utf8encodings.c

   1 /*-------------------------------------------------------------------------*/
   2 /* utf8encodings.c --- translation tables between an encoding and UTF-8    */
   3 /* Copyright (c) : see below                                               */
   4 /*-------------------------------------------------------------------------*/
   5 /*-------------------------------------------------------------------------*/
   6 /* written (ie. converted the tables) by Erik van der Wal, May 2014        */
   7 /*-------------------------------------------------------------------------*/
   8
   9 #include <stdio.h>
  10
  11 /*-------------------------------------------------------------------------*/
  12 /* The data in this file is based on a number of files from unicode.org    */
  13 /* The specific source of the translation is referenced in the heading     */
  14 /* above each encoding                                                     */
  15 /* All these files contained the following copyright notice:               */
  16 /* (I underlined the relevant part)                                        */
  17 /*-------------------------------------------------------------------------*/
  18 /*
  19 #       Copyright (c) 1991-1999 Unicode, Inc.  All Rights reserved.
  20 #
  21 #       This file is provided as-is by Unicode, Inc. (The Unicode Consortium).
  22 #       No claims are made as to fitness for any particular purpose.  No
  23 #       warranties of any kind are expressed or implied.  The recipient
  24 #       agrees to determine applicability of information provided.  If this
  25 #       file has been provided on optical media by Unicode, Inc., the sole
  26 #       remedy for any claim will be exchange of defective media within 90
  27 #       days of receipt.
  28 #
  29 #       Unicode, Inc. hereby grants the right to freely use the information
  30 #                                                --------------------------
  31 #       supplied in this file in the creation of products supporting the
  32 #       Unicode Standard, and to make copies of this file in any form for
  33 #       internal or external distribution as long as this notice remains
  34 #                                         ------------------------------
  35 #       attached.
  36 #       --------
  37 */
  38
  39
  40 /* Since the character at index 0 can never be used, it is used to store   */
  41 /* the identification of the given encoding                                */
  42 /* The encoding can be placed in random order, as long as it does start    */
  43 /* with the standard encoding and it ends with a NULL, to indicate the end */
  44 /* of the list.                                                            */
  45 /* Non-existent characters are encoded with the unicode character:         */
  46 /* 'REPLACEMENT CHARACTER' (typically a boxed question mark).              */
  47
  48 #define REPLC_CHAR "\xef\xbf\xbd"
  49
  50 const char *utf8encodings[][256] = {
  51 /*--------------------------------------------------------------------------*/
  52 /* stdenc                                                                   */
  53 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/ADOBE/stdenc.txt */
  54 /*--------------------------------------------------------------------------*/
  55 {
  56 /* 0x00 */ "Standard",     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  57 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  58 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  59 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  60 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  61 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  62 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  63 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  64 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&","\xe2\x80\x99",
  65 /* 0x28 */ "(",    ")",    "*",    "+",    ",", "\xc2\xad",".",    "/",
  66 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
  67 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
  68 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
  69 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
  70 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
  71 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
  72 /* 0x60 */ "\xe2\x80\x98","a","b", "c",    "d",    "e",    "f",    "g",
  73 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
  74 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
  75 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    REPLC_CHAR,
  76 /* 0x80 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  77 /* 0x84 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  78 /* 0x88 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  79 /* 0x8c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  80 /* 0x90 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  81 /* 0x94 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  82 /* 0x98 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  83 /* 0x9c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  84 /* 0xa0 */ REPLC_CHAR,     "\xc2\xa1",     "\xc2\xa2",     "\xc2\xa3",
