mdw@git.distorted.org.uk Git - sgt/charset/blob - big5enc.c

   1 /*
   2  * big5enc.c - multibyte encoding of Big5
   3  */
   4
   5 #ifndef ENUM_CHARSETS
   6
   7 #include "charset.h"
   8 #include "internal.h"
   9
  10 /*
  11  * Big5 has no associated data, so `charset' may be ignored.
  12  */
  13
  14 static void read_big5(charset_spec const *charset, long int input_chr,
  15                       charset_state *state,
  16                       void (*emit)(void *ctx, long int output), void *emitctx)
  17 {
  18     UNUSEDARG(charset);
  19
  20     /*
  21      * For reading Big5, state->s0 simply contains the single
  22      * stored lead byte when we are half way through a double-byte
  23      * character, or 0 if we aren't.
  24      */
  25
  26     if (state->s0 == 0) {
  27         if (input_chr >= 0xA1 && input_chr <= 0xFE) {
  28             /*
  29              * Lead byte. Just store it.
  30              */
  31             state->s0 = input_chr;
  32         } else {
  33             /*
  34              * Anything else we pass straight through unchanged.
  35              */
  36             emit(emitctx, input_chr);
  37         }
  38     } else {
  39         /*
  40          * We have a stored lead byte. We expect a valid followup
  41          * byte.
  42          */
  43         if ((input_chr >= 0x40 && input_chr <= 0x7E) ||
  44             (input_chr >= 0xA1 && input_chr <= 0xFE)) {
  45             emit(emitctx, big5_to_unicode(state->s0 - 0xA1, input_chr - 0x40));
  46         } else {
  47             emit(emitctx, ERROR);
  48         }
  49         state->s0 = 0;
  50     }
  51 }
  52
  53 /*
  54  * Big5 is a stateless multi-byte encoding (in the sense that just
  55  * after any character has been completed, the state is always the
  56  * same); hence when writing it, there is no need to use the
  57  * charset_state.
  58  */
  59
  60 static int write_big5(charset_spec const *charset, long int input_chr,
  61                       charset_state *state,
  62                       void (*emit)(void *ctx, long int output), void *emitctx)
  63 {
  64     UNUSEDARG(charset);
  65     UNUSEDARG(state);
  66
  67     if (input_chr == -1)
  68         return TRUE;                   /* stateless; no cleanup required */
  69
  70     if (input_chr < 0x80) {
  71         emit(emitctx, input_chr);
  72         return TRUE;
  73     } else {
  74         int r, c;
  75         if (unicode_to_big5(input_chr, &r, &c)) {
  76             emit(emitctx, r + 0xA1);
  77             emit(emitctx, c + 0x40);
  78             return TRUE;
  79         } else {
  80             return FALSE;
  81         }
  82     }
  83 }
  84
  85 const charset_spec charset_CS_BIG5 = {
  86     CS_BIG5, read_big5, write_big5, NULL
  87 };
  88
  89 #else /* ENUM_CHARSETS */
  90
  91 ENUM_CHARSET(CS_BIG5)
  92
  93 #endif /* ENUM_CHARSETS */