source: trunk/Packages/Common/UTF8.php

<?php
/*
  ISO8859-2 <-> UTF-8 conversion functions designed for use in PHP-GTK apps
  Adam Rambousek - rambousek@volny.cz

  version history:
  1.03 --- 12/02/2002
   * added Win1257 support
  1.02 --- 30/11/2001
   * added ISO8859-1 support
  1.01
   * added Win1250 support
  1.00

  string to_utf8(string string [, string charset])
  string from_utf8(string string [, string charset])

  supported charsets: name of charset you must use in script
  ISO8859-2: iso2   (this is the default charset, you don't have to specify it)
  Windows1250: win1250
  ISO8859-1: iso1
  Windows1257: win1257

  example:  $new_string=to_utf8($some_string,"win1250");
*/


/*
  translation table - actually, it's array where key is hexadecimal number of
  character in ISO8859-2/Windows1250 and value is its two byte representation in UTF-8
*/

class Encoding
{
  function __construct()
  {
    $this->CharTable = array(
  'iso2' => array(
  0x80=>"\xc2\x80",
  0x81=>"\xc2\x81",
  0x82=>"\xc2\x82",
  0x83=>"\xc2\x83",
  0x84=>"\xc2\x84",
  0x85=>"\xc2\x85",
  0x86=>"\xc2\x86",
  0x87=>"\xc2\x87",
  0x88=>"\xc2\x88",
  0x89=>"\xc2\x89",
  0x8A=>"\xc2\x8a",
  0x8B=>"\xc2\x8b",
  0x8C=>"\xc2\x8c",
  0x8D=>"\xc2\x8d",
  0x8E=>"\xc2\x8e",
  0x8F=>"\xc2\x8f",
  0x90=>"\xc2\x90",
  0x91=>"\xc2\x91",
  0x92=>"\xc2\x92",
  0x93=>"\xc2\x93",
  0x94=>"\xc2\x94",
  0x95=>"\xc2\x95",
  0x96=>"\xc2\x96",
  0x97=>"\xc2\x97",
  0x98=>"\xc2\x98",
  0x99=>"\xc2\x99",
  0x9A=>"\xc2\x9a",
  0x9B=>"\xc2\x9b",
  0x9C=>"\xc2\x9c",
  0x9D=>"\xc2\x9d",
  0x9E=>"\xc2\x9e",
  0x9F=>"\xc2\x9f",
  0xA0=>"\xc2\xa0",
  0xA1=>"\xc4\x84",
  0xA2=>"\xcb\x98",
  0xA3=>"\xc5\x81",
  0xA4=>"\xc2\xa4",
  0xA5=>"\xc4\xbd",
  0xA6=>"\xc5\x9a",
  0xA7=>"\xc2\xa7",
  0xA8=>"\xc2\xa8",
  0xA9=>"\xc5\xa0",
  0xAA=>"\xc5\x9e",
  0xAB=>"\xc5\xa4",
  0xAC=>"\xc5\xb9",
  0xAD=>"\xc2\xad",
  0xAE=>"\xc5\xbd",
  0xAF=>"\xc5\xbb",
  0xB0=>"\xc2\xb0",
  0xB1=>"\xc4\x85",
