FeatureBuilder::clearText

static function clearText($input, $maxWords = 0)
{
	//$input = Encoding::convertEncoding($input, "utf8", "cp1251");
	$result = HTMLToTxt($input);

	// strip BBCode
	$result = preg_replace('/[[\/\!]*?[^\[\]]*?]/si', ' ', $result);

	// strip punctuation
	$result = preg_replace("/[[:punct:]]/", ' ', $result);

	// replace multiple spaces with single one
	$result = preg_replace("/[[:space:]]+/", ' ', $result);

	// remove short words
	$words = explode(" ", $result);
	$words = array_filter($words, function($word) {return mb_strlen($word) > 3 && \Bitrix\Main\Text\UtfSafeString::checkEncoding($word);});

	if($maxWords > 0)
	{
		$words = array_slice($words, 0, $maxWords);
	}

	return join(" ", $words);
}