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(int argc, char **argv)
  24 {
  25     long int c;
  26     int verbose = FALSE;
  27
  28     if (argc > 1 && !strcmp(argv[1], "-v"))
  29         verbose = TRUE;
  30
  31     for (c = 0; c < 0x30000; c++) {
  32         int i, plane, row, col, chr;
  33         char const *sep = "";
  34
  35         printf("U+%04x:", c);
  36
  37         /*
  38          * Look up in SBCSes.
  39          */
  40         for (i = 0; i < lenof(cs_table); i++)
  41             if (cs_table[i]->read == read_sbcs &&
  42                 (chr = sbcs_from_unicode(cs_table[i]->data, c)) != ERROR) {
  43                 printf("%s %s", sep,
  44                        charset_to_localenc(cs_table[i]->charset));
  45                 if (verbose)
  46                     printf("[%d]", chr);
  47                 sep = ";";
  48             }
  49
  50         /*
  51          * Look up individually in MBCS base charsets.
  52          */
  53         if (unicode_to_big5(c, &row, &col)) {
  54             printf("%s Big5", sep);
  55             if (verbose)
  56                 printf("[%d,%d]", row, col);
  57             sep = ";";
  58         }
  59         if (unicode_to_gb2312(c, &row, &col)) {
  60             printf("%s GB2312", sep);
  61             if (verbose)
  62                 printf("[%d,%d]", row, col);
  63             sep = ";";
  64         }
  65
  66         if (unicode_to_jisx0208(c, &row, &col)) {
  67             printf("%s JIS X 0208", sep);
  68             if (verbose)
  69                 printf("[%d,%d]", row, col);
  70             sep = ";";
  71         }
  72
  73         if (unicode_to_ksx1001(c, &row, &col)) {
  74             printf("%s KS X 1001", sep);
  75             if (verbose)
  76                 printf("[%d,%d]", row, col);
  77             sep = ";";
  78         }
  79
  80         if (unicode_to_cp949(c, &row, &col)) {
  81             printf("%s CP949", sep);
  82             if (verbose)
  83                 printf("[%d,%d]", row, col);
  84             sep = ";";
  85         }
  86
  87         if (unicode_to_cns11643(c, &plane, &row, &col)) {
  88             printf("%s CNS11643", sep);
  89             if (verbose)
  90                 printf("[%d,%d,%d]", plane, row, col);
  91             sep = ";";
  92         }
  93
  94         if (!*sep)
  95             printf(" unicode-only");
  96
  97         printf("\n");
  98     }
  99
 100     return 0;
 101 }