xref: /freebsd/contrib/xz/src/common/tuklib_mbstr_width.c (revision fcb560670601b2a4d87bb31d7531c8dcc37ee71b)
1 ///////////////////////////////////////////////////////////////////////////////
2 //
3 /// \file       tuklib_mstr_width.c
4 /// \brief      Calculate width of a multibyte string
5 //
6 //  Author:     Lasse Collin
7 //
8 //  This file has been put into the public domain.
9 //  You can do whatever you want with this file.
10 //
11 ///////////////////////////////////////////////////////////////////////////////
12 
13 #include "tuklib_mbstr.h"
14 
15 #if defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
16 #	include <wchar.h>
17 #endif
18 
19 
20 extern size_t
21 tuklib_mbstr_width(const char *str, size_t *bytes)
22 {
23 	const size_t len = strlen(str);
24 	if (bytes != NULL)
25 		*bytes = len;
26 
27 #if !(defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH))
28 	// In single-byte mode, the width of the string is the same
29 	// as its length.
30 	return len;
31 
32 #else
33 	mbstate_t state;
34 	memset(&state, 0, sizeof(state));
35 
36 	size_t width = 0;
37 	size_t i = 0;
38 
39 	// Convert one multibyte character at a time to wchar_t
40 	// and get its width using wcwidth().
41 	while (i < len) {
42 		wchar_t wc;
43 		const size_t ret = mbrtowc(&wc, str + i, len - i, &state);
44 		if (ret < 1 || ret > len)
45 			return (size_t)-1;
46 
47 		i += ret;
48 
49 		const int wc_width = wcwidth(wc);
50 		if (wc_width < 0)
51 			return (size_t)-1;
52 
53 		width += wc_width;
54 	}
55 
56 	// Require that the string ends in the initial shift state.
57 	// This way the caller can be combine the string with other
58 	// strings without needing to worry about the shift states.
59 	if (!mbsinit(&state))
60 		return (size_t)-1;
61 
62 	return width;
63 #endif
64 }
65