tools/make_charset_table.c

   1 /* make_charset_table.c
   2  * sample program to generate tables for charsets.c using iconv
   3  *
   4  * public domain
   5  */
   6
   7 #include <stdio.h>
   8 #include <stdint.h>
   9 #include <errno.h>
  10 #include <iconv.h>
  11
  12 #define UNREPL 0xFFFD
  13
  14 int main(int argc, char **argv) {
  15         /* for now only UCS-2 */
  16         uint16_t table[0x100];
  17
  18         iconv_t conv;
  19         const char *charset;
  20         int i, j;
  21
  22         /* 0x00 ... 0x7F same as ASCII? */
  23         int ascii_based = 1;
  24         /* 0x00 ... 0x9F same as ISO? */
  25         int iso_based = 1;
  26
  27         if (argc != 2) {
  28                 printf("usage: %s <charset>\n", argv[0]);
  29                 return 1;
  30         }
  31
  32         charset = argv[1];
  33
  34         conv = iconv_open("UCS-2", charset);
  35         if (conv == (iconv_t) -1) {
  36                 perror("iconv_open");
  37                 return 2;
  38         }
  39         iconv_close(conv);
  40
  41         for (i = 0x00; i < 0x100; i++) {
  42                 unsigned char in[1], out[2];
  43                 size_t inlen = 1, outlen = 2;
  44
  45                 char *inbuf = (char *) in;
  46                 char *outbuf = (char *) out;
  47
  48                 size_t ret;
  49
  50                 in[0] = i;
  51
  52                 conv = iconv_open("UCS-2BE", charset);
  53
  54                 if (conv == (iconv_t) -1) {
  55                         /* shouldn't fail now */
  56                         perror("iconv_open");
  57                         return 2;
  58                 }
  59
  60                 ret = iconv(conv, &inbuf, &inlen, &outbuf, &outlen);
  61
  62                 if (ret == (size_t) -1 && errno == EILSEQ) {
  63                         table[i] = UNREPL;
  64                         iconv_close(conv);
  65                         continue;
  66                 }
  67
  68                 if (ret == (size_t) -1) {
  69                         perror("iconv");
  70                         iconv_close(conv);
  71                         return 4;
  72                 }
  73
  74                 iconv_close(conv);
  75
  76                 if (ret != 0 || inlen != 0 || outlen != 0) {
  77                         fprintf(stderr, "%d: smth went wrong: %zu %zu %zu\n", i, ret, inlen, outlen);
  78                         return 3;
  79                 }
  80
  81                 if (i < 0x80 && (out[0] != 0 || out[1] != i))
  82                         ascii_based = 0;
  83
  84                 if (i < 0xA0 && (out[0] != 0 || out[1] != i))
  85                         iso_based = 0;
  86
  87                 table[i] = (out[0] << 8) | out[1];
  88         }
  89
  90         /* iso_based not supported */
  91         iso_based = 0;
  92
  93         printf("/* generated by %s %s */\n", argv[0], charset);
  94
  95         if (iso_based)
  96                 i = 0xA0;
  97         else if (ascii_based)
  98                 i = 0x80;
  99         else
 100                 i = 0;
 101
 102         printf("const gunichar2 charset_table_%s[0x%x] = {\n", charset, 0x100 - i);
 103         while (i < 0x100) {
 104                 int start = i;
 105
 106                 printf("    ");
 107
 108                 for (j = 0; j < 8; j++, i++) {
 109                         if (table[i] == UNREPL)
 110                                 printf("UNREPL, ");
 111                         else
 112                                 printf("0x%.4x, ", table[i]);
 113                 }
 114
 115                 if ((start & 0xf) == 0)
 116                         printf("       /* 0x%.2X -      */", start);
 117                 else
 118                         printf("       /*      - 0x%.2X */", i - 1);
 119
 120                 printf("\n");
 121         }
 122         printf("};\n");
 123
 124         return 0;
 125 }