functions/decode/cp1251.php

   1 <?php
   2
   3 /**
   4  * decode/cp1251.php
   5  *
   6  * This file contains cp1251 decoding function that is needed to read
   7  * cp1251 encoded mails in non-cp1251 locale.
   8  *
   9  * Original data taken from:
  10  *  ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1250.TXT
  11  *
  12  *   Name:     cp1251 to Unicode table
  13  *   Unicode version: 2.0
  14  *   Table version: 2.01
  15  *   Table format:  Format A
  16  *   Date:          04/15/98
  17  *   Contact:       cpxlate@microsoft.com
  18  *
  19  * @copyright &copy; 2003-2006 The SquirrelMail Project Team
  20  * @license http://opensource.org/licenses/gpl-license.php GNU Public License
  21  * @version $Id$
  22  * @package squirrelmail
  23  * @subpackage decode
  24  */
  25
  26 /**
  27  * Decode cp1251-encoded string
  28  * @param string $string Encoded string
  29  * @return string $string Decoded string
  30  */
  31 function charset_decode_cp1251 ($string) {
  32     // don't do decoding when there are no 8bit symbols
  33     if (! sq_is8bit($string,'windows-1251'))
  34         return $string;
  35
  36     $cp1251 = array(
  37         "\x80" => '&#1026;',
  38         "\x81" => '&#1027;',
  39         "\x82" => '&#8218;',
  40         "\x83" => '&#1107;',
  41         "\x84" => '&#8222;',
  42         "\x85" => '&#8230;',
  43         "\x86" => '&#8224;',
  44         "\x87" => '&#8225;',
  45         "\x88" => '&#8364;',
  46         "\x89" => '&#8240;',
  47         "\x8A" => '&#1033;',
  48         "\x8B" => '&#8249;',
  49         "\x8C" => '&#1034;',
  50         "\x8D" => '&#1036;',
  51         "\x8E" => '&#1035;',
  52         "\x8F" => '&#1039;',
  53         "\x90" => '&#1106;',
  54         "\x91" => '&#8216;',
  55         "\x92" => '&#8217;',
  56         "\x93" => '&#8220;',
  57         "\x94" => '&#8221;',
  58         "\x95" => '&#8226;',
  59         "\x96" => '&#8211;',
  60         "\x97" => '&#8212;',
  61         "\x98" => '&#65533;',
  62         "\x99" => '&#8482;',
  63         "\x9A" => '&#1113;',
  64         "\x9B" => '&#8250;',
  65         "\x9C" => '&#1114;',
  66         "\x9D" => '&#1116;',
  67         "\x9E" => '&#1115;',
  68         "\x9F" => '&#1119;',
  69         "\xA0" => '&#160;',
  70         "\xA1" => '&#1038;',
  71         "\xA2" => '&#1118;',
  72         "\xA3" => '&#1032;',
  73         "\xA4" => '&#164;',
  74         "\xA5" => '&#1168;',
  75         "\xA6" => '&#166;',
  76         "\xA7" => '&#167;',
  77         "\xA8" => '&#1025;',
  78         "\xA9" => '&#169;',
  79         "\xAA" => '&#1028;',
  80         "\xAB" => '&#171;',
  81         "\xAC" => '&#172;',
  82         "\xAD" => '&#173;',
  83         "\xAE" => '&#174;',
  84         "\xAF" => '&#1031;',
  85         "\xB0" => '&#176;',
  86         "\xB1" => '&#177;',
  87         "\xB2" => '&#1030;',
  88         "\xB3" => '&#1110;',
  89         "\xB4" => '&#1169;',
  90         "\xB5" => '&#181;',
  91         "\xB6" => '&#182;',
  92         "\xB7" => '&#183;',
  93         "\xB8" => '&#1105;',
  94         "\xB9" => '&#8470;',
  95         "\xBA" => '&#1108;',
  96         "\xBB" => '&#187;',
  97         "\xBC" => '&#1112;',
  98         "\xBD" => '&#1029;',
  99         "\xBE" => '&#1109;',
 100         "\xBF" => '&#1111;',
 101         "\xC0" => '&#1040;',
 102         "\xC1" => '&#1041;',
 103         "\xC2" => '&#1042;',
 104         "\xC3" => '&#1043;',
 105         "\xC4" => '&#1044;',
 106         "\xC5" => '&#1045;',
 107         "\xC6" => '&#1046;',
 108         "\xC7" => '&#1047;',
 109         "\xC8" => '&#1048;',
 110         "\xC9" => '&#1049;',
 111         "\xCA" => '&#1050;',
 112         "\xCB" => '&#1051;',
 113         "\xCC" => '&#1052;',
 114         "\xCD" => '&#1053;',
 115         "\xCE" => '&#1054;',
 116         "\xCF" => '&#1055;',
 117         "\xD0" => '&#1056;',
 118         "\xD1" => '&#1057;',
 119         "\xD2" => '&#1058;',
 120         "\xD3" => '&#1059;',
 121         "\xD4" => '&#1060;',
 122         "\xD5" => '&#1061;',
 123         "\xD6" => '&#1062;',
 124         "\xD7" => '&#1063;',
 125         "\xD8" => '&#1064;',
 126         "\xD9" => '&#1065;',
 127         "\xDA" => '&#1066;',
 128         "\xDB" => '&#1067;',
 129         "\xDC" => '&#1068;',
 130         "\xDD" => '&#1069;',
 131         "\xDE" => '&#1070;',
 132         "\xDF" => '&#1071;',
 133         "\xE0" => '&#1072;',
 134         "\xE1" => '&#1073;',
 135         "\xE2" => '&#1074;',
 136         "\xE3" => '&#1075;',
 137         "\xE4" => '&#1076;',
 138         "\xE5" => '&#1077;',
 139         "\xE6" => '&#1078;',
 140         "\xE7" => '&#1079;',
 141         "\xE8" => '&#1080;',
 142         "\xE9" => '&#1081;',
 143         "\xEA" => '&#1082;',
 144         "\xEB" => '&#1083;',
 145         "\xEC" => '&#1084;',
 146         "\xED" => '&#1085;',
 147         "\xEE" => '&#1086;',
 148         "\xEF" => '&#1087;',
 149         "\xF0" => '&#1088;',
 150         "\xF1" => '&#1089;',
 151         "\xF2" => '&#1090;',
 152         "\xF3" => '&#1091;',
 153         "\xF4" => '&#1092;',
 154         "\xF5" => '&#1093;',
 155         "\xF6" => '&#1094;',
 156         "\xF7" => '&#1095;',
 157         "\xF8" => '&#1096;',
 158         "\xF9" => '&#1097;',
 159         "\xFA" => '&#1098;',
 160         "\xFB" => '&#1099;',
 161         "\xFC" => '&#1100;',
 162         "\xFD" => '&#1101;',
 163         "\xFE" => '&#1102;',
 164         "\xFF" => '&#1103;'
 165     );
 166
 167     $string = str_replace(array_keys($cp1251), array_values($cp1251), $string);
 168
 169     return $string;
 170 }