functions/i18n.php

   1 <?
   2
   3 /**
   4  ** i18n.php
   5  **
   6  ** This file contains variuos functions that are needed to do
   7  ** internationalization of SquirrelMail.
   8  **
   9  ** Internally iso-8859-1 is used as character set. Other characters
  10  ** are encoded using Unicode entities according to HTML 4.0.
  11  **
  12  **/
  13
  14    $i18n_php = true;
  15
  16    // This array specifies the available languages.
  17    $languages[0]["NAME"] = "English";
  18    $languages[0]["CODE"] = "en";
  19    $languages[1]["NAME"] = "Norsk";
  20    $languages[1]["CODE"] = "no";
  21    $languages[2]["NAME"] = "Deutcsh";
  22    $languages[2]["CODE"] = "de";
  23
  24    // Decodes a string to the internal encoding from the given charset
  25    function charset_decode ($charset, $string) {
  26       // All HTML special characters are 7 bit and can be replaced first
  27       $string = htmlspecialchars ($string);
  28
  29       $charset = strtolower($charset);
  30
  31       if (ereg("iso-8859-(.*)", $charset, $res)) {
  32          if ($res[1] == "1")
  33             return charset_decode_iso_8859_1 ($string);
  34          if ($res[1] == "7")
  35             return charset_decode_iso_8859_7 ($string);
  36          else if ($res[1] == "15")
  37             return charset_decode_iso_8859_15 ($string);
  38          else
  39             return charset_decode_iso_8859_default ($string);
  40       } else if ($charset == "ns_4551-1") {
  41          return charset_decode_ns_4551_1 ($string);
  42       } else
  43          return $string;
  44    }
  45
  46    // iso-8859-1 is the same as Latin 1 and is normally used
  47    // in western europe.
  48    function charset_decode_iso_8859_1 ($string) {
  49       // This is only debug code as long as the internal
  50       // character set is iso-8859-1
  51
  52       // Latin small letter o with stroke
  53       while (ereg("\370", $string))
  54          $string = ereg_replace ("\370", "&#248;", $string);
  55
  56       return ($string);
  57    }
  58
  59    // iso-8859-7 is Greek.
  60    function charset_decode_iso_8859_7 ($string) {
  61       // Could not find Unicode equivalent of 0xA1 and 0xA2
  62       // 0xA4, 0xA5, 0xAA, 0xAE, 0xD2 and 0xFF should not be used
  63       $string = strtr($string, "\241\242\244\245\252\256\322\377",
  64                       "????????");
  65
  66       // Horizontal bar (parentheki pavla)
  67       while (ereg("\257", $string))
  68          $string = ereg_replace ("\257", "&#8213;", $string);
  69
  70       // ISO-8859-7 characters from 11/04 (0xB4) to 11/06 (0xB6)
  71       // These are Unicode 900-902
  72       while (ereg("([\264-\266])", $string, $res)) {
  73          $replace = "&#" . (ord($res[1])+720) . ";";
  74          $string = ereg_replace($res[1], $replace, $string);
  75       }
  76
  77       // 11/07 (0xB7) Middle dot is the same in iso-8859-1
  78
  79       // ISO-8859-7 characters from 11/08 (0xB8) to 11/10 (0xBA)
  80       // These are Unicode 900-902
  81       while (ereg("([\270-\272])", $string, $res)) {
  82          $replace = "&#" . (ord($res[1])+720) . ";";
  83          $string = ereg_replace($res[1], $replace, $string);
  84       }
  85
  86       // 11/11 (0xBB) Right angle quotation mark is the same as in
  87       // iso-8859-1
  88
  89       // And now the rest of the charset
  90       while (ereg("([\273-\376])", $string, $res)) {
  91          $replace = "&#" . (ord($res[1])+720) . ";";
  92          $string = ereg_replace($res[1], $replace, $string);
  93       }
  94
  95       return $string;
  96    }
  97
  98    // iso-8859-15 is Latin 15 and has very much the same use as Latin 1
  99    // but has the Euro symbol and some characters needed for French.
 100    function charset_decode_iso_8859_15 ($string) {
 101       // Euro sign
 102       while (ereg("\244", $replace))
 103          $string = ereg_replace ("\244", "&#8364;", $string);
 104       // Latin capital letter S with caron
 105       while (ereg("\246", $string))
 106          $string = ereg_replace ("\244", "&#352;", $string);
 107       // Latin small letter s with caron
 108       while (ereg("\250", $string))
 109          $string = ereg_replace ("\250", "&#353;", $string);
 110       // Latin capital letter Z with caron
 111       while (ereg("\264", $string))
 112          $string = ereg_replace ("\264", "&#381;", $string);
 113       // Latin small letter z with caron
 114       while (ereg("\270", $string))
 115          $string = ereg_replace ("\270", "&#382;", $string);
 116       // Latin capital ligature OE
 117       while (ereg("\274", $string))
 118          $string = ereg_replace ("\274", "&#338;", $string);
 119       // Latin small ligature oe
 120       while (ereg("\275", $string))
 121          $string = ereg_replace ("\275", "&#339;", $string);
 122       // Latin capital letter Y with diaeresis
 123       while (ereg("\276", $string))
 124          $string = ereg_replace ("\276", "&#376;", $string);
 125
 126       return ($string);
 127    }
 128
 129    // Remove all 8 bit characters from all other ISO-8859 character sets
 130    function charset_decode_iso_8859_default ($string) {
 131       return (strtr($string, "\240\241\242\243\244\245\246\247".
 132                     "\250\251\252\253\254\255\256\257".
 133                     "\260\261\262\263\264\265\266\267".
 134                     "\270\271\272\273\274\275\276\277".
 135                     "\300\301\302\303\304\305\306\307".
 136                     "\310\311\312\313\314\315\316\317".
 137                     "\320\321\322\323\324\325\326\327".
 138                     "\330\331\332\333\334\335\336\337".
 139                     "\340\341\342\343\344\345\346\347".
 140                     "\350\351\352\353\354\355\356\357".
 141                     "\360\361\362\363\364\365\366\367".
 142                     "\370\371\372\373\374\375\376\377",
 143                     "????????????????????????????????????????".
 144                     "????????????????????????????????????????".
 145                     "????????????????????????????????????????".
 146                     "????????"));
 147
 148    }
 149
 150    // This is the same as ISO-646-NO and is used by some
 151    // Microsoft programs when sending Norwegian characters
 152    function charset_decode_ns_4551_1 ($string) {
 153       // These characters are:
 154       // Latin capital letter AE
 155       // Latin capital letter O with stroke
 156       // Latin capital letter A with ring above
 157       // and the same as small letters
 158       return strtr ($string, "[\\]{|}", "ÆØÅæøå");
 159    }
 160
 161 ?>