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 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 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 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