apps/konsole/src/konsole_wcwidth.cpp

   1 /* $XFree86: xc/programs/xterm/wcwidth.character,v 1.3 2001/07/29 22:08:16 tsi Exp $ */
   2 /*
   3  * This is an implementation of wcwidth() and wcswidth() as defined in
   4  * "The Single UNIX Specification, Version 2, The Open Group, 1997"
   5  * <http://www.UNIX-systems.org/online.html>
   6  *
   7  * Markus Kuhn -- 2001-01-12 -- public domain
   8  */
   9
  10 #include "konsole_wcwidth.h"
  11
  12 struct interval {
  13   unsigned short first;
  14   unsigned short last;
  15 };
  16
  17 /* auxiliary function for binary search in interval table */
  18 static int bisearch(quint16 ucs, const struct interval *table, int max) {
  19   int min = 0;
  20   int mid;
  21
  22   if (ucs < table[0].first || ucs > table[max].last)
  23     return 0;
  24   while (max >= min) {
  25     mid = (min + max) / 2;
  26     if (ucs > table[mid].last)
  27       min = mid + 1;
  28     else if (ucs < table[mid].first)
  29       max = mid - 1;
  30     else
  31       return 1;
  32   }
  33
  34   return 0;
  35 }
  36
  37
  38 /* The following functions define the column width of an ISO 10646
  39  * character as follows:
  40  *
  41  *    - The null character (U+0000) has a column width of 0.
  42  *
  43  *    - Other C0/C1 control characters and DEL will lead to a return
  44  *      value of -1.
  45  *
  46  *    - Non-spacing and enclosing combining characters (general
  47  *      category code Mn or Me in the Unicode database) have a
  48  *      column width of 0.
  49  *
  50  *    - Other format characters (general category code Cf in the Unicode
  51  *      database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
  52  *
  53  *    - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
  54  *      have a column width of 0.
  55  *
  56  *    - Spacing characters in the East Asian Wide (W) or East Asian
  57  *      FullWidth (F) category as defined in Unicode Technical
  58  *      Report #11 have a column width of 2.
  59  *
  60  *    - All remaining characters (including all printable
  61  *      ISO 8859-1 and WGL4 characters, Unicode control characters,
  62  *      etc.) have a column width of 1.
  63  *
  64  * This implementation assumes that quint16 characters are encoded
  65  * in ISO 10646.
  66  */
  67
  68 int konsole_wcwidth(quint16 ucs)
  69 {
  70   /* sorted list of non-overlapping intervals of non-spacing characters */
  71   static const struct interval combining[] = {
  72     { 0x0300, 0x034E }, { 0x0360, 0x0362 }, { 0x0483, 0x0486 },
  73     { 0x0488, 0x0489 }, { 0x0591, 0x05A1 }, { 0x05A3, 0x05B9 },
  74     { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
  75     { 0x05C4, 0x05C4 }, { 0x064B, 0x0655 }, { 0x0670, 0x0670 },
  76     { 0x06D6, 0x06E4 }, { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED },
  77     { 0x070F, 0x070F }, { 0x0711, 0x0711 }, { 0x0730, 0x074A },
  78     { 0x07A6, 0x07B0 }, { 0x0901, 0x0902 }, { 0x093C, 0x093C },
  79     { 0x0941, 0x0948 }, { 0x094D, 0x094D }, { 0x0951, 0x0954 },
  80     { 0x0962, 0x0963 }, { 0x0981, 0x0981 }, { 0x09BC, 0x09BC },
  81     { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 },
  82     { 0x0A02, 0x0A02 }, { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 },
  83     { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D }, { 0x0A70, 0x0A71 },
  84     { 0x0A81, 0x0A82 }, { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 },
  85     { 0x0AC7, 0x0AC8 }, { 0x0ACD, 0x0ACD }, { 0x0B01, 0x0B01 },
  86     { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 },
