1186681Sed/* 2186681Sed * Markus Kuhn -- 2007-05-26 (Unicode 5.0) 3186681Sed * 4186681Sed * Permission to use, copy, modify, and distribute this software 5186681Sed * for any purpose and without fee is hereby granted. The author 6186681Sed * disclaims all warranties with regard to this software. 7186681Sed * 8186681Sed * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c 9186681Sed * 10186681Sed * $FreeBSD$ 11186681Sed */ 12186681Sed 13186681Sedstruct interval { 14186681Sed teken_char_t first; 15186681Sed teken_char_t last; 16186681Sed}; 17186681Sed 18186681Sed/* auxiliary function for binary search in interval table */ 19186681Sedstatic int bisearch(teken_char_t ucs, const struct interval *table, int max) { 20186681Sed int min = 0; 21186681Sed int mid; 22186681Sed 23186681Sed if (ucs < table[0].first || ucs > table[max].last) 24186681Sed return 0; 25186681Sed while (max >= min) { 26186681Sed mid = (min + max) / 2; 27186681Sed if (ucs > table[mid].last) 28186681Sed min = mid + 1; 29186681Sed else if (ucs < table[mid].first) 30186681Sed max = mid - 1; 31186681Sed else 32186681Sed return 1; 33186681Sed } 34186681Sed 35186681Sed return 0; 36186681Sed} 37186681Sed 38186681Sedstatic int teken_wcwidth(teken_char_t ucs) 39186681Sed{ 40186681Sed /* sorted list of non-overlapping intervals of non-spacing characters */ 41186681Sed /* generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c" */ 42186681Sed static const struct interval combining[] = { 43186681Sed { 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 }, 44186681Sed { 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 }, 45186681Sed { 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 }, 46186681Sed { 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 }, 47186681Sed { 0x06D6, 0x06E4 }, { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, 48186681Sed { 0x070F, 0x070F }, { 0x0711, 0x0711 }, { 0x0730, 0x074A }, 49186681Sed { 0x07A6, 0x07B0 }, { 0x07EB, 0x07F3 }, { 0x0901, 0x0902 }, 50186681Sed { 0x093C, 0x093C }, { 0x0941, 0x0948 }, { 0x094D, 0x094D }, 51186681Sed { 0x0951, 0x0954 }, { 0x0962, 0x0963 }, { 0x0981, 0x0981 }, 52186681Sed { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD }, 53186681Sed { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 }, { 0x0A3C, 0x0A3C }, 54186681Sed { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D }, 55186681Sed { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 }, { 0x0ABC, 0x0ABC }, 56186681Sed { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 }, { 0x0ACD, 0x0ACD }, 57186681Sed { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 }, { 0x0B3C, 0x0B3C }, 58186681Sed { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 }, { 0x0B4D, 0x0B4D }, 59186681Sed { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 }, { 0x0BC0, 0x0BC0 }, 60186681Sed { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 }, { 0x0C46, 0x0C48 }, 61186681Sed { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 }, { 0x0CBC, 0x0CBC }, 62186681Sed { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD }, 63186681Sed { 0x0CE2, 0x0CE3 }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D }, 64186681Sed { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 }, 65186681Sed { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E }, 66186681Sed { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC }, 67186681Sed { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 }, 68186681Sed { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E }, 69186681Sed { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 }, 70186681Sed { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 }, 71186681Sed { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 }, 72186681Sed { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x135F, 0x135F }, 73186681Sed { 0x1712, 0x1714 }, { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, 74186681Sed { 0x1772, 0x1773 }, { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, 75186681Sed { 0x17C6, 0x17C6 }, { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, 76186681Sed { 0x180B, 0x180D }, { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, 77186681Sed { 0x1927, 0x1928 }, { 0x1932, 0x1932 }, { 0x1939, 0x193B }, 78186681Sed { 0x1A17, 0x1A18 }, { 0x1B00, 0x1B03 }, { 0x1B34, 0x1B34 }, 79186681Sed { 0x1B36, 0x1B3A }, { 0x1B3C, 0x1B3C }, { 0x1B42, 0x1B42 }, 80186681Sed { 0x1B6B, 0x1B73 }, { 0x1DC0, 0x1DCA }, { 0x1DFE, 0x1DFF }, 81186681Sed { 0x200B, 0x200F }, { 0x202A, 0x202E }, { 0x2060, 0x2063 }, 82186681Sed { 0x206A, 0x206F }, { 0x20D0, 0x20EF }, { 0x302A, 0x302F }, 83186681Sed { 0x3099, 0x309A }, { 0xA806, 0xA806 }, { 0xA80B, 0xA80B }, 84186681Sed { 0xA825, 0xA826 }, { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, 85186681Sed { 0xFE20, 0xFE23 }, { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, 86186681Sed { 0x10A01, 0x10A03 }, { 0x10A05, 0x10A06 }, { 0x10A0C, 0x10A0F }, 87186681Sed { 0x10A38, 0x10A3A }, { 0x10A3F, 0x10A3F }, { 0x1D167, 0x1D169 }, 88186681Sed { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD }, 89186681Sed { 0x1D242, 0x1D244 }, { 0xE0001, 0xE0001 }, { 0xE0020, 0xE007F }, 90186681Sed { 0xE0100, 0xE01EF } 91186681Sed }; 92186681Sed 93186681Sed /* test for 8-bit control characters */ 94186681Sed if (ucs == 0) 95186681Sed return 0; 96186681Sed if (ucs < 32 || (ucs >= 0x7f && ucs < 0xa0)) 97186681Sed return -1; 98186681Sed 99186681Sed /* binary search in table of non-spacing characters */ 100186681Sed if (bisearch(ucs, combining, 101186681Sed sizeof(combining) / sizeof(struct interval) - 1)) 102186681Sed return 0; 103186681Sed 104186681Sed /* if we arrive here, ucs is not a combining or C0/C1 control character */ 105186681Sed 106186681Sed return 1 + 107186681Sed (ucs >= 0x1100 && 108186681Sed (ucs <= 0x115f || /* Hangul Jamo init. consonants */ 109186681Sed ucs == 0x2329 || ucs == 0x232a || 110186681Sed (ucs >= 0x2e80 && ucs <= 0xa4cf && 111186681Sed ucs != 0x303f) || /* CJK ... Yi */ 112186681Sed (ucs >= 0xac00 && ucs <= 0xd7a3) || /* Hangul Syllables */ 113186681Sed (ucs >= 0xf900 && ucs <= 0xfaff) || /* CJK Compatibility Ideographs */ 114186681Sed (ucs >= 0xfe10 && ucs <= 0xfe19) || /* Vertical forms */ 115186681Sed (ucs >= 0xfe30 && ucs <= 0xfe6f) || /* CJK Compatibility Forms */ 116186681Sed (ucs >= 0xff00 && ucs <= 0xff60) || /* Fullwidth Forms */ 117186681Sed (ucs >= 0xffe0 && ucs <= 0xffe6) || 118186681Sed (ucs >= 0x20000 && ucs <= 0x2fffd) || 119186681Sed (ucs >= 0x30000 && ucs <= 0x3fffd))); 120186681Sed} 121