tools/codepages.c

   1 /***************************************************************************
   2  *             __________               __   ___.
   3  *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
   4  *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
   5  *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
   6  *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
   7  *                     \/            \/     \/    \/            \/
   8  *
   9  *
  10  * Copyright (C) 2005 by Frank Dischner
  11  *
  12  * This program is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU General Public License
  14  * as published by the Free Software Foundation; either version 2
  15  * of the License, or (at your option) any later version.
  16  *
  17  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  18  * KIND, either express or implied.
  19  *
  20  ****************************************************************************/
  21 #include <stdio.h>
  22 #include <stdlib.h>
  23 #include <string.h>
  24 #include "codepage_tables.h"
  25
  26 #define MAX_TABLE_SIZE     32768
  27
  28 static const int mini_index[7] = {
  29     0, 1, 3, 6, 7, 8, 9
  30 };
  31
  32 static unsigned short iso_table[MAX_TABLE_SIZE];
  33
  34 unsigned short iso_decode(unsigned char *latin1, int cp, int count)
  35 {
  36     unsigned short ucs = 0;
  37
  38     /* cp tells us which codepage to convert from */
  39     switch (cp) {
  40         case 0x01: /* Greek (ISO-8859-7) */
  41             while (count--) {
  42                 /* first convert to unicode */
  43                 if (*latin1 < 0xA1)
  44                     ucs = *latin1++;
  45                 else if (*latin1 > 0xB7)
  46                     ucs = *latin1++ + 0x02D0;
  47                 else
  48                     ucs = iso8859_7_to_uni[*latin1++ - 0xA1];
  49             }
  50             break;
  51
  52         case 0x02: /* Hebrew (ISO-8859-8) */
  53             while (count--) {
  54                 /* first convert to unicode */
  55                 if (*latin1 == 0xAA) {
  56                     ucs = 0xD7;
  57                     latin1++;
  58                 } else if (*latin1 == 0xBA) {
  59                     ucs = 0xF7;
  60                     latin1++;
  61                 } else if (*latin1 == 0xDF) {
  62                     ucs = 0x2017;
  63                     latin1++;
  64                 } else if (*latin1 < 0xC0)
  65                     ucs = *latin1++;
  66                 else
  67                     ucs = *latin1++ + 0x04F0;
  68             }
  69             break;
  70
  71         case 0x03: /* Cyrillic (CP1251) */
  72             while (count--) {
  73                 /* first convert to unicode */
  74                 if (*latin1 < 0x80)
  75                     ucs = *latin1++;
  76                 else if (*latin1 > 0xBF)
  77                     ucs = *latin1++ + 0x0350;
  78                 else
  79                     ucs = cp1251_to_uni[*latin1++ - 0x80];
  80             }
  81             break;
  82
  83         case 0x04: /* Thai (ISO-8859-11) */
  84             while (count--) {
  85                 /* first convert to unicode */
  86                 if (*latin1 < 0xA1)
  87                     ucs = *latin1++;
  88                 else
  89                     ucs = *latin1++ + 0x0D60;
  90             }
  91             break;
  92
  93         case 0x05: /* Arabic (CP1256) */
  94             while (count--) {
  95                 /* first convert to unicode */
  96                 if (*latin1 < 0x80)
  97                     ucs = *latin1++;
  98                 else
  99                     ucs = cp1256_to_uni[*latin1++ - 0x80];
 100             }
 101             break;
 102
 103         case 0x06: /* Turkish (ISO-8859-9) */
 104             while (count--) {
 105                 /* first convert to unicode */
 106                 switch (*latin1) {
 107                     case 0xD0:
 108                         ucs = 0x011E;
 109                         break;
 110                     case 0xDD:
 111                         ucs = 0x0130;
 112                         break;
 113                     case 0xDE:
 114                         ucs = 0x015E;
 115                         break;
 116                     case 0xF0:
 117                         ucs = 0x011F;
 118                         break;