  0xB2=>"\xcb\x9b",
  0xB3=>"\xc5\x82",
  0xB4=>"\xc2\xb4",
  0xB5=>"\xc4\xbe",
  0xB6=>"\xc5\x9b",
  0xB7=>"\xcb\x87",
  0xB8=>"\xc2\xb8",
  0xB9=>"\xc5\xa1",
  0xBA=>"\xc5\x9f",
  0xBB=>"\xc5\xa5",
  0xBC=>"\xc5\xba",
  0xBD=>"\xcb\x9d",
  0xBE=>"\xc5\xbe",
  0xBF=>"\xc5\xbc",
  0xC0=>"\xc5\x94",
  0xC1=>"\xc3\x81",
  0xC2=>"\xc3\x82",
  0xC3=>"\xc4\x82",
  0xC4=>"\xc3\x84",
  0xC5=>"\xc4\xb9",
  0xC6=>"\xc4\x86",
  0xC7=>"\xc3\x87",
  0xC8=>"\xc4\x8c",
  0xC9=>"\xc3\x89",
  0xCA=>"\xc4\x98",
  0xCB=>"\xc3\x8b",
  0xCC=>"\xc4\x9a",
  0xCD=>"\xc3\x8d",
  0xCE=>"\xc3\x8e",
  0xCF=>"\xc4\x8e",
  0xD0=>"\xc4\x90",
  0xD1=>"\xc5\x83",
  0xD2=>"\xc5\x87",
  0xD3=>"\xc3\x93",
  0xD4=>"\xc3\x94",
  0xD5=>"\xc5\x90",
  0xD6=>"\xc3\x96",
  0xD7=>"\xc3\x97",
  0xD8=>"\xc5\x98",
  0xD9=>"\xc5\xae",
  0xDA=>"\xc3\x9a",
  0xDB=>"\xc5\xb0",
  0xDC=>"\xc3\x9c",
  0xDD=>"\xc3\x9d",
  0xDE=>"\xc5\xa2",
  0xDF=>"\xc3\x9f",
  0xE0=>"\xc5\x95",
  0xE1=>"\xc3\xa1",
  0xE2=>"\xc3\xa2",
  0xE3=>"\xc4\x83",
  0xE4=>"\xc3\xa4",
  0xE5=>"\xc4\xba",
  0xE6=>"\xc4\x87",
  0xE7=>"\xc3\xa7",
  0xE8=>"\xc4\x8d",
  0xE9=>"\xc3\xa9",
  0xEA=>"\xc4\x99",
  0xEB=>"\xc3\xab",
  0xEC=>"\xc4\x9b",
  0xED=>"\xc3\xad",
  0xEE=>"\xc3\xae",
  0xEF=>"\xc4\x8f",
  0xF0=>"\xc4\x91",
  0xF1=>"\xc5\x84",
  0xF2=>"\xc5\x88",
  0xF3=>"\xc3\xb3",
  0xF4=>"\xc3\xb4",
  0xF5=>"\xc5\x91",
  0xF6=>"\xc3\xb6",
  0xF7=>"\xc3\xb7",
  0xF8=>"\xc5\x99",
  0xF9=>"\xc5\xaf",
  0xFA=>"\xc3\xba",
  0xFB=>"\xc5\xb1",
  0xFC=>"\xc3\xbc",
  0xFD=>"\xc3\xbd",
  0xFE=>"\xc5\xa3",
  0xFF=>"\xcb\x99"
  ),
  'win1250' => array(
  0x80=>"\xc2\x80",
  0x81=>"\xc2\x81",
  0x82=>"\xe2\x80\x9a",
  0x83=>"\xc2\x83",
  0x84=>"\xe2\x80\x9e",
  0x85=>"\xe2\x80\xa6",
  0x86=>"\xe2\x80\xa0",
  0x87=>"\xe2\x80\xa1",
  0x88=>"\xc2\x88",
  0x89=>"\xe2\x80\xb0",
  0x8a=>"\xc5\xa0",
  0x8b=>"\xe2\x80\xb9",
  0x8c=>"\xc5\x9a",
  0x8d=>"\xc5\xa4",
  0x8e=>"\xc5\xbd",
  0x8f=>"\xc5\xb9",
  0x90=>"\xc2\x90",
  0x91=>"\xe2\x80\x98",
