xref: /freebsd/sys/teken/teken_wcwidth.h (revision 90367ba750bcbf3f9ac4609c3ec8df4ab95a22af)
1 /*
2  * Markus Kuhn -- 2007-05-26 (Unicode 5.0)
3  *
4  * Permission to use, copy, modify, and distribute this software
5  * for any purpose and without fee is hereby granted. The author
6  * disclaims all warranties with regard to this software.
7  *
8  * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c
9  */
10 
11 #define TEKEN_UTF8_INVALID_CODEPOINT -1
12 
13 struct interval {
14 	teken_char_t first;
15 	teken_char_t last;
16 };
17 
18 /* Auxiliary function for binary search in interval table */
19 static inline int
bisearch(teken_char_t ucs,const struct interval * table,int max)20 bisearch(teken_char_t ucs, const struct interval *table, int max)
21 {
22 	int min = 0;
23 	int mid;
24 
25 	if (ucs < table[0].first || ucs > table[max].last)
26 		return (0);
27 	while (max >= min) {
28 		mid = (min + max) / 2;
29 		if (ucs > table[mid].last)
30 			min = mid + 1;
31 		else if (ucs < table[mid].first)
32 			max = mid - 1;
33 		else
34 			return (1);
35 	}
36 
37 	return (0);
38 }
39 
40 static inline int
teken_wcwidth(teken_char_t ucs)41 teken_wcwidth(teken_char_t ucs)
42 {
43 	/*
44 	 * Sorted list of non-overlapping intervals of non-spacing characters.
45 	 * Generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c"
46 	 */
47 	static const struct interval combining[] = {
48 		{ 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 },
49 		{ 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
50 		{ 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 },
51 		{ 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 },
52 		{ 0x06D6, 0x06E4 }, { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED },
53 		{ 0x070F, 0x070F }, { 0x0711, 0x0711 }, { 0x0730, 0x074A },
54 		{ 0x07A6, 0x07B0 }, { 0x07EB, 0x07F3 }, { 0x0901, 0x0902 },
55 		{ 0x093C, 0x093C }, { 0x0941, 0x0948 }, { 0x094D, 0x094D },
56 		{ 0x0951, 0x0954 }, { 0x0962, 0x0963 }, { 0x0981, 0x0981 },
57 		{ 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD },
58 		{ 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 }, { 0x0A3C, 0x0A3C },
59 		{ 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D },
60 		{ 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 }, { 0x0ABC, 0x0ABC },
61 		{ 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 }, { 0x0ACD, 0x0ACD },
62 		{ 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 }, { 0x0B3C, 0x0B3C },
63 		{ 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 }, { 0x0B4D, 0x0B4D },
64 		{ 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 }, { 0x0BC0, 0x0BC0 },
65 		{ 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 }, { 0x0C46, 0x0C48 },
66 		{ 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 }, { 0x0CBC, 0x0CBC },
67 		{ 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD },
68 		{ 0x0CE2, 0x0CE3 }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
69 		{ 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
70 		{ 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
71 		{ 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
72 		{ 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
73 		{ 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
74 		{ 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
75 		{ 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
76 		{ 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
77 		{ 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x135F, 0x135F },
78 		{ 0x1712, 0x1714 }, { 0x1732, 0x1734 }, { 0x1752, 0x1753 },
79 		{ 0x1772, 0x1773 }, { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD },
80 		{ 0x17C6, 0x17C6 }, { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD },
81 		{ 0x180B, 0x180D }, { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 },
82 		{ 0x1927, 0x1928 }, { 0x1932, 0x1932 }, { 0x1939, 0x193B },
83 		{ 0x1A17, 0x1A18 }, { 0x1B00, 0x1B03 }, { 0x1B34, 0x1B34 },
84 		{ 0x1B36, 0x1B3A }, { 0x1B3C, 0x1B3C }, { 0x1B42, 0x1B42 },
85 		{ 0x1B6B, 0x1B73 }, { 0x1DC0, 0x1DCA }, { 0x1DFE, 0x1DFF },
86 		{ 0x200B, 0x200F }, { 0x202A, 0x202E }, { 0x2060, 0x2063 },
87 		{ 0x206A, 0x206F }, { 0x20D0, 0x20EF }, { 0x302A, 0x302F },
88 		{ 0x3099, 0x309A }, { 0xA806, 0xA806 }, { 0xA80B, 0xA80B },
89 		{ 0xA825, 0xA826 }, { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F },
90 		{ 0xFE20, 0xFE23 }, { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB },
91 		{ 0x10A01, 0x10A03 }, { 0x10A05, 0x10A06 }, { 0x10A0C, 0x10A0F },
92 		{ 0x10A38, 0x10A3A }, { 0x10A3F, 0x10A3F }, { 0x1D167, 0x1D169 },
93 		{ 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD },
94 		{ 0x1D242, 0x1D244 }, { 0xE0001, 0xE0001 }, { 0xE0020, 0xE007F },
95 		{ 0xE0100, 0xE01EF }
96 	};
97 
98 	/* Test for 8-bit control characters */
99 	if (ucs == 0)
100 		return (0);
101 	if (ucs < 32 || (ucs >= 0x7f && ucs < 0xa0))
102 		return (-1);
103 
104 	/* Binary search in table of non-spacing characters */
105 	if (bisearch(ucs, combining, sizeof(combining) /
106 	    sizeof(struct interval) - 1))
107 		return (0);
108 
109 	/*
110 	 * If we arrive here, ucs is not a combining or C0/C1 control character
111 	 */
112 	return (1 +
113 	    (ucs >= 0x1100 &&
114 	    (ucs <= 0x115f ||			/* Hangul Jamo init. consonants */
115 	    ucs == 0x2329 || ucs == 0x232a ||
116 	    (ucs >= 0x2e80 && ucs <= 0xa4cf &&
117 	    ucs != 0x303f) ||			/* CJK ... Yi */
118 	    (ucs >= 0xac00 && ucs <= 0xd7a3) ||	/* Hangul Syllables */
119 	    (ucs >= 0xf900 && ucs <= 0xfaff) ||	/* CJK Compatibility Ideographs */
120 	    (ucs >= 0xfe10 && ucs <= 0xfe19) ||	/* Vertical forms */
121 	    (ucs >= 0xfe30 && ucs <= 0xfe6f) ||	/* CJK Compatibility Forms */
122 	    (ucs >= 0xff00 && ucs <= 0xff60) ||	/* Fullwidth Forms */
123 	    (ucs >= 0xffe0 && ucs <= 0xffe6) ||
124 	    (ucs >= 0x20000 && ucs <= 0x2fffd) ||
125 	    (ucs >= 0x30000 && ucs <= 0x3fffd))));
126 }
127 
128 /*
129  * Converts an UTF-8 byte sequence to a codepoint as specified in
130  * https://datatracker.ietf.org/doc/html/rfc3629#section-3 . The function
131  * expects the 'bytes' array to start with the leading character.
132  */
133 static inline teken_char_t
teken_utf8_bytes_to_codepoint(uint8_t bytes[4],int nbytes)134 teken_utf8_bytes_to_codepoint(uint8_t bytes[4], int nbytes)
135 {
136 	/*
137 	 * Check for malformed characters by comparing 'nbytes' to the byte
138 	 * length of the character.
139 	 *
140 	 * The table in section 3 of RFC 3629 defines 4 different values
141 	 * indicating the length of a UTF-8 byte sequence.
142 	 *
143 	 * 0xxxxxxx -> 1 byte
144 	 * 110xxxxx -> 2 bytes
145 	 * 1110xxxx -> 3 bytes
146 	 * 11110xxx -> 4 bytes
147 	 *
148 	 * The length is determined by the higher-order bits in the leading
149 	 * octet (except in the first case, where an MSB of 0 means a byte
150 	 * length of 1). Here we flip the 4 upper bits and count the leading
151 	 * zeros using __builtin_clz() to determine the number of bytes.
152 	 */
153 	if (__builtin_clz(~(bytes[0] & 0xf0) << 24) != nbytes)
154 		return (TEKEN_UTF8_INVALID_CODEPOINT);
155 
156 	switch (nbytes) {
157 	case 1:
158 		return (bytes[0] & 0x7f);
159 	case 2:
160 		return (bytes[0] & 0x1f) << 6 | (bytes[1] & 0x3f);
161 	case 3:
162 		return (bytes[0] & 0xf) << 12 | (bytes[1] & 0x3f) << 6 |
163 		    (bytes[2] & 0x3f);
164 	case 4:
165 		return (bytes[0] & 0x7) << 18 | (bytes[1] & 0x3f) << 12 |
166 		    (bytes[2] & 0x3f) << 6 | (bytes[3] & 0x3f);
167 	default:
168 		return (TEKEN_UTF8_INVALID_CODEPOINT);
169 	}
170 }
171