/**
* breaks up the text into words
* @param string $text
* @return string
*/
protected static function break_words($text)
{
$text = strip_tags($text);
$text = strip_bbcode($text, true, false, false, true);
// multibyte
$is_mb = preg_match('/[^\\x00-\\x7F]/', $text);
if (!$is_mb) {
$is_mb = strlen($text) != vB_String::vbStrlen($text);
}
if ($is_mb) {
$text = vB_String::toUtf8($text, vB_String::getCharSet());
if (preg_match('/&#([0-9]+);|[^\\x00-\\x7F]/esiU', $text) and function_exists('mb_decode_numericentity')) {
$text = mb_decode_numericentity($text, array(0x0, 0x2ffff, 0, 0xffff), 'UTF-8');
}
$pattern = '/[\\s,.!?@#$%^&*\\(\\)\\/<>"\';:\\[\\]\\{\\}\\+|-]/';
} else {
$pattern = '/[^a-z0-9_]+/i';
}
$words = preg_split($pattern, vB_String::vBStrToLower($text), -1, PREG_SPLIT_NO_EMPTY);
foreach ($words as $index => $word) {
if (!vB_Api_Search::is_index_word($word, true)) {
unset($words[$index]);
continue;
}
//$words[$index] = $is_mb ? vB_String::toUtf8($word) : $word;
if (empty($words[$index])) {
unset($words[$index]);
continue;
}
}
return $words;
}