  0x92=>"\xe2\x80\x99",
  0x93=>"\xe2\x80\x9c",
  0x94=>"\xe2\x80\x9d",
  0x95=>"\xe2\x80\xa2",
  0x96=>"\xe2\x80\x93",
  0x97=>"\xe2\x80\x94",
  0x98=>"\xe2\x80\x98",
  0x99=>"\xe2\x84\xa2",
  0x9a=>"\xc5\xa1",
  0x9b=>"\xe2\x80\xba",
  0x9c=>"\xc5\x9b",
  0x9d=>"\xc5\xa5",
  0x9e=>"\xc5\xbe",
  0x9f=>"\xc5\xba",
  0xa0=>"\xc2\xa0",
  0xa1=>"\xcb\x87",
  0xa2=>"\xcb\x98",
  0xa3=>"\xc5\x81",
  0xa4=>"\xc2\xa4",
  0xa5=>"\xc4\x84",
  0xa6=>"\xc2\xa6",
  0xa7=>"\xc2\xa7",
  0xa8=>"\xc2\xa8",
  0xa9=>"\xc2\xa9",
  0xaa=>"\xc5\x9e",
  0xab=>"\xc2\xab",
  0xac=>"\xc2\xac",
  0xad=>"\xc2\xad",
  0xae=>"\xc2\xae",
  0xaf=>"\xc5\xbb",
  0xb0=>"\xc2\xb0",
  0xb1=>"\xc2\xb1",
  0xb2=>"\xcb\x9b",
  0xb3=>"\xc5\x82",
  0xb4=>"\xc2\xb4",
  0xb5=>"\xc2\xb5",
  0xb6=>"\xc2\xb6",
  0xb7=>"\xc2\xb7",
  0xb8=>"\xc2\xb8",
  0xb9=>"\xc4\x85",
  0xba=>"\xc5\x9f",
  0xbb=>"\xc2\xbb",
  0xbc=>"\xc4\xbd",
  0xbd=>"\xcb\x9d",
  0xbe=>"\xc4\xbe",
  0xbf=>"\xc5\xbc",
  0xc0=>"\xc5\x94",
  0xc1=>"\xc3\x81",
  0xc2=>"\xc3\x82",
  0xc3=>"\xc4\x82",
  0xc4=>"\xc3\x84",
  0xc5=>"\xc4\xb9",
  0xc6=>"\xc4\x86",
  0xc7=>"\xc3\x87",
  0xc8=>"\xc4\x8c",
  0xc9=>"\xc3\x89",
  0xca=>"\xc4\x98",
  0xcb=>"\xc3\x8b",
  0xcc=>"\xc4\x9a",
  0xcd=>"\xc3\x8d",
  0xce=>"\xc3\x8e",
  0xcf=>"\xc4\x8e",
  0xd0=>"\xc4\x90",
  0xd1=>"\xc5\x83",
  0xd2=>"\xc5\x87",
  0xd3=>"\xc3\x93",
  0xd4=>"\xc3\x94",
  0xd5=>"\xc5\x90",
  0xd6=>"\xc3\x96",
  0xd7=>"\xc3\x97",
  0xd8=>"\xc5\x98",
  0xd9=>"\xc5\xae",
  0xda=>"\xc3\x9a",
  0xdb=>"\xc5\xb0",
  0xdc=>"\xc3\x9c",
  0xdd=>"\xc3\x9d",
  0xde=>"\xc5\xa2",
  0xdf=>"\xc3\x9f",
  0xe0=>"\xc5\x95",
  0xe1=>"\xc3\xa1",
  0xe2=>"\xc3\xa2",
  0xe3=>"\xc4\x83",
  0xe4=>"\xc3\xa4",
  0xe5=>"\xc4\xba",
  0xe6=>"\xc4\x87",
  0xe7=>"\xc3\xa7",
  0xe8=>"\xc4\x8d",
  0xe9=>"\xc3\xa9",
  0xea=>"\xc4\x99",
  0xeb=>"\xc3\xab",
  0xec=>"\xc4\x9b",
  0xed=>"\xc3\xad",
  0xee=>"\xc3\xae",
  0xef=>"\xc4\x8f",
  0xf0=>"\xc4\x91",
  0xf1=>"\xc5\x84",