  85 /* 0xa4 */ "\xe2\x81\x84", "\xc2\xa5",     "\xc6\x92",     "\xc2\xa7",
  86 /* 0xa8 */ "\xc2\xa4",     "'",            "\xe2\x80\x9c", "\xc2\xab",
  87 /* 0xac */ "\xe2\x80\xb9", "\xe2\x80\xba", "\xef\xac\x81", "\xef\xac\x82",
  88 /* 0xb0 */ REPLC_CHAR,     "\xe2\x80\x93", "\xe2\x80\xa0", "\xe2\x80\xa1",
  89 /* 0xb4 */ "\xc2\xb7",     REPLC_CHAR,     "\xc2\xb6",     "\xe2\x80\xa2",
  90 /* 0xb8 */ "\xe2\x80\x9a", "\xe2\x80\x9e", "\xe2\x80\x9d", "\xc2\xbb",
  91 /* 0xbc */ "\xe2\x80\xa6", "\xe2\x80\xb0", REPLC_CHAR,     "\xc2\xbf",
  92 /* 0xc0 */ REPLC_CHAR,     "`",            "\xc2\xb4",     "\xcb\x86",
  93 /* 0xc4 */ "\xcb\x9c",     "\xc2\xaf",     "\xcb\x98",     "\xcb\x99",
  94 /* 0xc8 */ "\xc2\xa8",     REPLC_CHAR,     "\xcb\x9a",     "\xc2\xb8",
  95 /* 0xcc */ REPLC_CHAR,     "\xcb\x9d",     "\xcb\x9b",     "\xcb\x87",
  96 /* 0xd0 */ "\xe2\x80\x94", REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  97 /* 0xd4 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  98 /* 0xd8 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
  99 /* 0xdc */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 100 /* 0xe0 */ REPLC_CHAR,     "\xc3\x86",     REPLC_CHAR,     "\xc2\xaa",
 101 /* 0xe4 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 102 /* 0xe8 */ "\xc5\x81",     "\xc3\x98",     "\xc5\x92",     "\xc2\xba",
 103 /* 0xec */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 104 /* 0xf0 */ REPLC_CHAR,     "\xc3\xa6",     REPLC_CHAR,     REPLC_CHAR,
 105 /* 0xf4 */ REPLC_CHAR,     "\xc4\xb1",     REPLC_CHAR,     REPLC_CHAR,
 106 /* 0xf8 */ "\xc5\x82",     "\xc3\xb8",     "\xc5\x93",     "\xc3\x9f",
 107 /* 0xfc */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR
 108 },
 109 /*--------------------------------------------------------------------------*/
 110 /* symbol                                                                   */
 111 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/ADOBE/symbol.txt */
 112 /*--------------------------------------------------------------------------*/
 113 {
 114 /* 0x00 */ "Symbol",       REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 115 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 116 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 117 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 118 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 119 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 120 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 121 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 122 /* 0x20 */ " ",            "!",            "\xe2\x88\x80", "#",
 123 /* 0x24 */ "\xe2\x88\x83", "%",            "&",            "\xe2\x88\x8b",
 124 /* 0x28 */ "(",            ")",            "\xe2\x88\x97", "+",
 125 /* 0x2c */ ",",            "\xe2\x88\x92", ".",            "/",
 126 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 127 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 128 /* 0x40 */ "\xe2\x89\x85", "\xce\x91",     "\xce\x92",     "\xce\xa7",
 129 /* 0x44 */ "\xce\x94",     "\xce\x95",     "\xce\xa6",     "\xce\x93",
 130 /* 0x48 */ "\xce\x97",     "\xce\x99",     "\xcf\x91",     "\xce\x9a",
 131 /* 0x4c */ "\xce\x9b",     "\xce\x9c",     "\xce\x9d",     "\xce\x9f",
 132 /* 0x50 */ "\xce\xa0",     "\xce\x98",     "\xce\xa1",     "\xce\xa3",
 133 /* 0x54 */ "\xce\xa4",     "\xce\xa5",     "\xcf\x82",     "\xe2\x84\xa6",
 134 /* 0x58 */ "\xce\x9e",     "\xce\xa8",     "\xce\x96",     "[",
 135 /* 0x5c */ "\xe2\x88\xb4", "]",            "\xe2\x8a\xa5", "_",
 136 /* 0x60 */ "\xef\xa3\xa5", "\xce\xb1",     "\xce\xb2",     "\xcf\x87",
 137 /* 0x64 */ "\xce\xb4",     "\xce\xb5",     "\xcf\x86",     "\xce\xb3",
 138 /* 0x68 */ "\xce\xb7",     "\xce\xb9",     "\xcf\x95",     "\xce\xba",
 139 /* 0x6c */ "\xce\xbb",     "\xc2\xb5",     "\xce\xbd",     "\xce\xbf",
 140 /* 0x70 */ "\xcf\x80",     "\xce\xb8",     "\xcf\x81",     "\xcf\x83",
 141 /* 0x74 */ "\xcf\x84",     "\xcf\x85",     "\xcf\x96",     "\xcf\x89",
 142 /* 0x78 */ "\xce\xbe",     "\xcf\x88",     "\xce\xb6",     "{",
 143 /* 0x7c */ "|",            "}",            "\xe2\x88\xbc", REPLC_CHAR,
 144 /* 0x80 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 145 /* 0x84 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 146 /* 0x88 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 147 /* 0x8c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 148 /* 0x90 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 149 /* 0x94 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 150 /* 0x98 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 151 /* 0x9c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 152 /* 0xa0 */ "\xe2\x82\xac", "\xcf\x92",     "\xe2\x80\xb2", "\xe2\x89\xa4",
 153 /* 0xa4 */ "\xe2\x88\x95", "\xe2\x88\x9e", "\xc6\x92",     "\xe2\x99\xa3",
 154 /* 0xa8 */ "\xe2\x99\xa6", "\xe2\x99\xa5", "\xe2\x99\xa0", "\xe2\x86\x94",
 155 /* 0xac */ "\xe2\x86\x90", "\xe2\x86\x91", "\xe2\x86\x92", "\xe2\x86\x93",
 156 /* 0xb0 */ "\xc2\xb0",     "\xc2\xb1",     "\xe2\x80\xb3", "\xe2\x89\xa5",
 157 /* 0xb4 */ "\xc3\x97",     "\xe2\x88\x9d", "\xe2\x88\x82", "\xe2\x80\xa2",
