charset/toucs.c

   1 /*
   2  * toucs.c - convert charsets to Unicode.
   3  */
   4
   5 #include "charset.h"
   6 #include "internal.h"
   7
   8 struct unicode_emit_param {
   9     wchar_t *output;
  10     int outlen;
  11     const wchar_t *errstr;
  12     int errlen;
  13     int stopped;
  14 };
  15
  16 static void unicode_emit(void *ctx, long int output)
  17 {
  18     struct unicode_emit_param *param = (struct unicode_emit_param *)ctx;
  19     wchar_t outval;
  20     wchar_t const *p;
  21     int outlen;
  22
  23     if (output == ERROR) {
  24         if (param->errstr) {
  25             p = param->errstr;
  26             outlen = param->errlen;
  27         } else {
  28             outval = 0xFFFD;           /* U+FFFD REPLACEMENT CHARACTER */
  29             p = &outval;
  30             outlen = 1;
  31         }
  32     } else {
  33         outval = output;
  34         p = &outval;
  35         outlen = 1;
  36     }
  37
  38     if (param->outlen >= outlen) {
  39         while (outlen > 0) {
  40             *param->output++ = *p++;
  41             param->outlen--;
  42             outlen--;
  43         }
  44     } else {
  45         param->stopped = 1;
  46     }
  47 }
  48
  49 int charset_to_unicode(char **input, int *inlen, wchar_t *output, int outlen,
  50                        int charset, charset_state *state,
  51                        const wchar_t *errstr, int errlen)
  52 {
  53     charset_spec const *spec = charset_find_spec(charset);
  54     charset_state localstate;
  55     struct unicode_emit_param param;
  56
  57     param.output = output;
  58     param.outlen = outlen;
  59     param.errstr = errstr;
  60     param.errlen = errlen;
  61     param.stopped = 0;
  62
  63     if (!state) {
  64         localstate.s0 = 0;
  65     } else {
  66         localstate = *state;           /* structure copy */
  67     }
  68
  69     while (*inlen > 0) {
  70         int lenbefore = param.output - output;
  71         spec->read(spec, (unsigned char)**input, &localstate,
  72                    unicode_emit, &param);
  73         if (param.stopped) {
  74             /*
  75              * The emit function has _tried_ to output some
  76              * characters, but ran up against the end of the
  77              * buffer. Leave immediately, and return what happened
  78              * _before_ attempting to process this character.
  79              */
  80             return lenbefore;
  81         }
  82         if (state)
  83             *state = localstate;   /* structure copy */
  84         (*input)++;
  85         (*inlen)--;
  86     }
  87
  88     return param.output - output;
  89 }