  87     { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 },
  88     { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 },
  89     { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 },
  90     { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD },
  91     { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D }, { 0x0DCA, 0x0DCA },
  92     { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 }, { 0x0E31, 0x0E31 },
  93     { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E }, { 0x0EB1, 0x0EB1 },
  94     { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC }, { 0x0EC8, 0x0ECD },
  95     { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 }, { 0x0F37, 0x0F37 },
  96     { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E }, { 0x0F80, 0x0F84 },
  97     { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 }, { 0x0F99, 0x0FBC },
  98     { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 }, { 0x1032, 0x1032 },
  99     { 0x1036, 0x1037 }, { 0x1039, 0x1039 }, { 0x1058, 0x1059 },
 100     { 0x1160, 0x11FF }, { 0x17B7, 0x17BD }, { 0x17C6, 0x17C6 },
 101     { 0x17C9, 0x17D3 }, { 0x180B, 0x180E }, { 0x18A9, 0x18A9 },
 102     { 0x200B, 0x200F }, { 0x202A, 0x202E }, { 0x206A, 0x206F },
 103     { 0x20D0, 0x20E3 }, { 0x302A, 0x302F }, { 0x3099, 0x309A },
 104     { 0xFB1E, 0xFB1E }, { 0xFE20, 0xFE23 }, { 0xFEFF, 0xFEFF },
 105     { 0xFFF9, 0xFFFB }
 106   };
 107
 108   /* test for 8-bit control characters */
 109   if (ucs == 0)
 110     return 0;
 111   if (ucs < 32 || (ucs >= 0x7f && ucs < 0xa0))
 112     return -1;
 113
 114   /* binary search in table of non-spacing characters */
 115   if (bisearch(ucs, combining,
 116            sizeof(combining) / sizeof(struct interval) - 1))
 117     return 0;
 118
 119   /* if we arrive here, ucs is not a combining or C0/C1 control character */
 120
 121   return 1 +
 122     (ucs >= 0x1100 &&
 123      (ucs <= 0x115f ||                    /* Hangul Jamo init. consonants */
 124       (ucs >= 0x2e80 && ucs <= 0xa4cf && (ucs & ~0x0011) != 0x300a &&
 125        ucs != 0x303f) ||                  /* CJK ... Yi */
 126       (ucs >= 0xac00 && ucs <= 0xd7a3) || /* Hangul Syllables */
 127       (ucs >= 0xf900 && ucs <= 0xfaff) || /* CJK Compatibility Ideographs */
 128       (ucs >= 0xfe30 && ucs <= 0xfe6f) || /* CJK Compatibility Forms */
 129       (ucs >= 0xff00 && ucs <= 0xff5f) || /* Fullwidth Forms */
 130       (ucs >= 0xffe0 && ucs <= 0xffe6) /* do not compare UINT16 with 0x20000 ||
 131       (ucs >= 0x20000 && ucs <= 0x2ffff) */));
 132 }
 133
 134 #if 0
 135 /*
 136  * The following function is the same as konsole_wcwidth(), except that
 137  * spacing characters in the East Asian Ambiguous (A) category as
 138  * defined in Unicode Technical Report #11 have a column width of 2.
 139  * This experimental variant might be useful for users of CJK legacy
 140  * encodings who want to migrate to UCS. It is not otherwise
 141  * recommended for general use.
 142  */
 143 int konsole_wcwidth_cjk(quint16 ucs)
 144 {
 145   /* sorted list of non-overlapping intervals of East Asian Ambiguous
 146    * characters */
 147   static const struct interval ambiguous[] = {
 148     { 0x00A1, 0x00A1 }, { 0x00A4, 0x00A4 }, { 0x00A7, 0x00A8 },
 149     { 0x00AA, 0x00AA }, { 0x00AD, 0x00AD }, { 0x00B0, 0x00B4 },
 150     { 0x00B6, 0x00BA }, { 0x00BC, 0x00BF }, { 0x00C6, 0x00C6 },
 151     { 0x00D0, 0x00D0 }, { 0x00D7, 0x00D8 }, { 0x00DE, 0x00E1 },
 152     { 0x00E6, 0x00E6 }, { 0x00E8, 0x00EA }, { 0x00EC, 0x00ED },