 158 /* 0xb8 */ "\xc3\xb7",     "\xe2\x89\xa0", "\xe2\x89\xa1", "\xe2\x89\x88",
 159 /* 0xbc */ "\xe2\x80\xa6", "\xef\xa3\xa6", "\xef\xa3\xa7", "\xe2\x86\xb5",
 160 /* 0xc0 */ "\xe2\x84\xb5", "\xe2\x84\x91", "\xe2\x84\x9c", "\xe2\x84\x98",
 161 /* 0xc4 */ "\xe2\x8a\x97", "\xe2\x8a\x95", "\xe2\x88\x85", "\xe2\x88\xa9",
 162 /* 0xc8 */ "\xe2\x88\xaa", "\xe2\x8a\x83", "\xe2\x8a\x87", "\xe2\x8a\x84",
 163 /* 0xcc */ "\xe2\x8a\x82", "\xe2\x8a\x86", "\xe2\x88\x88", "\xe2\x88\x89",
 164 /* 0xd0 */ "\xe2\x88\xa0", "\xe2\x88\x87", "\xef\x9b\x9a", "\xef\x9b\x99",
 165 /* 0xd4 */ "\xef\x9b\x9b", "\xe2\x88\x8f", "\xe2\x88\x9a", "\xe2\x8b\x85",
 166 /* 0xd8 */ "\xc2\xac",     "\xe2\x88\xa7", "\xe2\x88\xa8", "\xe2\x87\x94",
 167 /* 0xdc */ "\xe2\x87\x90", "\xe2\x87\x91", "\xe2\x87\x92", "\xe2\x87\x93",
 168 /* 0xe0 */ "\xe2\x97\x8a", "\xe2\x8c\xa9", "\xef\xa3\xa8", "\xef\xa3\xa9",
 169 /* 0xe4 */ "\xef\xa3\xaa", "\xe2\x88\x91", "\xef\xa3\xab", "\xef\xa3\xac",
 170 /* 0xe8 */ "\xef\xa3\xad", "\xef\xa3\xae", "\xef\xa3\xaf", "\xef\xa3\xb0",
 171 /* 0xec */ "\xef\xa3\xb1", "\xef\xa3\xb2", "\xef\xa3\xb3", "\xef\xa3\xb4",
 172 /* 0xf0 */ REPLC_CHAR,     "\xe2\x8c\xaa", "\xe2\x88\xab", "\xe2\x8c\xa0",
 173 /* 0xf4 */ "\xef\xa3\xb5", "\xe2\x8c\xa1", "\xef\xa3\xb6", "\xef\xa3\xb7",
 174 /* 0xf8 */ "\xef\xa3\xb8", "\xef\xa3\xb9", "\xef\xa3\xba", "\xef\xa3\xbb",
 175 /* 0xfc */ "\xef\xa3\xbc", "\xef\xa3\xbd", "\xef\xa3\xbe", REPLC_CHAR
 176 },
 177 /*--------------------------------------------------------------------------*/
 178 /* iso 8859-1                                                               */
 179 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/ISO8859/8859-1.TXT       */
 180 /*--------------------------------------------------------------------------*/
 181 {
 182 /* 0x00 */ "ISOLatin1",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 183 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 184 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 185 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 186 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 187 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 188 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 189 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 190 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 191 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 192 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 193 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 194 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 195 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 196 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 197 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 198 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 199 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 200 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 201 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 202 /* 0x80 */ "\xc2\x80",     "\xc2\x81",     "\xc2\x82",     "\xc2\x83",
 203 /* 0x84 */ "\xc2\x84",     "\xc2\x85",     "\xc2\x86",     "\xc2\x87",
 204 /* 0x88 */ "\xc2\x88",     "\xc2\x89",     "\xc2\x8a",     "\xc2\x8b",
 205 /* 0x8c */ "\xc2\x8c",     "\xc2\x8d",     "\xc2\x8e",     "\xc2\x8f",
 206 /* 0x90 */ "\xc2\x90",     "\xc2\x91",     "\xc2\x92",     "\xc2\x93",
 207 /* 0x94 */ "\xc2\x94",     "\xc2\x95",     "\xc2\x96",     "\xc2\x97",
 208 /* 0x98 */ "\xc2\x98",     "\xc2\x99",     "\xc2\x9a",     "\xc2\x9b",
 209 /* 0x9c */ "\xc2\x9c",     "\xc2\x9d",     "\xc2\x9e",     "\xc2\x9f",
 210 /* 0xa0 */ "\xc2\xa0",     "\xc2\xa1",     "\xc2\xa2",     "\xc2\xa3",
 211 /* 0xa4 */ "\xc2\xa4",     "\xc2\xa5",     "\xc2\xa6",     "\xc2\xa7",
 212 /* 0xa8 */ "\xc2\xa8",     "\xc2\xa9",     "\xc2\xaa",     "\xc2\xab",
 213 /* 0xac */ "\xc2\xac",     "\xc2\xad",     "\xc2\xae",     "\xc2\xaf",
 214 /* 0xb0 */ "\xc2\xb0",     "\xc2\xb1",     "\xc2\xb2",     "\xc2\xb3",
 215 /* 0xb4 */ "\xc2\xb4",     "\xc2\xb5",     "\xc2\xb6",     "\xc2\xb7",
 216 /* 0xb8 */ "\xc2\xb8",     "\xc2\xb9",     "\xc2\xba",     "\xc2\xbb",
 217 /* 0xbc */ "\xc2\xbc",     "\xc2\xbd",     "\xc2\xbe",     "\xc2\xbf",
 218 /* 0xc0 */ "\xc3\x80",     "\xc3\x81",     "\xc3\x82",     "\xc3\x83",
 219 /* 0xc4 */ "\xc3\x84",     "\xc3\x85",     "\xc3\x86",     "\xc3\x87",
 220 /* 0xc8 */ "\xc3\x88",     "\xc3\x89",     "\xc3\x8a",     "\xc3\x8b",
 221 /* 0xcc */ "\xc3\x8c",     "\xc3\x8d",     "\xc3\x8e",     "\xc3\x8f",
 222 /* 0xd0 */ "\xc3\x90",     "\xc3\x91",     "\xc3\x92",     "\xc3\x93",
 223 /* 0xd4 */ "\xc3\x94",     "\xc3\x95",     "\xc3\x96",     "\xc3\x97",
 224 /* 0xd8 */ "\xc3\x98",     "\xc3\x99",     "\xc3\x9a",     "\xc3\x9b",
 225 /* 0xdc */ "\xc3\x9c",     "\xc3\x9d",     "\xc3\x9e",     "\xc3\x9f",
 226 /* 0xe0 */ "\xc3\xa0",     "\xc3\xa1",     "\xc3\xa2",     "\xc3\xa3",
 227 /* 0xe4 */ "\xc3\xa4",     "\xc3\xa5",     "\xc3\xa6",     "\xc3\xa7",
 228 /* 0xe8 */ "\xc3\xa8",     "\xc3\xa9",     "\xc3\xaa",     "\xc3\xab",
 229 /* 0xec */ "\xc3\xac",     "\xc3\xad",     "\xc3\xae",     "\xc3\xaf",
 230 /* 0xf0 */ "\xc3\xb0",     "\xc3\xb1",     "\xc3\xb2",     "\xc3\xb3",