  0xf2=>"\xc5\x88",
  0xf3=>"\xc3\xb3",
  0xf4=>"\xc3\xb4",
  0xf5=>"\xc5\x91",
  0xf6=>"\xc3\xb6",
  0xf7=>"\xc3\xb7",
  0xf8=>"\xc5\x99",
  0xf9=>"\xc5\xaf",
  0xfa=>"\xc3\xba",
  0xfb=>"\xc5\xb1",
  0xfc=>"\xc3\xbc",
  0xfd=>"\xc3\xbd",
  0xfe=>"\xc5\xa3",
  0xff=>"\xcb\x99"
  ),
  'iso1' => array(
  0xA0=>"\xc2\xa0",
  0xA1=>"\xc2\xa1",
  0xA2=>"\xc2\xa2",
  0xA3=>"\xc2\xa3",
  0xA4=>"\xc2\xa4",
  0xA5=>"\xc2\xa5",
  0xA6=>"\xc2\xa6",
  0xA7=>"\xc2\xa7",
  0xA8=>"\xc2\xa8",
  0xA9=>"\xc2\xa9",
  0xAA=>"\xc2\xaa",
  0xAB=>"\xc2\xab",
  0xAC=>"\xc2\xac",
  0xAD=>"\xc2\xad",
  0xAE=>"\xc2\xae",
  0xAF=>"\xc2\xaf",
  0xB0=>"\xc2\xb0",
  0xB1=>"\xc2\xb1",
  0xB2=>"\xc2\xb2",
  0xB3=>"\xc2\xb3",
  0xB4=>"\xc2\xb4",
  0xB5=>"\xc2\xb5",
  0xB6=>"\xc2\xb6",
  0xB7=>"\xc2\xb7",
  0xB8=>"\xc2\xb8",
  0xB9=>"\xc2\xb9",
  0xBA=>"\xc2\xba",
  0xBB=>"\xc2\xbb",
  0xBC=>"\xc2\xbc",
  0xBD=>"\xc2\xbd",
  0xBE=>"\xc2\xbe",
  0xBF=>"\xc2\xbf",
  0xC0=>"\xc3\x80",
  0xC1=>"\xc3\x81",
  0xC2=>"\xc3\x82",
  0xC3=>"\xc3\x83",
  0xC4=>"\xc3\x84",
  0xC5=>"\xc3\x85",
  0xC6=>"\xc3\x86",
  0xC7=>"\xc3\x87",
  0xC8=>"\xc3\x88",
  0xC9=>"\xc3\x89",
  0xCA=>"\xc3\x8a",
  0xCB=>"\xc3\x8b",
  0xCC=>"\xc3\x8c",
  0xCD=>"\xc3\x8d",
  0xCE=>"\xc3\x8e",
  0xCF=>"\xc3\x8f",
  0xD0=>"\xc3\x90",
  0xD1=>"\xc3\x91",
  0xD2=>"\xc3\x92",
  0xD3=>"\xc3\x93",
  0xD4=>"\xc3\x94",
  0xD5=>"\xc3\x95",
  0xD6=>"\xc3\x96",
  0xD7=>"\xc3\x97",
  0xD8=>"\xc3\x98",
  0xD9=>"\xc3\x99",
  0xDA=>"\xc3\x9a",
  0xDB=>"\xc3\x9b",
  0xDC=>"\xc3\x9c",
  0xDD=>"\xc3\x9d",
  0xDE=>"\xc3\x9e",
  0xDF=>"\xc3\x9f",
  0xE0=>"\xc3\xa0",
  0xE1=>"\xc3\xa1",
  0xE2=>"\xc3\xa2",
  0xE3=>"\xc3\xa3",
  0xE4=>"\xc3\xa4",
  0xE5=>"\xc3\xa5",
  0xE6=>"\xc3\xa6",
  0xE7=>"\xc3\xa7",
  0xE8=>"\xc3\xa8",
  0xE9=>"\xc3\xa9",
  0xEA=>"\xc3\xaa",
  0xEB=>"\xc3\xab",
  0xEC=>"\xc3\xac",
  0xED=>"\xc3\xad",
  0xEE=>"\xc3\xae",
  0xEF=>"\xc3\xaf",
  0xF0=>"\xc3\xb0",
  0xF1=>"\xc3\xb1",
  0xF2=>"\xc3\xb2",
  0xF3=>"\xc3\xb3",
