php-ml/src/Tokenization/WhitespaceTokenizer.php
2018-10-28 07:44:52 +01:00

21 lines
462 B
PHP

<?php
declare(strict_types=1);
namespace Phpml\Tokenization;
use Phpml\Exception\InvalidArgumentException;
class WhitespaceTokenizer implements Tokenizer
{
public function tokenize(string $text): array
{
$substrings = preg_split('/[\pZ\pC]+/u', $text, -1, PREG_SPLIT_NO_EMPTY);
if ($substrings === false) {
throw new InvalidArgumentException('preg_split failed on: '.$text);
}
return $substrings;
}
}