 231 /* 0xf4 */ "\xc3\xb4",     "\xc3\xb5",     "\xc3\xb6",     "\xc3\xb7",
 232 /* 0xf8 */ "\xc3\xb8",     "\xc3\xb9",     "\xc3\xba",     "\xc3\xbb",
 233 /* 0xfc */ "\xc3\xbc",     "\xc3\xbd",     "\xc3\xbe",     "\xc3\xbf"
 234 },
 235 /*--------------------------------------------------------------------------*/
 236 /* iso 8859-2                                                               */
 237 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/ISO8859/8859-2.TXT       */
 238 /*--------------------------------------------------------------------------*/
 239 {
 240 /* 0x00 */ "ISOLatin2",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 241 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 242 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 243 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 244 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 245 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 246 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 247 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 248 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 249 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 250 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 251 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 252 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 253 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 254 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 255 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 256 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 257 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 258 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 259 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 260 /* 0x80 */ "\xc2\x80",     "\xc2\x81",     "\xc2\x82",     "\xc2\x83",
 261 /* 0x84 */ "\xc2\x84",     "\xc2\x85",     "\xc2\x86",     "\xc2\x87",
 262 /* 0x88 */ "\xc2\x88",     "\xc2\x89",     "\xc2\x8a",     "\xc2\x8b",
 263 /* 0x8c */ "\xc2\x8c",     "\xc2\x8d",     "\xc2\x8e",     "\xc2\x8f",
 264 /* 0x90 */ "\xc2\x90",     "\xc2\x91",     "\xc2\x92",     "\xc2\x93",
 265 /* 0x94 */ "\xc2\x94",     "\xc2\x95",     "\xc2\x96",     "\xc2\x97",
 266 /* 0x98 */ "\xc2\x98",     "\xc2\x99",     "\xc2\x9a",     "\xc2\x9b",
 267 /* 0x9c */ "\xc2\x9c",     "\xc2\x9d",     "\xc2\x9e",     "\xc2\x9f",
 268 /* 0xa0 */ "\xc2\xa0",     "\xc4\x84",     "\xcb\x98",     "\xc5\x81",
 269 /* 0xa4 */ "\xc2\xa4",     "\xc4\xbd",     "\xc5\x9a",     "\xc2\xa7",
 270 /* 0xa8 */ "\xc2\xa8",     "\xc5\xa0",     "\xc5\x9e",     "\xc5\xa4",
 271 /* 0xac */ "\xc5\xb9",     "\xc2\xad",     "\xc5\xbd",     "\xc5\xbb",
 272 /* 0xb0 */ "\xc2\xb0",     "\xc4\x85",     "\xcb\x9b",     "\xc5\x82",
 273 /* 0xb4 */ "\xc2\xb4",     "\xc4\xbe",     "\xc5\x9b",     "\xcb\x87",
 274 /* 0xb8 */ "\xc2\xb8",     "\xc5\xa1",     "\xc5\x9f",     "\xc5\xa5",
 275 /* 0xbc */ "\xc5\xba",     "\xcb\x9d",     "\xc5\xbe",     "\xc5\xbc",
 276 /* 0xc0 */ "\xc5\x94",     "\xc3\x81",     "\xc3\x82",     "\xc4\x82",
 277 /* 0xc4 */ "\xc3\x84",     "\xc4\xb9",     "\xc4\x86",     "\xc3\x87",
 278 /* 0xc8 */ "\xc4\x8c",     "\xc3\x89",     "\xc4\x98",     "\xc3\x8b",
 279 /* 0xcc */ "\xc4\x9a",     "\xc3\x8d",     "\xc3\x8e",     "\xc4\x8e",
 280 /* 0xd0 */ "\xc4\x90",     "\xc5\x83",     "\xc5\x87",     "\xc3\x93",
 281 /* 0xd4 */ "\xc3\x94",     "\xc5\x90",     "\xc3\x96",     "\xc3\x97",
 282 /* 0xd8 */ "\xc5\x98",     "\xc5\xae",     "\xc3\x9a",     "\xc5\xb0",
 283 /* 0xdc */ "\xc3\x9c",     "\xc3\x9d",     "\xc5\xa2",     "\xc3\x9f",
 284 /* 0xe0 */ "\xc5\x95",     "\xc3\xa1",     "\xc3\xa2",     "\xc4\x83",
 285 /* 0xe4 */ "\xc3\xa4",     "\xc4\xba",     "\xc4\x87",     "\xc3\xa7",
 286 /* 0xe8 */ "\xc4\x8d",     "\xc3\xa9",     "\xc4\x99",     "\xc3\xab",
 287 /* 0xec */ "\xc4\x9b",     "\xc3\xad",     "\xc3\xae",     "\xc4\x8f",
 288 /* 0xf0 */ "\xc4\x91",     "\xc5\x84",     "\xc5\x88",     "\xc3\xb3",
 289 /* 0xf4 */ "\xc3\xb4",     "\xc5\x91",     "\xc3\xb6",     "\xc3\xb7",
 290 /* 0xf8 */ "\xc5\x99",     "\xc5\xaf",     "\xc3\xba",     "\xc5\xb1",
 291 /* 0xfc */ "\xc3\xbc",     "\xc3\xbd",     "\xc5\xa3",     "\xcb\x99"
 292 },
 293 /*--------------------------------------------------------------------------*/
 294 /* iso 8859-3                                                               */
 295 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/ISO8859/8859-3.TXT       */
 296 /*--------------------------------------------------------------------------*/
 297 {