  0xF4=>"\xc3\xb4",
  0xF5=>"\xc3\xb5",
  0xF6=>"\xc3\xb6",
  0xF7=>"\xc3\xb7",
  0xF8=>"\xc3\xb8",
  0xF9=>"\xc3\xb9",
  0xFA=>"\xc3\xba",
  0xFB=>"\xc3\xbb",
  0xFC=>"\xc3\xbc",
  0xFD=>"\xc3\xbd",
  0xFE=>"\xc3\xbe"
  ),
  'win1257' => array(
  0x80=>"\xe2\x82\xac",
  0x81=>"\xc0\x4",
  0x82=>"\xe2\x80\x9a",
  0x83=>"\xc0\x4",
  0x84=>"\xe2\x80\x9e",
  0x85=>"\xe2\x80\xa6",
  0x86=>"\xe2\x80\xa0",
  0x87=>"\xe2\x80\xa1",
  0x88=>"\xc0\x4",
  0x89=>"\xe2\x80\xb0",
  0x8A=>"\xc0\x4",
  0x8B=>"\xe2\x80\xb9",
  0x8C=>"\xc0\x4",
  0x8D=>"\xc2\xa8",
  0x8E=>"\xcb\x87",
  0x8F=>"\xc2\xb8",
  0x90=>"\xc0\x4",
  0x91=>"\xe2\x80\x98",
  0x92=>"\xe2\x80\x99",
  0x93=>"\xe2\x80\x9c",
  0x94=>"\xe2\x80\x9d",
  0x95=>"\xe2\x80\xa2",
  0x96=>"\xe2\x80\x93",
  0x97=>"\xe2\x80\x94",
  0x98=>"\xc0\x4",
  0x99=>"\xe2\x84\xa2",
  0x9A=>"\xc0\x4",
  0x9B=>"\xe2\x80\xba",
  0x9C=>"\xc0\x4",
  0x9D=>"\xc2\xaf",
  0x9E=>"\xcb\x9b",
  0x9F=>"\xc0\x4",
  0xA0=>"\xc2\xa0",
  0xA1=>"\xc0\x4",
  0xA2=>"\xc2\xa2",
  0xA3=>"\xc2\xa3",
  0xA4=>"\xc2\xa4",
  0xA5=>"\xc0\x4",
  0xA6=>"\xc2\xa6",
  0xA7=>"\xc2\xa7",
  0xA8=>"\xc3\x98",
  0xA9=>"\xc2\xa9",
  0xAA=>"\xc5\x96",
  0xAB=>"\xc2\xab",
  0xAC=>"\xc2\xac",
  0xAD=>"\xc2\xad",
  0xAE=>"\xc2\xae",
  0xAF=>"\xc3\x86",
  0xB0=>"\xc2\xb0",
  0xB1=>"\xc2\xb1",
  0xB2=>"\xc2\xb2",
  0xB3=>"\xc2\xb3",
  0xB4=>"\xc2\xb4",
  0xB5=>"\xc2\xb5",
  0xB6=>"\xc2\xb6",
  0xB7=>"\xc2\xb7",
  0xB8=>"\xc3\xb8",
  0xB9=>"\xc2\xb9",
  0xBA=>"\xc5\x97",
  0xBB=>"\xc2\xbb",
  0xBC=>"\xc2\xbc",
  0xBD=>"\xc2\xbd",
  0xBE=>"\xc2\xbe",
  0xBF=>"\xc3\xa6",
  0xC0=>"\xc4\x84",
  0xC1=>"\xc4\xae",
  0xC2=>"\xc4\x80",
  0xC3=>"\xc4\x86",
  0xC4=>"\xc3\x84",
  0xC5=>"\xc3\x85",
  0xC6=>"\xc4\x98",
  0xC7=>"\xc4\x92",
  0xC8=>"\xc4\x8c",
  0xC9=>"\xc3\x89",
  0xCA=>"\xc5\xb9",
  0xCB=>"\xc4\x96",
  0xCC=>"\xc4\xa2",
  0xCD=>"\xc4\xb6",
  0xCE=>"\xc4\xaa",
  0xCF=>"\xc4\xbb",
  0xD0=>"\xc5\xa0",
  0xD1=>"\xc5\x83",
  0xD2=>"\xc5\x85",
