functions/decode/cp10017.php

   1 <?php
   2 /**
   3  * decode/cp10017.php
   4  * $Id$
   5  *
   6  * Copyright (c) 2003 The SquirrelMail Project Team
   7  * Licensed under the GNU GPL. For full terms see the file COPYING.
   8  *
   9  * This file contains cp10017 (MacUkrainian) decoding function that
  10  * is needed to read cp10017 encoded mails in non-cp10017 locale.
  11  *
  12  * Apple states [2] that x-mac-ukrainian differs from x-mac-cyrillic [1]
  13  * only in two places. According to [3] these symbols are:
  14  *  0x92 - U+1168 - CYRILLIC CAPITAL LETTER GHE WITH UPTURN
  15  *  0xD6 - U+1169 - CYRILLIC SMALL LETTER GHE WITH UPTURN
  16  *
  17  * References:
  18  * 1. ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/MAC/CYRILLIC.TXT
  19  * 2. http://developer.apple.com/documentation/macos8/TextIntlSvcs/TextEncodingConversionManager/TEC1.5/TEC.b0.html
  20  * 3. http://shlimazl.nm.ru/rus/cptable.htm (page in Russian)
  21  * @package squirrelmail
  22  * @subpackage decode
  23  */
  24
  25 /**
  26  * Decode a cp10017 (MacUkrainian) string
  27  * @param string $string Encoded string
  28  * @return string $string Decoded string
  29  */
  30 function charset_decode_cp10017 ($string) {
  31     global $default_charset;
  32
  33     if (strtolower($default_charset) == 'x-mac-ukrainian')
  34         return $string;
  35
  36     /* Only do the slow convert if there are 8-bit characters */
  37     /* avoid using 0xA0 (\240) in ereg ranges. RH73 does not like that */
  38     if (! ereg("[\200-\237]", $string) and ! ereg("[\241-\377]", $string) )
  39         return $string;
  40
  41     $cp10017 = array(
  42         "\x80" => '&#1040;',
  43         "\x81" => '&#1041;',
  44         "\x82" => '&#1042;',
  45         "\x83" => '&#1043;',
  46         "\x84" => '&#1044;',
  47         "\x85" => '&#1045;',
  48         "\x86" => '&#1046;',
  49         "\x87" => '&#1047;',
  50         "\x88" => '&#1048;',
  51         "\x89" => '&#1049;',
  52         "\x8A" => '&#1050;',
  53         "\x8B" => '&#1051;',
  54         "\x8C" => '&#1052;',
  55         "\x8D" => '&#1053;',
  56         "\x8E" => '&#1054;',
  57         "\x8F" => '&#1055;',
  58         "\x90" => '&#1056;',
  59         "\x91" => '&#1057;',
  60         "\x92" => '&#1168;',
  61         "\x93" => '&#1059;',
  62         "\x94" => '&#1060;',
  63         "\x95" => '&#1061;',
  64         "\x96" => '&#1062;',
  65         "\x97" => '&#1063;',
  66         "\x98" => '&#1064;',
  67         "\x99" => '&#1065;',
  68         "\x9A" => '&#1066;',
  69         "\x9B" => '&#1067;',
  70         "\x9C" => '&#1068;',
  71         "\x9D" => '&#1069;',
  72         "\x9E" => '&#1070;',
  73         "\x9F" => '&#1071;',
  74         "\xA0" => '&#8224;',
  75         "\xA1" => '&#176;',
  76         "\xA2" => '&#162;',
  77         "\xA3" => '&#163;',
  78         "\xA4" => '&#167;',
  79         "\xA5" => '&#8226;',
  80         "\xA6" => '&#182;',
  81         "\xA7" => '&#1030;',
  82         "\xA8" => '&#174;',
  83         "\xA9" => '&#169;',
