xref: /freebsd/contrib/xz/src/common/tuklib_mbstr_width.c (revision fe815331bb40604ba31312acf7e4619674631777)
1 ///////////////////////////////////////////////////////////////////////////////
2 //
3 /// \file       tuklib_mbstr_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 #include <string.h>
15 
16 #if defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH)
17 #	include <wchar.h>
18 #endif
19 
20 
21 extern size_t
22 tuklib_mbstr_width(const char *str, size_t *bytes)
23 {
24 	const size_t len = strlen(str);
25 	if (bytes != NULL)
26 		*bytes = len;
27 
28 #if !(defined(HAVE_MBRTOWC) && defined(HAVE_WCWIDTH))
29 	// In single-byte mode, the width of the string is the same
30 	// as its length.
31 	return len;
32 
33 #else
34 	mbstate_t state;
35 	memset(&state, 0, sizeof(state));
36 
37 	size_t width = 0;
38 	size_t i = 0;
39 
40 	// Convert one multibyte character at a time to wchar_t
41 	// and get its width using wcwidth().
42 	while (i < len) {
43 		wchar_t wc;
44 		const size_t ret = mbrtowc(&wc, str + i, len - i, &state);
45 		if (ret < 1 || ret > len)
46 			return (size_t)-1;
47 
48 		i += ret;
49 
50 		const int wc_width = wcwidth(wc);
51 		if (wc_width < 0)
52 			return (size_t)-1;
53 
54 		width += (size_t)wc_width;
55 	}
56 
57 	// Require that the string ends in the initial shift state.
58 	// This way the caller can be combine the string with other
59 	// strings without needing to worry about the shift states.
60 	if (!mbsinit(&state))
61 		return (size_t)-1;
62 
63 	return width;
64 #endif
65 }
66