 153     { 0x00F0, 0x00F0 }, { 0x00F2, 0x00F3 }, { 0x00F7, 0x00FA },
 154     { 0x00FC, 0x00FC }, { 0x00FE, 0x00FE }, { 0x0101, 0x0101 },
 155     { 0x0111, 0x0111 }, { 0x0113, 0x0113 }, { 0x011B, 0x011B },
 156     { 0x0126, 0x0127 }, { 0x012B, 0x012B }, { 0x0131, 0x0133 },
 157     { 0x0138, 0x0138 }, { 0x013F, 0x0142 }, { 0x0144, 0x0144 },
 158     { 0x0148, 0x014A }, { 0x014D, 0x014D }, { 0x0152, 0x0153 },
 159     { 0x0166, 0x0167 }, { 0x016B, 0x016B }, { 0x01CE, 0x01CE },
 160     { 0x01D0, 0x01D0 }, { 0x01D2, 0x01D2 }, { 0x01D4, 0x01D4 },
 161     { 0x01D6, 0x01D6 }, { 0x01D8, 0x01D8 }, { 0x01DA, 0x01DA },
 162     { 0x01DC, 0x01DC }, { 0x0251, 0x0251 }, { 0x0261, 0x0261 },
 163     { 0x02C7, 0x02C7 }, { 0x02C9, 0x02CB }, { 0x02CD, 0x02CD },
 164     { 0x02D0, 0x02D0 }, { 0x02D8, 0x02DB }, { 0x02DD, 0x02DD },
 165     { 0x0391, 0x03A1 }, { 0x03A3, 0x03A9 }, { 0x03B1, 0x03C1 },
 166     { 0x03C3, 0x03C9 }, { 0x0401, 0x0401 }, { 0x0410, 0x044F },
 167     { 0x0451, 0x0451 }, { 0x2010, 0x2010 }, { 0x2013, 0x2016 },
 168     { 0x2018, 0x2019 }, { 0x201C, 0x201D }, { 0x2020, 0x2021 },
 169     { 0x2025, 0x2027 }, { 0x2030, 0x2030 }, { 0x2032, 0x2033 },
 170     { 0x2035, 0x2035 }, { 0x203B, 0x203B }, { 0x2074, 0x2074 },
 171     { 0x207F, 0x207F }, { 0x2081, 0x2084 }, { 0x20AC, 0x20AC },
 172     { 0x2103, 0x2103 }, { 0x2105, 0x2105 }, { 0x2109, 0x2109 },
 173     { 0x2113, 0x2113 }, { 0x2121, 0x2122 }, { 0x2126, 0x2126 },
 174     { 0x212B, 0x212B }, { 0x2154, 0x2155 }, { 0x215B, 0x215B },
 175     { 0x215E, 0x215E }, { 0x2160, 0x216B }, { 0x2170, 0x2179 },
 176     { 0x2190, 0x2199 }, { 0x21D2, 0x21D2 }, { 0x21D4, 0x21D4 },
 177     { 0x2200, 0x2200 }, { 0x2202, 0x2203 }, { 0x2207, 0x2208 },
 178     { 0x220B, 0x220B }, { 0x220F, 0x220F }, { 0x2211, 0x2211 },
 179     { 0x2215, 0x2215 }, { 0x221A, 0x221A }, { 0x221D, 0x2220 },
 180     { 0x2223, 0x2223 }, { 0x2225, 0x2225 }, { 0x2227, 0x222C },
 181     { 0x222E, 0x222E }, { 0x2234, 0x2237 }, { 0x223C, 0x223D },
 182     { 0x2248, 0x2248 }, { 0x224C, 0x224C }, { 0x2252, 0x2252 },
 183     { 0x2260, 0x2261 }, { 0x2264, 0x2267 }, { 0x226A, 0x226B },
 184     { 0x226E, 0x226F }, { 0x2282, 0x2283 }, { 0x2286, 0x2287 },
 185     { 0x2295, 0x2295 }, { 0x2299, 0x2299 }, { 0x22A5, 0x22A5 },
 186     { 0x22BF, 0x22BF }, { 0x2312, 0x2312 }, { 0x2460, 0x24BF },
 187     { 0x24D0, 0x24E9 }, { 0x2500, 0x254B }, { 0x2550, 0x2574 },
 188     { 0x2580, 0x258F }, { 0x2592, 0x2595 }, { 0x25A0, 0x25A1 },
 189     { 0x25A3, 0x25A9 }, { 0x25B2, 0x25B3 }, { 0x25B6, 0x25B7 },
 190     { 0x25BC, 0x25BD }, { 0x25C0, 0x25C1 }, { 0x25C6, 0x25C8 },
 191     { 0x25CB, 0x25CB }, { 0x25CE, 0x25D1 }, { 0x25E2, 0x25E5 },
 192     { 0x25EF, 0x25EF }, { 0x2605, 0x2606 }, { 0x2609, 0x2609 },
 193     { 0x260E, 0x260F }, { 0x261C, 0x261C }, { 0x261E, 0x261E },
 194     { 0x2640, 0x2640 }, { 0x2642, 0x2642 }, { 0x2660, 0x2661 },
 195     { 0x2663, 0x2665 }, { 0x2667, 0x266A }, { 0x266C, 0x266D },
 196     { 0x266F, 0x266F }, { 0x300A, 0x300B }, { 0x301A, 0x301B },
 197     { 0xE000, 0xF8FF }, { 0xFFFD, 0xFFFD }
 198   };
 199
 200   /* binary search in table of non-spacing characters */
 201   if (bisearch(ucs, ambiguous,
 202            sizeof(ambiguous) / sizeof(struct interval) - 1))
 203     return 2;
 204
 205   return konsole_wcwidth(ucs);
 206 }
 207 #endif
 208
 209 // single byte char: +1, multi byte char: +2
 210 int string_width( const QString &txt )
 211 {
 212   int w = 0;
 213   for ( int i = 0; i < txt.length(); ++i )
 214      w += konsole_wcwidth( txt[ i ].unicode() );
 215  return w;
 216 }