* The ultimate convenience function! Clean up invalid UTF-8 sequences,

* and convert to Normal Form C, canonical composition.

*

* @return string The same string, all shiny and cleaned-up

*/

{

// The string below is the list of all autorized characters, sorted by frequency in latin text

$pos = strspn($str, "\x20\x65\x69\x61\x73\x6E\x74\x72\x6F\x6C\x75\x64\x5D\x5B\x63\x6D\x70\x27\x0A\x67\x7C\x68\x76\x2E\x66\x62\x2C\x3A\x3D\x2D\x71\x31\x30\x43\x32\x2A\x79\x78\x29\x28\x4C\x39\x41\x53\x2F\x50\x22\x45\x6A\x4D\x49\x6B\x33\x3E\x35\x54\x3C\x44\x34\x7D\x42\x7B\x38\x46\x77\x52\x36\x37\x55\x47\x4E\x3B\x4A\x7A\x56\x23\x48\x4F\x57\x5F\x26\x21\x4B\x3F\x58\x51\x25\x59\x5C\x09\x5A\x2B\x7E\x5E\x24\x40\x60\x7F\x0D");

if ($pos == $len)

{

// ASCII strings with no special chars return immediately

}

// Note: we do not check for $GLOBALS['utf_canonical_decomp']. It is assumed they are always loaded together

// Replace any byte in the range 0x00..0x1F, except for \r, \n and \t

// We replace those characters with a 0xFF byte, which is illegal in UTF-8 and will in turn be replaced with a UTF replacement char

);

}

/**

* Validate and normalize a UTF string to NFC

*

* @return string The string, validated and in normal form

*/

{

$pos = strspn($str, UTF8_ASCII_RANGE);

$len = strlen($str);

if ($pos == $len)

{

// ASCII strings return immediately

}

if (!isset($GLOBALS['utf_nfc_qc']))

include($phpbb_root_path . 'includes/utf/data/utf_nfc_qc.' . $phpEx);

}

}

/**

* Validate and normalize a UTF string to NFKC

*

* @return string The string, validated and in normal form

*/

{

$pos = strspn($str, UTF8_ASCII_RANGE);

$len = strlen($str);

if ($pos == $len)

{

// ASCII strings return immediately

}

if (!isset($GLOBALS['utf_nfkc_qc']))

include($phpbb_root_path . 'includes/utf/data/utf_canonical_comp.' . $phpEx);

}

}

/**

* Validate and normalize a UTF string to NFD

*

* @return string The string, validated and in normal form

*/

{

$pos = strspn($str, UTF8_ASCII_RANGE);

$len = strlen($str);

if ($pos == $len)

{

// ASCII strings return immediately

}

if (!isset($GLOBALS['utf_canonical_decomp']))

include($phpbb_root_path . 'includes/utf/data/utf_canonical_decomp.' . $phpEx);

}

}

/**

* Validate and normalize a UTF string to NFKD

*

* @return string The string, validated and in normal form

*/

{

$pos = strspn($str, UTF8_ASCII_RANGE);

$len = strlen($str);

if ($pos == $len)

{

// ASCII strings return immediately

}

if (!isset($GLOBALS['utf_compatibility_decomp']))

include($phpbb_root_path . 'includes/utf/data/utf_compatibility_decomp.' . $phpEx);

}

}

/**

* Recompose a UTF string

*

*

* @access private

*/

$tmp = '';

$i = $tmp_pos = $last_cc = 0;

// UTF char length array

// This array is used to determine the length of a UTF character.

// has been encoded in a five- or six- byte sequence

if ($utf_char[0] >= "\xF8")

{

{

$trailing_bytes = 4;

}

{

$trailing_bytes = 0;

}

/**

* Decompose a UTF string

*

*

* @access private

*/

function decompose($str, $pos, $len, &$decomp_map)

{

// Load some commonly-used tables

if (!isset($utf_combining_class))

ksort($utf_sort);

}

{

$tmp .= implode('', $utf_chars);

}

// LIndex can only range from 0 to 18, therefore it cannot influence the first two bytes of the L Jamo, which allows us to hardcode them (based on LBase).

//

// The same goes for VIndex, but for TIndex there's a catch: the value of the third byte could exceed 0xBF and we would have to increment the second byte

{

{

$utf_char = "\xE1\x84\x00\xE1\x85\x00\xE1\x86\x00";

}

else

{

$utf_char = "\xE1\x84\x00\xE1\x85\x00\xE1\x87\x00";

}

}

else

}

return $tmp;

}

else if ($tmp_pos)

{