  0xD3=>"\xc3\x93",
  0xD4=>"\xc5\x8c",
  0xD5=>"\xc3\x95",
  0xD6=>"\xc3\x96",
  0xD7=>"\xc3\x97",
  0xD8=>"\xc5\xb2",
  0xD9=>"\xc5\x81",
  0xDA=>"\xc5\x9a",
  0xDB=>"\xc5\xaa",
  0xDC=>"\xc3\x9c",
  0xDD=>"\xc5\xbb",
  0xDE=>"\xc5\xbd",
  0xDF=>"\xc3\x9f",
  0xE0=>"\xc4\x85",
  0xE1=>"\xc4\xaf",
  0xE2=>"\xc4\x81",
  0xE3=>"\xc4\x87",
  0xE4=>"\xc3\xa4",
  0xE5=>"\xc3\xa5",
  0xE6=>"\xc4\x99",
  0xE7=>"\xc4\x93",
  0xE8=>"\xc4\x8d",
  0xE9=>"\xc3\xa9",
  0xEA=>"\xc5\xba",
  0xEB=>"\xc4\x97",
  0xEC=>"\xc4\xa3",
  0xED=>"\xc4\xb7",
  0xEE=>"\xc4\xab",
  0xEF=>"\xc4\xbc",
  0xF0=>"\xc5\xa1",
  0xF1=>"\xc5\x84",
  0xF2=>"\xc5\x86",
  0xF3=>"\xc3\xb3",
  0xF4=>"\xc5\x8d",
  0xF5=>"\xc3\xb5",
  0xF6=>"\xc3\xb6",
  0xF7=>"\xc3\xb7",
  0xF8=>"\xc5\xb3",
  0xF9=>"\xc5\x82",
  0xFA=>"\xc5\x9b",
  0xFB=>"\xc5\xab",
  0xFC=>"\xc3\xbc",
  0xFD=>"\xc5\xbc",
  0xFE=>"\xc5\xbe",
  0xFF=>"\xcb\x99"
  ),
);
  }

  function ToUTF8string(string $String, string $Charset = 'iso2'): string
  {
    $Result = '';
    for ($I = 0; $I < strlen($String); $I++)
    {
       if (ord($String[$I]) < 128) $Result .= $String[$I];
       else if (ord($String[$I]) > 127)
       {
         $Result .= $this->CharTable[$Charset][ord($String[$I])];
       }
    }
    return $Result;
  }

  function FromUTF8(string $String, string $Charset = 'iso2'): string
  {
    $Result = '';
    $UTFPrefix = '';
    for ($I = 0; $I < strlen($String); $I++)
    {
      if (ord($String[$I]) & 0x80) // UTF control character
      {
        if (ord($String[$I]) & 0x40) // First
        {
          if ($UTFPrefix != '') $Result .= chr(array_search($UTFPrefix, $this->CharTable[$Charset]));
          $UTFPrefix = $String[$I];
        }
        else $UTFPrefix .= $String[$I]; // Next
      } else
      {
        if ($UTFPrefix != '') $Result .= chr(array_search($UTFPrefix, $this->CharTable[$Charset]));
        $UTFPrefix = '';
        $Result .= $String[$I];
      }
    }
    return $Result;
  }
}