Divide a string that contains scores

If they were just spaces, it would be a case of

$partes = explode( ' ', $todo );

One solution, depending on what you want, would be to force a space before the characters you want to treat as isolates:

$todo = str_replace( array( '.', ',' ,'?' ), array( ' .', ' ,', ' ?'), $todo );
$partes = explode( ' ', $todo );

See working at IDEONE .

Note that I've placed valid separators directly in replace, but if you want to do this with a series of characters, it makes up for a more complex function.

If you prefer to consider all of the alphanumeric characters separated from the symbols, you can use a RegEx , and solve on one line:

preg_match_all('~\w+|[^\s\w]+~u', $todo, $partes ); 

See working at IDEONE .

In addition, it would be the case to add spaces before and after the symbols, remove double spaces, depends on the criteria. The intention of the response was to take an initial turn.

A more general approach would be to use regex to solve the problem.

$string = "Eu irei amanhã à casa. E tu vens?";

/*
    Adiciona um espaço em todos os boundaries da string
    Ex.: Início e fim de palavras, pontuações, etc...
    O modificador 'u' e para tratar a string como Unicode
*/
$resultado = preg_replace('/\b/u', ' ', $string);

// Cria um array usando como delimitador um regex que casa com qualquer espaço
$resultado = preg_split('/\s+/', trim($resultado));

var_dump($resultado);

output :

array(10) {
  [0]=>
  string(2) "Eu"
  [1]=>
  string(4) "irei"
  [2]=>
  string(7) "amanhã"
  [3]=>
  string(2) "à"
  [4]=>
  string(4) "casa"
  [5]=>
  string(1) "."
  [6]=>
  string(1) "E"
  [7]=>
  string(2) "tu"
  [8]=>
  string(4) "vens"
  [9]=>
  string(1) "?"
}

I made this sample code to illustrate.

Based on 99.9% of @bacco's response ... blatantly!

preg_match_all('~\b\w[\w\-.*#]*\w\b|\w|\.\.\.|[,.:;()[\]?!]|\S~u', $t, $ps);
print_r($ps)

( [0] => Array
   (    [0] => Baseando-me
        [1] => 99.9
        [2] => %
        [3] => na
        [4] => resposta
        [5] => do
        [6] => @
        [7] => bacco
        [8] => ...
        [9] => descaradamente
        [10] => !
    )
)

Upgrade Actually tokenize text in your elements is sometimes complex: text is not just simple words ...

A somewhat more robust approach (employing% of better readability):

  preg_match_all('~
           https?://\S+             ## url
         | \d+/\d+/\d+              ## data
         | \b\w [\w\-.*#]* \w\b     ## vou-me 12.2 f.html
         | \w
         | \.\.\.                   ## ...
         | [!?]+                    ## ???   ?!
         | [,.:;()[\]]
         | \S
         ~ux', $todo, $partes );
     print_r($partes)

Danger: (not tested ...)