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 <stdio.h>
  11 #include <string.h>
  12
  13 #include "charset.h"
  14 #include "internal.h"
  15 #include "sbcsdat.h"
  16
  17 #define ENUM_CHARSET(x) extern charset_spec const charset_##x;
  18 #include "enum.c"
  19 #undef ENUM_CHARSET
  20 static charset_spec const *const cs_table[] = {
  21 #define ENUM_CHARSET(x) &charset_##x,
  22 #include "enum.c"
  23 #undef ENUM_CHARSET
  24 };
  25 static const char *const cs_names[] = {
  26 #define ENUM_CHARSET(x) #x,
  27 #include "enum.c"
  28 #undef ENUM_CHARSET
  29 };
  30
  31 int main(int argc, char **argv)
  32 {
  33     long int c;
  34     int internal_names = FALSE;
  35     int verbose = FALSE;
  36
  37     while (--argc) {
  38         char *p = *++argv;
  39         if (!strcmp(p, "-i"))
  40             internal_names = TRUE;
  41         else if (!strcmp(p, "-v"))
  42             verbose = TRUE;
  43     }
  44
  45     for (c = 0; c < 0x30000; c++) {
  46         int i, plane, row, col, chr;
  47         char const *sep = "";
  48
  49         printf("U+%04x:", (unsigned)c);
  50
  51         /*
  52          * Look up in SBCSes.
  53          */
  54         for (i = 0; i < lenof(cs_table); i++)
  55             if (cs_table[i]->read == read_sbcs &&
  56                 (chr = sbcs_from_unicode(cs_table[i]->data, c)) != ERROR) {
  57                 printf("%s %s", sep,
  58                        (internal_names ? cs_names[i] :
  59                         charset_to_localenc(cs_table[i]->charset)));
  60                 if (verbose)
  61                     printf("[%d]", chr);
  62                 sep = ";";
  63             }
  64
  65         /*
  66          * Look up individually in MBCS base charsets. The
  67          * `internal_names' flag does not affect these, because
  68          * MBCS base charsets aren't directly encoded by CS_*
  69          * constants.
  70          */
  71         if (unicode_to_big5(c, &row, &col)) {
  72             printf("%s Big5", sep);
  73             if (verbose)
  74                 printf("[%d,%d]", row, col);
  75             sep = ";";
  76         }
  77
  78         if (unicode_to_gb2312(c, &row, &col)) {
  79             printf("%s GB2312", sep);
  80             if (verbose)
  81                 printf("[%d,%d]", row, col);
  82             sep = ";";
  83         }
  84
  85         if (unicode_to_jisx0208(c, &row, &col)) {
  86             printf("%s JIS X 0208", sep);
  87             if (verbose)
  88                 printf("[%d,%d]", row, col);
  89             sep = ";";
  90         }
  91
  92         if (unicode_to_ksx1001(c, &row, &col)) {
  93             printf("%s KS X 1001", sep);
  94             if (verbose)
  95                 printf("[%d,%d]", row, col);
  96             sep = ";";
  97         }
  98
  99         if (unicode_to_cp949(c, &row, &col)) {
 100             printf("%s CP949", sep);
 101             if (verbose)
 102                 printf("[%d,%d]", row, col);
 103             sep = ";";
 104         }
 105
 106         if (unicode_to_cns11643(c, &plane, &row, &col)) {
 107             printf("%s CNS11643", sep);
 108             if (verbose)
 109                 printf("[%d,%d,%d]", plane, row, col);
 110             sep = ";";
 111         }
 112
 113         if (!*sep)
 114             printf(" unicode-only");
 115
 116         printf("\n");
 117     }
 118
 119     return 0;
 120 }