 298 /* 0x00 */ "ISOLatin3",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 299 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 300 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 301 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 302 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 303 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 304 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 305 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 306 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 307 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 308 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 309 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 310 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 311 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 312 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 313 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 314 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 315 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 316 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 317 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 318 /* 0x80 */ "\xc2\x80",     "\xc2\x81",     "\xc2\x82",     "\xc2\x83",
 319 /* 0x84 */ "\xc2\x84",     "\xc2\x85",     "\xc2\x86",     "\xc2\x87",
 320 /* 0x88 */ "\xc2\x88",     "\xc2\x89",     "\xc2\x8a",     "\xc2\x8b",
 321 /* 0x8c */ "\xc2\x8c",     "\xc2\x8d",     "\xc2\x8e",     "\xc2\x8f",
 322 /* 0x90 */ "\xc2\x90",     "\xc2\x91",     "\xc2\x92",     "\xc2\x93",
 323 /* 0x94 */ "\xc2\x94",     "\xc2\x95",     "\xc2\x96",     "\xc2\x97",
 324 /* 0x98 */ "\xc2\x98",     "\xc2\x99",     "\xc2\x9a",     "\xc2\x9b",
 325 /* 0x9c */ "\xc2\x9c",     "\xc2\x9d",     "\xc2\x9e",     "\xc2\x9f",
 326 /* 0xa0 */ "\xc2\xa0",     "\xc4\xa6",     "\xcb\x98",     "\xc2\xa3",
 327 /* 0xa4 */ "\xc2\xa4",     REPLC_CHAR,     "\xc4\xa4",     "\xc2\xa7",
 328 /* 0xa8 */ "\xc2\xa8",     "\xc4\xb0",     "\xc5\x9e",     "\xc4\x9e",
 329 /* 0xac */ "\xc4\xb4",     "\xc2\xad",     REPLC_CHAR,     "\xc5\xbb",
 330 /* 0xb0 */ "\xc2\xb0",     "\xc4\xa7",     "\xc2\xb2",     "\xc2\xb3",
 331 /* 0xb4 */ "\xc2\xb4",     "\xc2\xb5",     "\xc4\xa5",     "\xc2\xb7",
 332 /* 0xb8 */ "\xc2\xb8",     "\xc4\xb1",     "\xc5\x9f",     "\xc4\x9f",
 333 /* 0xbc */ "\xc4\xb5",     "\xc2\xbd",     REPLC_CHAR,     "\xc5\xbc",
 334 /* 0xc0 */ "\xc3\x80",     "\xc3\x81",     "\xc3\x82",     REPLC_CHAR,
 335 /* 0xc4 */ "\xc3\x84",     "\xc4\x8a",     "\xc4\x88",     "\xc3\x87",
 336 /* 0xc8 */ "\xc3\x88",     "\xc3\x89",     "\xc3\x8a",     "\xc3\x8b",
 337 /* 0xcc */ "\xc3\x8c",     "\xc3\x8d",     "\xc3\x8e",     "\xc3\x8f",
 338 /* 0xd0 */ REPLC_CHAR,     "\xc3\x91",     "\xc3\x92",     "\xc3\x93",
 339 /* 0xd4 */ "\xc3\x94",     "\xc4\xa0",     "\xc3\x96",     "\xc3\x97",
 340 /* 0xd8 */ "\xc4\x9c",     "\xc3\x99",     "\xc3\x9a",     "\xc3\x9b",
 341 /* 0xdc */ "\xc3\x9c",     "\xc5\xac",     "\xc5\x9c",     "\xc3\x9f",
 342 /* 0xe0 */ "\xc3\xa0",     "\xc3\xa1",     "\xc3\xa2",     REPLC_CHAR,
 343 /* 0xe4 */ "\xc3\xa4",     "\xc4\x8b",     "\xc4\x89",     "\xc3\xa7",
 344 /* 0xe8 */ "\xc3\xa8",     "\xc3\xa9",     "\xc3\xaa",     "\xc3\xab",
 345 /* 0xec */ "\xc3\xac",     "\xc3\xad",     "\xc3\xae",     "\xc3\xaf",
 346 /* 0xf0 */ REPLC_CHAR,     "\xc3\xb1",     "\xc3\xb2",     "\xc3\xb3",
 347 /* 0xf4 */ "\xc3\xb4",     "\xc4\xa1",     "\xc3\xb6",     "\xc3\xb7",
 348 /* 0xf8 */ "\xc4\x9d",     "\xc3\xb9",     "\xc3\xba",     "\xc3\xbb",
 349 /* 0xfc */ "\xc3\xbc",     "\xc5\xad",     "\xc5\x9d",     "\xcb\x99"
 350 },
 351 /*--------------------------------------------------------------------------*/
 352 /* iso 8859-4                                                               */
 353 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/ISO8859/8859-4.TXT       */
 354 /*--------------------------------------------------------------------------*/
 355 {
 356 /* 0x00 */ "ISOLatin4",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 357 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 358 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 359 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 360 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 361 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 362 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 363 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 364 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 365 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 366 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 367 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 368 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 369 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 370 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 371 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 372 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 373 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 374 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 375 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 376 /* 0x80 */ "\xc2\x80",     "\xc2\x81",     "\xc2\x82",     "\xc2\x83",
 377 /* 0x84 */ "\xc2\x84",     "\xc2\x85",     "\xc2\x86",     "\xc2\x87",
 378 /* 0x88 */ "\xc2\x88",     "\xc2\x89",     "\xc2\x8a",     "\xc2\x8b",
 379 /* 0x8c */ "\xc2\x8c",     "\xc2\x8d",     "\xc2\x8e",     "\xc2\x8f",
 380 /* 0x90 */ "\xc2\x90",     "\xc2\x91",     "\xc2\x92",     "\xc2\x93",
 381 /* 0x94 */ "\xc2\x94",     "\xc2\x95",     "\xc2\x96",     "\xc2\x97",
 382 /* 0x98 */ "\xc2\x98",     "\xc2\x99",     "\xc2\x9a",     "\xc2\x9b",
 383 /* 0x9c */ "\xc2\x9c",     "\xc2\x9d",     "\xc2\x9e",     "\xc2\x9f",
 384 /* 0xa0 */ "\xc2\xa0",     "\xc4\x84",     "\xc4\xb8",     "\xc5\x96",
