voku / portable-utf8

   *

   * @return string

   */

  public static function fix_simple_utf8($str)

  {

    // init

    $str = (string)$str;

    if (!isset($str[0])) {

      return '';

    }

    static $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = null;

    static $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = null;

    if ($BROKEN_UTF8_TO_UTF8_KEYS_CACHE === null) {

      $BROKEN_UTF8_TO_UTF8_KEYS_CACHE = array_keys(self::$BROKEN_UTF8_FIX);

      $BROKEN_UTF8_TO_UTF8_VALUES_CACHE = array_values(self::$BROKEN_UTF8_FIX);

    }

    return str_replace($BROKEN_UTF8_TO_UTF8_KEYS_CACHE, $BROKEN_UTF8_TO_UTF8_VALUES_CACHE, $str);

  }

  /**

   * Fix a double (or multiple) encoded UTF8 string.

   *

   * @return string

   */

  public static function normalize_msword($str)

  {

    $str = (string)$str;

    if (!isset($str[0])) {

      return '';

    }

    static $UTF8_MSWORD_KEYS_CACHE = null;

    static $UTF8_MSWORD_VALUES_CACHE = null;

    if ($UTF8_MSWORD_KEYS_CACHE === null) {

      $UTF8_MSWORD_KEYS_CACHE = array_keys(self::$UTF8_MSWORD);

      $UTF8_MSWORD_VALUES_CACHE = array_values(self::$UTF8_MSWORD);

    }

    return str_replace($UTF8_MSWORD_KEYS_CACHE, $UTF8_MSWORD_VALUES_CACHE, $str);

  }

  /**

   * Normalize the whitespace.

   *