*

* @package phpBB3

* @version $Id$

* @license http://opensource.org/licenses/gpl-license.php GNU Public License

*

*/

/**

* UTF-8 tools

*

* Whenever possible, these functions will try to use PHP's built-in functions or

if (!extension_loaded('xml'))

{

/**

function utf8_encode($str)

{

$out = '';

}

/**

function utf8_decode($string)

{

$newcharstring = '';

$offset = 0;

$stringlength = strlen($string);

while ($offset < $stringlength)

{

$ord = ord($string{$offset});

$charval = false;

$offset += 1;

}

if ($charval !== false)

{

$newcharstring .= (($charval < 256) ? chr($charval) : '?');

}

}

return $newcharstring;

}

}

{

return false;

}

return mb_strrpos($str, $search);

}

else

* @param string needle

* @param integer offset in characters (from left)

* @return mixed integer position or FALSE on failure

function utf8_strpos($str, $needle, $offset = null)

{

if ($offset === false)

{

$ar = explode($needle, $str);

{

// Pop off the end of the string where the last match was made

array_pop($ar);

return utf8_strlen($str);

}

return false;

trigger_error('utf8_strrpos expects parameter 3 to be long', E_USER_WARNING);

return false;

}

$str = utf8_substr($str, $offset);

if (false !== ($pos = utf8_strrpos($str, $needle)))

{

return $pos + $offset;

}

return false;

}

}

* @param string needle

* @param integer offset in characters (from left)

* @return mixed integer position or FALSE on failure

function utf8_strpos($str, $needle, $offset = null)

{

// native

if (is_null($offset))

{

$ar = explode($needle, $str);

{

return utf8_strlen($ar[0]);

}

}

}

/**

* UTF-8 aware alternative to strtolower

function utf8_strtolower($string)

{

global $UTF8_UPPER_TO_LOWER;

$uni = utf8_to_unicode($string);

if (!$uni)

{

return false;

}

{

if (isset($UTF8_UPPER_TO_LOWER[$uni[$i]]))

{

function utf8_strtoupper($str)

{

global $UTF8_LOWER_TO_UPPER;

$uni = utf8_to_unicode($string);

if (!$uni)

{

return false;

}

{

if (isset($UTF8_LOWER_TO_UPPER[$uni[$i]]))

{

trigger_error('utf8_substr expects parameter 3 to be long', E_USER_WARNING);

return false;

}

$strlen = strlen(utf8_decode($str));

if ($offset > $strlen)

{

return '';

}

if (($offset + $length) > $strlen)

{

$length = '*';

$length = '{' . $length . '}';

}

}

if (!preg_match('/^[0-9]+$/', $offset))

{

trigger_error('utf8_substr expects parameter 2 to be long', E_USER_WARNING);

return false;

}

$pattern = '/^.{' . $offset . '}(.' . $length . ')/us';

preg_match($pattern, $str, $matches);

if (isset($matches[1]))

{

return $matches[1];

}

return false;

}

else

{

// From: http://www.php.net/manual/en/function.substr.php#44838

preg_match_all('/./u', $str, $ar);

if ($length !== null)

{

return join('', array_slice($ar[0], $offset, $length));

// Since utf8_decode is replacing multibyte characters to ? strlen works fine

return strlen(utf8_decode($text));

}

}

/**

{

return false;

}

$len = utf8_strlen($str);

if ($len <= $split_len)

{

function utf8_strspn($str, $mask, $start = null, $length = null)

{

$mask = preg_replace('!([\\\\\\-\\]\\[/^])!', '\\\${1}', $mask);

if ($start !== null || $length !== null)

{

$str = utf8_substr($str, $start, $length);

preg_match('/^[' . $mask . ']+/u', $str, $matches);

if (isset($matches[0]))

return utf8_strlen($matches[0]);

}

}

/**

* If the encoding is not supported, the string is returned as-is

*

* @param string $string Original string

* @return string The string, encoded in UTF-8

*/

function utf8_recode($string, $encoding)

return $string;

}

// start with something simple

if ($encoding == 'iso-8859-1')

{

}

/**

function utf8_ord($chr)

{

switch (strlen($chr))

}

/**

function utf8_chr($cp)

{

if ($cp > 0xFFFF)

{

return chr(0xF0 | ($cp >> 18)) . chr(0x80 | (($cp >> 12) & 0x3F)) . chr(0x80 | (($cp >> 6) & 0x3F)) . chr(0x80 | ($cp & 0x3F));

}

{

return chr(0xE0 | ($cp >> 12)) . chr(0x80 | (($cp >> 6) & 0x3F)) . chr(0x80 | ($cp & 0x3F));

}

{

return chr(0xC0 | ($cp >> 6)) . chr(0x80 | ($cp & 0x3F));

}

}

/**

function utf8_to_unicode($string)

{

$unicode = array();

$offset = 0;

$stringlength = strlen($string);

while ($offset < $stringlength)

{

$ord = ord($string{$offset});

}

/**

function utf8_from_unicode($array)

{

$str = '';