 385 /* 0xa4 */ "\xc2\xa4",     "\xc4\xa8",     "\xc4\xbb",     "\xc2\xa7",
 386 /* 0xa8 */ "\xc2\xa8",     "\xc5\xa0",     "\xc4\x92",     "\xc4\xa2",
 387 /* 0xac */ "\xc5\xa6",     "\xc2\xad",     "\xc5\xbd",     "\xc2\xaf",
 388 /* 0xb0 */ "\xc2\xb0",     "\xc4\x85",     "\xcb\x9b",     "\xc5\x97",
 389 /* 0xb4 */ "\xc2\xb4",     "\xc4\xa9",     "\xc4\xbc",     "\xcb\x87",
 390 /* 0xb8 */ "\xc2\xb8",     "\xc5\xa1",     "\xc4\x93",     "\xc4\xa3",
 391 /* 0xbc */ "\xc5\xa7",     "\xc5\x8a",     "\xc5\xbe",     "\xc5\x8b",
 392 /* 0xc0 */ "\xc4\x80",     "\xc3\x81",     "\xc3\x82",     "\xc3\x83",
 393 /* 0xc4 */ "\xc3\x84",     "\xc3\x85",     "\xc3\x86",     "\xc4\xae",
 394 /* 0xc8 */ "\xc4\x8c",     "\xc3\x89",     "\xc4\x98",     "\xc3\x8b",
 395 /* 0xcc */ "\xc4\x96",     "\xc3\x8d",     "\xc3\x8e",     "\xc4\xaa",
 396 /* 0xd0 */ "\xc4\x90",     "\xc5\x85",     "\xc5\x8c",     "\xc4\xb6",
 397 /* 0xd4 */ "\xc3\x94",     "\xc3\x95",     "\xc3\x96",     "\xc3\x97",
 398 /* 0xd8 */ "\xc3\x98",     "\xc5\xb2",     "\xc3\x9a",     "\xc3\x9b",
 399 /* 0xdc */ "\xc3\x9c",     "\xc5\xa8",     "\xc5\xaa",     "\xc3\x9f",
 400 /* 0xe0 */ "\xc4\x81",     "\xc3\xa1",     "\xc3\xa2",     "\xc3\xa3",
 401 /* 0xe4 */ "\xc3\xa4",     "\xc3\xa5",     "\xc3\xa6",     "\xc4\xaf",
 402 /* 0xe8 */ "\xc4\x8d",     "\xc3\xa9",     "\xc4\x99",     "\xc3\xab",
 403 /* 0xec */ "\xc4\x97",     "\xc3\xad",     "\xc3\xae",     "\xc4\xab",
 404 /* 0xf0 */ "\xc4\x91",     "\xc5\x86",     "\xc5\x8d",     "\xc4\xb7",
 405 /* 0xf4 */ "\xc3\xb4",     "\xc3\xb5",     "\xc3\xb6",     "\xc3\xb7",
 406 /* 0xf8 */ "\xc3\xb8",     "\xc5\xb3",     "\xc3\xba",     "\xc3\xbb",
 407 /* 0xfc */ "\xc3\xbc",     "\xc5\xa9",     "\xc5\xab",     "\xcb\x99"
 408 },
 409 /*--------------------------------------------------------------------------*/
 410 /* CP1254 (ISO Latin-5)                                                     */
 411 /* based on:                                                                */
 412 /* ftp://ftp.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1254.TXT  */
 413 /*--------------------------------------------------------------------------*/
 414 {
 415 /* 0x00 */ "ISOLatin5",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 416 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 417 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 418 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 419 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 420 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 421 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 422 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 423 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 424 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 425 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 426 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 427 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 428 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 429 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 430 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 431 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 432 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 433 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 434 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 435 /* 0x80 */ "\xe2\x82\xac", REPLC_CHAR,     "\xe2\x80\x9a", "\xc6\x92",
 436 /* 0x84 */ "\xe2\x80\x9e", "\xe2\x80\xa6", "\xe2\x80\xa0", "\xe2\x80\xa1",
 437 /* 0x88 */ "\xcb\x86",     "\xe2\x80\xb0", "\xc5\xa0",     "\xe2\x80\xb9",
 438 /* 0x8c */ "\xc5\x92",     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 439 /* 0x90 */ REPLC_CHAR,     "\xe2\x80\x98", "\xe2\x80\x99", "\xe2\x80\x9c",
 440 /* 0x94 */ "\xe2\x80\x9d", "\xe2\x80\xa2", "\xe2\x80\x93", "\xe2\x80\x94",
 441 /* 0x98 */ "\xcb\x9c",     "\xe2\x84\xa2", "\xc5\xa1",     "\xe2\x80\xba",
 442 /* 0x9c */ "\xc5\x93", REPLC_CHAR, REPLC_CHAR, "\xc5\xb8",
 443 /* 0xa0 */ "\xc2\xa0", "\xc2\xa1", "\xc2\xa2", "\xc2\xa3",
 444 /* 0xa4 */ "\xc2\xa4", "\xc2\xa5", "\xc2\xa6", "\xc2\xa7",
 445 /* 0xa8 */ "\xc2\xa8", "\xc2\xa9", "\xc2\xaa", "\xc2\xab",
 446 /* 0xac */ "\xc2\xac", "\xc2\xad", "\xc2\xae", "\xc2\xaf",
 447 /* 0xb0 */ "\xc2\xb0", "\xc2\xb1", "\xc2\xb2", "\xc2\xb3",
 448 /* 0xb4 */ "\xc2\xb4", "\xc2\xb5", "\xc2\xb6", "\xc2\xb7",
 449 /* 0xb8 */ "\xc2\xb8", "\xc2\xb9", "\xc2\xba", "\xc2\xbb",
 450 /* 0xbc */ "\xc2\xbc", "\xc2\xbd", "\xc2\xbe", "\xc2\xbf",
 451 /* 0xc0 */ "\xc3\x80", "\xc3\x81", "\xc3\x82", "\xc3\x83",
 452 /* 0xc4 */ "\xc3\x84", "\xc3\x85", "\xc3\x86", "\xc3\x87",
 453 /* 0xc8 */ "\xc3\x88", "\xc3\x89", "\xc3\x8a", "\xc3\x8b",
 454 /* 0xcc */ "\xc3\x8c", "\xc3\x8d", "\xc3\x8e", "\xc3\x8f",
 455 /* 0xd0 */ "\xc4\x9e", "\xc3\x91", "\xc3\x92", "\xc3\x93",
 456 /* 0xd4 */ "\xc3\x94", "\xc3\x95", "\xc3\x96", "\xc3\x97",
 457 /* 0xd8 */ "\xc3\x98", "\xc3\x99", "\xc3\x9a", "\xc3\x9b",
 458 /* 0xdc */ "\xc3\x9c", "\xc4\xb0", "\xc5\x9e", "\xc3\x9f",
 459 /* 0xe0 */ "\xc3\xa0", "\xc3\xa1", "\xc3\xa2", "\xc3\xa3",
 460 /* 0xe4 */ "\xc3\xa4", "\xc3\xa5", "\xc3\xa6", "\xc3\xa7",
