mdw@git.distorted.org.uk Git - sgt/charset/blob - cstable.c

   1 /*
   2  * cstable.c - libcharset supporting utility which draws up a map
   3  * of the whole Unicode BMP and annotates it with details of which
   4  * other character sets each character appears in.
   5  *
   6  * Note this is not a libcharset _client_; it is part of the
   7  * libcharset _package_, using libcharset internals.
   8  */
   9
  10 #include "charset.h"
  11 #include "internal.h"
  12 #include "sbcsdat.h"
  13
  14 #define ENUM_CHARSET(x) extern charset_spec const charset_##x;
  15 #include "enum.c"
  16 #undef ENUM_CHARSET
  17 static charset_spec const *const cs_table[] = {
  18 #define ENUM_CHARSET(x) &charset_##x,
  19 #include "enum.c"
  20 #undef ENUM_CHARSET
  21 };
  22
  23 int main(void)
  24 {
  25     long int c;
  26
  27     for (c = 0; c < 0x30000; c++) {
  28         int i, plane, row, col;
  29         char const *sep = "";
  30
  31         printf("U+%04x:", c);
  32
  33         /*
  34          * Look up in SBCSes.
  35          */
  36         for (i = 0; i < lenof(cs_table); i++)
  37             if (cs_table[i]->read == read_sbcs &&
  38                 sbcs_from_unicode(cs_table[i]->data, c) != ERROR) {
  39                 printf("%s %s", sep,
  40                        charset_to_localenc(cs_table[i]->charset));
  41                 sep = ";";
  42             }
  43
  44         /*
  45          * Look up individually in MBCS base charsets.
  46          */
  47         if (unicode_to_big5(c, &row, &col)) {
  48             printf("%s Big5", sep);
  49             sep = ";";
  50         }
  51         if (unicode_to_gb2312(c, &row, &col)) {
  52             printf("%s GB2312", sep);
  53             sep = ";";
  54         }
  55
  56         if (unicode_to_jisx0208(c, &row, &col)) {
  57             printf("%s JIS X 0208", sep);
  58             sep = ";";
  59         }
  60
  61         if (unicode_to_ksx1001(c, &row, &col)) {
  62             printf("%s KS X 1001", sep);
  63             sep = ";";
  64         }
  65
  66         if (unicode_to_cp949(c, &row, &col)) {
  67             printf("%s CP949", sep);
  68             sep = ";";
  69         }
  70
  71         if (unicode_to_cns11643(c, &plane, &row, &col)) {
  72             printf("%s CNS11643", sep);
  73             sep = ";";
  74         }
  75
  76         if (!*sep)
  77             printf(" unicode-only");
  78
  79         printf("\n");
  80     }
  81
  82     return 0;
  83 }