How can I split a string in PHP at the nth occurrence of a needle?

Question

There must be a fast and efficient way to split a (text) string at the "nth" occurrence of a needle, but I cannot find it. There is a fairly full set of functions

Answer 1

I've edited Galled's function to make it explode after every nth occurrences instead of just the first one.

function split2($string, $needle, $nth) {
  $max = strlen($string);
  $n = 0;
  $arr = array();

  //Loop trough each character
  for ($i = 0; $i < $max; $i++) {

    //if character == needle
    if ($string[$i] == $needle) {
      $n++;
      //Make a string for every n-th needle
      if ($n == $nth) {
        $arr[] = substr($string, $i-$nth, $i);
        $n=0; //reset n for next $nth
      }
      //Include last part of the string
      if(($i+$nth) >= $max) {
        $arr[] = substr($string, $i + 1, $max);
        break;
      }
    }
  }
  return $arr;
}

Answer 2

Personally I'd just split it into an array with explode, and then implode the first n-1 parts as the first half, and implode the remaining number as the second half.

Answer 3

Here's an approach that I would prefer over a regexp solution (see my other answer):

function split_nth($str, $delim, $n)
{
  return array_map(function($p) use ($delim) {
      return implode($delim, $p);
  }, array_chunk(explode($delim, $str), $n));
}

Just call it by:

split_nth("1 2 3 4 5 6", " ", 2);

Output:

array(3) {
  [0]=>
  string(3) "1 2"
  [1]=>
  string(3) "3 4"
  [2]=>
  string(3) "5 6"
}

Answer 4

If your needle will always be one character, use Galled's answer. It's going to be faster by quite a bit. If your $needle is a string, try this. It seems to work fine.

function splitn($string, $needle, $offset)
{
    $newString = $string;
    $totalPos = 0;
    $length = strlen($needle);
    for($i = 0; $i < $offset; $i++)
    {
        $pos = strpos($newString, $needle);

        // If you run out of string before you find all your needles
        if($pos === false)
            return false;
        $newString = substr($newString, $pos + $length);
        $totalPos += $pos + $length;
    }
    return array(substr($string, 0, $totalPos-$length), substr($string, $totalPos));
}

Answer 5

I really like Hamze GhaemPanah's answer for its brevity. However, there's a small bug in it.

In the original code:

$i = $pos = 0;
do {
    $pos = strpos($string, $needle, $pos+1);
} while( $i++ < $nth);

$nth in the do while loop should be replaced with ($nth-1) as it will incorrectly iterate one extra time - setting the $pos to the position of the $nth+1 instance of the needle. Here's an example playground to demonstrate. If this link fails here is the code:

$nth = 2;
$string = "44 E conway ave west horse";
$needle = " ";

echo"======= ORIGINAL =======\n";

$i = $pos = 0;
do {
    $pos = strpos($string, $needle, $pos + 1);
} while( $i++ < $nth);

echo "position: $pos \n";
echo substr($string, 0, $pos) . "\n\n";

/*
    Outputs:

    ======= ORIGINAL =======
    position: 11
    44 E conway
*/

echo"======= FIXED =======\n";

$i = $pos = 0;
do {
    $pos = strpos($string, $needle, $pos + 1);
} while( $i++ < ($nth-1) );

echo "position: $pos \n";
echo substr($string, 0, $pos);

/*
    Outputs:

    ======= FIXED =======
    position: 4
    44 E

*/

That is, when searching for the position of the second instance of our needle, our loop iterates one extra time setting $pos to the position of our third instance of our needle. So, when we split the string on the second instance of our needle - as the OP asked - we get the incorrect substring.

Answer 6

This is ugly, but it seems to work:

$foo = '1 2 3 4 5 6 7 8 9 10 11 12 13 14';

$parts = preg_split('!([^ ]* [^ ]* [^ ]*) !', $foo, -1,
            PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY);

var_dump($parts);

Output:

array(5) {
  [0]=>
  string(5) "1 2 3"
  [1]=>
  string(5) "4 5 6"
  [2]=>
  string(5) "7 8 9"
  [3]=>
  string(8) "10 11 12"
  [4]=>
  string(5) "13 14"
}

Replace the single spaces in the query with a single character you wish to split on. This expression won't work as-is with multiple characters as the delimiter.

This is hard coded for every third space. With a little tweaking, probably could be easily adjusted. Although a str_repeat to build a dynamic expression would work as well.