  84         "\xAA" => '&#8482;',
  85         "\xAB" => '&#1026;',
  86         "\xAC" => '&#1106;',
  87         "\xAD" => '&#8800;',
  88         "\xAE" => '&#1027;',
  89         "\xAF" => '&#1107;',
  90         "\xB0" => '&#8734;',
  91         "\xB1" => '&#177;',
  92         "\xB2" => '&#8804;',
  93         "\xB3" => '&#8805;',
  94         "\xB4" => '&#1110;',
  95         "\xB5" => '&#181;',
  96         "\xB6" => '&#8706;',
  97         "\xB7" => '&#1032;',
  98         "\xB8" => '&#1028;',
  99         "\xB9" => '&#1108;',
 100         "\xBA" => '&#1031;',
 101         "\xBB" => '&#1111;',
 102         "\xBC" => '&#1033;',
 103         "\xBD" => '&#1113;',
 104         "\xBE" => '&#1034;',
 105         "\xBF" => '&#1114;',
 106         "\xC0" => '&#1112;',
 107         "\xC1" => '&#1029;',
 108         "\xC2" => '&#172;',
 109         "\xC3" => '&#8730;',
 110         "\xC4" => '&#402;',
 111         "\xC5" => '&#8776;',
 112         "\xC6" => '&#8710;',
 113         "\xC7" => '&#171;',
 114         "\xC8" => '&#187;',
 115         "\xC9" => '&#8230;',
 116         "\xCA" => '&#160;',
 117         "\xCB" => '&#1035;',
 118         "\xCC" => '&#1115;',
 119         "\xCD" => '&#1036;',
 120         "\xCE" => '&#1116;',
 121         "\xCF" => '&#1109;',
 122         "\xD0" => '&#8211;',
 123         "\xD1" => '&#8212;',
 124         "\xD2" => '&#8220;',
 125         "\xD3" => '&#8221;',
 126         "\xD4" => '&#8216;',
 127         "\xD5" => '&#8217;',
 128         "\xD6" => '&#1169;',
 129         "\xD7" => '&#8222;',
 130         "\xD8" => '&#1038;',
 131         "\xD9" => '&#1118;',
 132         "\xDA" => '&#1039;',
 133         "\xDB" => '&#1119;',
 134         "\xDC" => '&#8470;',
 135         "\xDD" => '&#1025;',
 136         "\xDE" => '&#1105;',
 137         "\xDF" => '&#1103;',
 138         "\xE0" => '&#1072;',
 139         "\xE1" => '&#1073;',
 140         "\xE2" => '&#1074;',
 141         "\xE3" => '&#1075;',
 142         "\xE4" => '&#1076;',
 143         "\xE5" => '&#1077;',
 144         "\xE6" => '&#1078;',
 145         "\xE7" => '&#1079;',
 146         "\xE8" => '&#1080;',
 147         "\xE9" => '&#1081;',
 148         "\xEA" => '&#1082;',
 149         "\xEB" => '&#1083;',
 150         "\xEC" => '&#1084;',
 151         "\xED" => '&#1085;',
 152         "\xEE" => '&#1086;',
 153         "\xEF" => '&#1087;',
 154         "\xF0" => '&#1088;',
 155         "\xF1" => '&#1089;',
 156         "\xF2" => '&#1090;',
 157         "\xF3" => '&#1091;',
 158         "\xF4" => '&#1092;',
 159         "\xF5" => '&#1093;',
 160         "\xF6" => '&#1094;',
 161         "\xF7" => '&#1095;',
 162         "\xF8" => '&#1096;',
 163         "\xF9" => '&#1097;',
 164         "\xFA" => '&#1098;',
 165         "\xFB" => '&#1099;',
 166         "\xFC" => '&#1100;',
 167         "\xFD" => '&#1101;',
 168         "\xFE" => '&#1102;',
 169         "\xFF" => '&#164;'
 170     );
 171
 172     $string = str_replace(array_keys($cp10017), array_values($cp10017), $string);
 173
 174     return $string;
 175 }
 176 ?>