 461 /* 0xe8 */ "\xc3\xa8", "\xc3\xa9", "\xc3\xaa", "\xc3\xab",
 462 /* 0xec */ "\xc3\xac", "\xc3\xad", "\xc3\xae", "\xc3\xaf",
 463 /* 0xf0 */ "\xc4\x9f", "\xc3\xb1", "\xc3\xb2", "\xc3\xb3",
 464 /* 0xf4 */ "\xc3\xb4", "\xc3\xb5", "\xc3\xb6", "\xc3\xb7",
 465 /* 0xf8 */ "\xc3\xb8", "\xc3\xb9", "\xc3\xba", "\xc3\xbb",
 466 /* 0xfc */ "\xc3\xbc", "\xc4\xb1", "\xc5\x9f", "\xc3\xbf"
 467 },
 468 /*--------------------------------------------------------------------------*/
 469 /* iso 8859-6                                                               */
 470 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/ISO8859/8859-6.TXT       */
 471 /*--------------------------------------------------------------------------*/
 472 {
 473 /* 0x00 */ "ISOLatin6",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 474 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 475 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 476 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 477 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 478 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 479 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 480 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 481 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 482 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 483 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 484 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 485 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 486 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 487 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 488 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 489 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 490 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 491 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 492 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 493 /* 0x80 */ "\xc2\x80",     "\xc2\x81",     "\xc2\x82",     "\xc2\x83",
 494 /* 0x84 */ "\xc2\x84",     "\xc2\x85",     "\xc2\x86",     "\xc2\x87",
 495 /* 0x88 */ "\xc2\x88",     "\xc2\x89",     "\xc2\x8a",     "\xc2\x8b",
 496 /* 0x8c */ "\xc2\x8c",     "\xc2\x8d",     "\xc2\x8e",     "\xc2\x8f",
 497 /* 0x90 */ "\xc2\x90",     "\xc2\x91",     "\xc2\x92",     "\xc2\x93",
 498 /* 0x94 */ "\xc2\x94",     "\xc2\x95",     "\xc2\x96",     "\xc2\x97",
 499 /* 0x98 */ "\xc2\x98",     "\xc2\x99",     "\xc2\x9a",     "\xc2\x9b",
 500 /* 0x9c */ "\xc2\x9c",     "\xc2\x9d",     "\xc2\x9e",     "\xc2\x9f",
 501 /* 0xa0 */ "\xc2\xa0",     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 502 /* 0xa4 */ "\xc2\xa4",     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 503 /* 0xa8 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 504 /* 0xac */ "\xd8\x8c",     "\xc2\xad",     REPLC_CHAR,     REPLC_CHAR,
 505 /* 0xb0 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 506 /* 0xb4 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 507 /* 0xb8 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     "\xd8\x9b",
 508 /* 0xbc */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     "\xd8\x9f",
 509 /* 0xc0 */ REPLC_CHAR,     "\xd8\xa1",     "\xd8\xa2",     "\xd8\xa3",
 510 /* 0xc4 */ "\xd8\xa4",     "\xd8\xa5",     "\xd8\xa6",     "\xd8\xa7",
 511 /* 0xc8 */ "\xd8\xa8",     "\xd8\xa9",     "\xd8\xaa",     "\xd8\xab",
 512 /* 0xcc */ "\xd8\xac",     "\xd8\xad",     "\xd8\xae",     "\xd8\xaf",
 513 /* 0xd0 */ "\xd8\xb0",     "\xd8\xb1",     "\xd8\xb2",     "\xd8\xb3",
 514 /* 0xd4 */ "\xd8\xb4",     "\xd8\xb5",     "\xd8\xb6",     "\xd8\xb7",
 515 /* 0xd8 */ "\xd8\xb8",     "\xd8\xb9",     "\xd8\xba",     REPLC_CHAR,
 516 /* 0xdc */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 517 /* 0xe0 */ "\xd9\x80",     "\xd9\x81",     "\xd9\x82",     "\xd9\x83",
 518 /* 0xe4 */ "\xd9\x84",     "\xd9\x85",     "\xd9\x86",     "\xd9\x87",
 519 /* 0xe8 */ "\xd9\x88",     "\xd9\x89",     "\xd9\x8a",     "\xd9\x8b",
 520 /* 0xec */ "\xd9\x8c",     "\xd9\x8d",     "\xd9\x8e",     "\xd9\x8f",
 521 /* 0xf0 */ "\xd9\x90",     "\xd9\x91",     "\xd9\x92",     REPLC_CHAR,
 522 /* 0xf4 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 523 /* 0xf8 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 524 /* 0xfc */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR
 525 },
 526 /*--------------------------------------------------------------------------*/
 527 /* iso 8859-5                                                               */
 528 /* based on: ftp://ftp.unicode.org/Public/MAPPINGS/ISO8859/8859-5.TXT       */
 529 /*--------------------------------------------------------------------------*/
 530 {
 531 /* 0x00 */ "ISO8859-5",    REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 532 /* 0x04 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 533 /* 0x08 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 534 /* 0x0c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 535 /* 0x10 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 536 /* 0x14 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 537 /* 0x18 */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 538 /* 0x1c */ REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,     REPLC_CHAR,
