---

UPDATE: to incorporate this into your current regex change this part at the end:

|\ -  |\ - |\- |\-|\ -/

to

|\s?(?(?<=foo)\-(?!bar)|\-)\s?/

Answer 2

Though I make no guarantee that my solution is more efficient than nobody's double lookaround pattern for this case, I think my solution is slightly easier to read. (*SKIP)(*FAIL) effectively matches and discards the substrings that you wish to ignore. In some cases, this approach can be very useful/effective/maintainable.

Code: (Demo)

$string = 'I-like-candy-and-foo-bar-sandwiches';
var_export(preg_split('~foo-bar(*SKIP)(*FAIL)|-~', $string));

Output:

array (
  0 => 'I',
  1 => 'like',
  2 => 'candy',
  3 => 'and',
  4 => 'foo-bar',
  5 => 'sandwiches',
)

---

To be completely honest, I think nobody's answer is a bit over-engineered. It can be more simply written as a negated lookbehind and a negated lookahead ...no reason for the conditional syntax.

Code: (Demo)

$string = 'I-like-candy-and-foo-bar-sandwiches';
var_export(preg_split('~(?<!foo)-(?!bar)~', $string));

Output:

array (
  0 => 'I',
  1 => 'like',
  2 => 'candy',
  3 => 'and',
  4 => 'foo-bar',
  5 => 'sandwiches',
)

p.s. If you might have a hyphen at the start or end of your input string AND you don't want empty elements to be generated by preg_split(), then use 0 and PREG_SPLIT_NO_EMPTY as parameters 3 and 4 (respectively) in the function call.