<?php

function truncate_to_words($string, $len, $wordsafe = FALSE, $dots = FALSE) {
  static $separators = array(
    0xE1 => array(
      0x9A => array(
        0x80 => TRUE, // U+1680 Ogham Space Mark
      ),
    ),
    0xE2 => array(
      0x80 => array(
        0x80 => TRUE, // U+2000 en quad
        0x81 => TRUE, // U+2001 em quad
        0x82 => TRUE, // U+2002 en space
        0x83 => TRUE, // U+2003 em space
        0x84 => TRUE, // U+2004 three-per-em space
        0x85 => TRUE, // U+2005 four-per-em space
        0x86 => TRUE, // U+2006 six-per-em space
        0x87 => TRUE, // U+2007 figure space
        0x88 => TRUE, // U+2008 punctuation space
        0x89 => TRUE, // U+2009 thin space
        0x8A => TRUE, // U+200A hair space
        0xA8 => TRUE, // U+2028 line separator
        0xA9 => TRUE, // U+2029 paragraph separator
      ),
      0x81 => array(
        0x9F => TRUE, // U+205F medium mathematical space
      ),
    ),
    0xE3 => array(
      0x80 => array(
        0x80 => TRUE, // U+3000 ideographic space
      ),
    ),
  );
  $to_space = $return = '';
  $count = $index = 0;
  while ($count < $len) {
    $c = $string[$index++];
    $o = ord($c);
    if ($o < 0x80) {
      if ($o == 0x20) {
        $to_space = $return;
      }
      $return .= $c;
    }
    elseif ($o >= 0xC2 && $o <= 0xDF) {
      $c1 = $string[$index++];
      $return .= $c;
      $return .= $c1;
    }
    elseif ($o >= 0xE0 && $o <= 0xEF) {
      $c1 = $string[$index++];
      $c2 = $string[$index++];
      if (isset($separators[$o][ord($c1)][ord($c2)])) {
        $to_space = $return;
      }
      $return .= $c;
      $return .= $c1;
      $return .= $c2;
    }
    elseif ($o >= 0xF0 && $o <= 0xF4) {
      $return .= $c;
      $return .= $string[$index++];
      $return .= $string[$index++];
      $return .= $string[$index++];
    }
    $count++;
  }
  return ($wordsafe && $to_space)  ? $to_space : $return;
}

echo truncate_to_words('áéó űúí', 5, TRUE);
echo "!\n";
echo truncate_to_words('áéó űúí', 5, FALSE);
echo "!\n";