 539 /* 0x20 */ " ",    "!",    "\"",   "#",    "$",    "%",    "&",    "'",
 540 /* 0x28 */ "(",    ")",    "*",    "+",    ",",    "-",    ".",    "/",
 541 /* 0x30 */ "0",    "1",    "2",    "3",    "4",    "5",    "6",    "7",
 542 /* 0x38 */ "8",    "9",    ":",    ";",    "<",    "=",    ">",    "?",
 543 /* 0x40 */ "@",    "A",    "B",    "C",    "D",    "E",    "F",    "G",
 544 /* 0x48 */ "H",    "I",    "J",    "K",    "L",    "M",    "N",    "O",
 545 /* 0x50 */ "P",    "Q",    "R",    "S",    "T",    "U",    "V",    "W",
 546 /* 0x58 */ "X",    "Y",    "Z",    "[",    "\\",   "]",    "^",    "_",
 547 /* 0x60 */ "`",    "a",    "b",    "c",    "d",    "e",    "f",    "g",
 548 /* 0x68 */ "h",    "i",    "j",    "k",    "l",    "m",    "n",    "o",
 549 /* 0x70 */ "p",    "q",    "r",    "s",    "t",    "u",    "v",    "w",
 550 /* 0x78 */ "x",    "y",    "z",    "{",    "|",    "}",    "~",    "\x7f",
 551 /* 0x80 */ "\xc2\x80",     "\xc2\x81",     "\xc2\x82",     "\xc2\x83",
 552 /* 0x84 */ "\xc2\x84",     "\xc2\x85",     "\xc2\x86",     "\xc2\x87",
 553 /* 0x88 */ "\xc2\x88",     "\xc2\x89",     "\xc2\x8a",     "\xc2\x8b",
 554 /* 0x8c */ "\xc2\x8c",     "\xc2\x8d",     "\xc2\x8e",     "\xc2\x8f",
 555 /* 0x90 */ "\xc2\x90",     "\xc2\x91",     "\xc2\x92",     "\xc2\x93",
 556 /* 0x94 */ "\xc2\x94",     "\xc2\x95",     "\xc2\x96",     "\xc2\x97",
 557 /* 0x98 */ "\xc2\x98",     "\xc2\x99",     "\xc2\x9a",     "\xc2\x9b",
 558 /* 0x9c */ "\xc2\x9c",     "\xc2\x9d",     "\xc2\x9e",     "\xc2\x9f",
 559 /* 0xa0 */ "\xc2\xa0",     "\xd0\x81",     "\xd0\x82",     "\xd0\x83",
 560 /* 0xa4 */ "\xd0\x84",     "\xd0\x85",     "\xd0\x86",     "\xd0\x87",
 561 /* 0xa8 */ "\xd0\x88",     "\xd0\x89",     "\xd0\x8a",     "\xd0\x8b",
 562 /* 0xac */ "\xd0\x8c",     "\xc2\xad",     "\xd0\x8e",     "\xd0\x8f",
 563 /* 0xb0 */ "\xd0\x90",     "\xd0\x91",     "\xd0\x92",     "\xd0\x93",
 564 /* 0xb4 */ "\xd0\x94",     "\xd0\x95",     "\xd0\x96",     "\xd0\x97",
 565 /* 0xb8 */ "\xd0\x98",     "\xd0\x99",     "\xd0\x9a",     "\xd0\x9b",
 566 /* 0xbc */ "\xd0\x9c",     "\xd0\x9d",     "\xd0\x9e",     "\xd0\x9f",
 567 /* 0xc0 */ "\xd0\xa0",     "\xd0\xa1",     "\xd0\xa2",     "\xd0\xa3",
 568 /* 0xc4 */ "\xd0\xa4",     "\xd0\xa5",     "\xd0\xa6",     "\xd0\xa7",
 569 /* 0xc8 */ "\xd0\xa8",     "\xd0\xa9",     "\xd0\xaa",     "\xd0\xab",
 570 /* 0xcc */ "\xd0\xac",     "\xd0\xad",     "\xd0\xae",     "\xd0\xaf",
 571 /* 0xd0 */ "\xd0\xb0",     "\xd0\xb1",     "\xd0\xb2",     "\xd0\xb3",
 572 /* 0xd4 */ "\xd0\xb4",     "\xd0\xb5",     "\xd0\xb6",     "\xd0\xb7",
 573 /* 0xd8 */ "\xd0\xb8",     "\xd0\xb9",     "\xd0\xba",     "\xd0\xbb",
 574 /* 0xdc */ "\xd0\xbc",     "\xd0\xbd",     "\xd0\xbe",     "\xd0\xbf",
 575 /* 0xe0 */ "\xd1\x80",     "\xd1\x81",     "\xd1\x82",     "\xd1\x83",
 576 /* 0xe4 */ "\xd1\x84",     "\xd1\x85",     "\xd1\x86",     "\xd1\x87",
 577 /* 0xe8 */ "\xd1\x88",     "\xd1\x89",     "\xd1\x8a",     "\xd1\x8b",
 578 /* 0xec */ "\xd1\x8c",     "\xd1\x8d",     "\xd1\x8e",     "\xd1\x8f",
 579 /* 0xf0 */ "\xe2\x84\x96", "\xd1\x91",     "\xd1\x92",     "\xd1\x93",
 580 /* 0xf4 */ "\xd1\x94",     "\xd1\x95",     "\xd1\x96",     "\xd1\x97",
 581 /* 0xf8 */ "\xd1\x98",     "\xd1\x99",     "\xd1\x9a",     "\xd1\x9b",
 582 /* 0xfc */ "\xd1\x9c",     "\xc2\xa7",     "\xd1\x9e",     "\xd1\x9f"
 583 },
 584
 585 /*--------------------------------------------------------------------------*/
 586 /* Closing line should be NULL                                              */
 587 /*--------------------------------------------------------------------------*/
 588 {(char*) 0}
 589
 590 };
 591
 592 /*----------------------------------------------------------------------*/
 593 /* Routine to search for a UTF-8 code in the tables and return the      */
 594 /* associated 8-bit code (reverse lookup).                              */
 595 /*                                                                      */
 596 /* Note that this routine does not check whether or not the encoding    */
 597 /* scheme containing the UTF-8 character code is the encoding           */
 598 /* currently in effect in a text string.                                */
 599 /*----------------------------------------------------------------------*/
 600
 601 int utf8_reverse_lookup(char *utf8string)
 602 {
 603     int i, j;
 604     char *s1, *s2;
 605
 606     for (i = 0; ; i++) {
 607         if (utf8encodings[i][0] == NULL) break;
 608         for (j = 0; j < 256; j++) {
 609             s1 = (char *)utf8encodings[i][j];
 610             s2 = utf8string;
 611             while (*s1 == *s2) {
 612                 if (*s1 == '\0')
 613                     return j;
 614                 s1++;
 615                 s2++;
 616             }
 617             /* NOTE: Xutf8LookupString() results in strings longer than */
 618             /* expected for the UTF-8 code.  Not sure what they are,    */
 619             /* but the initial part matches the expected code.          */
 620             if ((*s1 == '\0') && (s2 > (utf8string + 1))) return j;
 621         }
 622     }
 623     return -1;
 624 }