 119                     case 0xFD:
 120                         ucs = 0x0131;
 121                         break;
 122                     case 0xFE:
 123                         ucs = 0x015F;
 124                         break;
 125                     default:
 126                         ucs = *latin1;
 127                         break;
 128                 }
 129
 130                 latin1++;
 131             }
 132             break;
 133
 134         case 0x07: /* Latin Extended (ISO-8859-2) */
 135             while (count--) {
 136                 /* first convert to unicode */
 137                 if (*latin1 < 0xA1)
 138                     ucs = *latin1++;
 139                 else
 140                     ucs = iso8859_2_to_uni[*latin1++ - 0xA1];
 141             }
 142             break;
 143
 144         case 0x08: /* Central European (CP1250) */
 145             while (count--) {
 146                 /* first convert to unicode */
 147                 if (*latin1 < 0x80)
 148                     ucs = *latin1++;
 149                 else
 150                     ucs = cp1250_to_uni[*latin1++ - 0x80];
 151             }
 152             break;
 153
 154         case 0x09: /* Western European (CP1252) */
 155             while (count--) {
 156                 /* first convert to unicode */
 157                 if (*latin1 < 0x80 || *latin1 >= 0xa0)
 158                     ucs = *latin1++;
 159                 else
 160                     ucs = cp1252_to_uni[*latin1++ - 0x80];
 161             }
 162             break;
 163
 164         default:
 165             break;
 166     }
 167     return ucs;
 168 }
 169
 170 int writeshort(FILE *f, unsigned short s)
 171 {
 172     putc(s, f);
 173     return putc(s>>8, f) != EOF;
 174 }
 175
 176 void print_usage(void)
 177 {
 178     printf("Usage: codepages [-m]\n"
 179            "\t-m       Create isomini.cp only\n");
 180     printf("build date: " __DATE__ "\n\n");
 181 }
 182
 183 int main(int argc, char **argv)
 184 {
 185     int mini = 0;
 186     int i, j;
 187     unsigned char k;
 188     unsigned short uni;
 189     FILE *of;
 190
 191     for (i = 1;i < argc;i++)
 192     {
 193         if (argv[i][0] == '-')
 194         {
 195             switch (argv[i][1])
 196             {
 197               case 'm':   /* create isomini.cp only */
 198                 mini = 1;
 199                 break;
 200
 201               case 'h':   /* help */
 202               case '?':
 203                 print_usage();
 204                 exit(1);
 205                 break;
 206
 207               default:
 208                 print_usage();
 209                 exit(1);
 210                 break;
 211             }
 212         }
 213     }
 214
 215     for (i=0; i < MAX_TABLE_SIZE; i++)
 216         iso_table[i] = 0;
 217
 218     if (mini) {
 219         of = fopen("isomini.cp", "wb");
 220         if (!of) return 1;
 221
 222         for (i=1; i<7; i++) {
 223
 224             for (j=0; j<128; j++) {
 225                 k = (unsigned char)j + 128;
 226                 uni = iso_decode(&k, mini_index[i], 1);
 227                 writeshort(of, uni);
 228             }
 229         }
 230         fclose(of);
 231     }
 232     else {
 233         of = fopen("iso.cp", "wb");
 234         if (!of) return 1;
 235
 236         for (i=1; i<10; i++) {
 237
 238             for (j=0; j<128; j++) {
 239                 k = (unsigned char)j + 128;
 240                 uni = iso_decode(&k, i, 1);
 241                 writeshort(of, uni);
 242             }
 243         }
 244         fclose(of);
 245
 246         of = fopen("932.cp", "wb");
 247         if (!of) return 1;
 248         for (i=0; i < MAX_TABLE_SIZE; i++)
 249             writeshort(of, cp932_table[i]);
 250         fclose(of);
 251
 252         of = fopen("936.cp", "wb");
 253         if (!of) return 1;
 254         for (i=0; i < MAX_TABLE_SIZE; i++)
 255             writeshort(of, cp936_table[i]);
 256         fclose(of);
 257
 258         of = fopen("949.cp", "wb");
 259         if (!of) return 1;
 260         for (i=0; i < MAX_TABLE_SIZE; i++)
 261             writeshort(of, cp949_table[i]);
 262         fclose(of);
 263
 264         of = fopen("950.cp", "wb");
 265         if (!of) return 1;
 266         for (i=0; i < MAX_TABLE_SIZE; i++)
 267             writeshort(of, cp950_table[i]);
 268         fclose(of);
 269     }
 270
 271     return 0;
 272 }