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 int writtenlen;
12 const wchar_t *errstr;
13 int errlen;
14 int stopped;
15 };
16
unicode_emit(void * ctx,long int output)17 static void unicode_emit(void *ctx, long int output)
18 {
19 struct unicode_emit_param *param = (struct unicode_emit_param *)ctx;
20 wchar_t outval;
21 wchar_t const *p;
22 int outlen;
23
24 if (output == ERROR) {
25 if (param->errstr) {
26 p = param->errstr;
27 outlen = param->errlen;
28 } else {
29 outval = 0xFFFD; /* U+FFFD REPLACEMENT CHARACTER */
30 p = &outval;
31 outlen = 1;
32 }
33 } else {
34 outval = output;
35 p = &outval;
36 outlen = 1;
37 }
38
39 if (param->outlen < 0 || param->outlen >= outlen) {
40 while (outlen > 0) {
41 if (param->output)
42 *param->output++ = *p++;
43 if (param->outlen > 0)
44 param->outlen--;
45 outlen--;
46 param->writtenlen++;
47 }
48 } else {
49 param->stopped = 1;
50 }
51 }
52
charset_to_unicode(const char ** input,int * inlen,wchar_t * output,int outlen,int charset,charset_state * state,const wchar_t * errstr,int errlen)53 int charset_to_unicode(const char **input, int *inlen,
54 wchar_t *output, int outlen,
55 int charset, charset_state *state,
56 const wchar_t *errstr, int errlen)
57 {
58 charset_spec const *spec = charset_find_spec(charset);
59 charset_state localstate = CHARSET_INIT_STATE;
60 struct unicode_emit_param param;
61
62 param.output = output;
63 param.outlen = outlen;
64 param.errstr = errstr;
65 param.errlen = errlen;
66 param.writtenlen = 0;
67 param.stopped = 0;
68
69 if (state)
70 localstate = *state; /* structure copy */
71
72 while (*inlen > 0) {
73 int lenbefore = param.writtenlen;
74 spec->read(spec, (unsigned char)**input, &localstate,
75 unicode_emit, ¶m);
76 if (param.stopped) {
77 /*
78 * The emit function has _tried_ to output some
79 * characters, but ran up against the end of the
80 * buffer. Leave immediately, and return what happened
81 * _before_ attempting to process this character.
82 */
83 return lenbefore;
84 }
85 if (state)
86 *state = localstate; /* structure copy */
87 (*input)++;
88 (*inlen)--;
89 }
90
91 return param.writtenlen;
92 }
93