mdw@git.distorted.org.uk Git - sgt/halibut/blob - wcwidth.c

   1 /*
   2  * A wrapper around wcswidth() to have it ignore characters which
   3  * are not represented in a particular output charset.
   4  */
   5
   6 #include <assert.h>
   7 #include <stdlib.h>
   8 #include "halibut.h"
   9
  10 #ifndef HAS_WCSWIDTH
  11
  12 /*
  13  * Markus Kuhn's public domain wcwidth() implementation, cribbed
  14  * from
  15  *
  16  *   http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c
  17  */
  18
  19 #define wcswidth mk_wcswidth
  20
  21 struct interval {
  22     int first;
  23     int last;
  24 };
  25
  26 /* auxiliary function for binary search in interval table */
  27 static int bisearch(wchar_t ucs, const struct interval *table, int max) {
  28     int min = 0;
  29     int mid;
  30
  31     if (ucs < table[0].first || ucs > table[max].last)
  32         return 0;
  33     while (max >= min) {
  34         mid = (min + max) / 2;
  35         if (ucs > table[mid].last)
  36             min = mid + 1;
  37         else if (ucs < table[mid].first)
  38             max = mid - 1;
  39         else
  40             return 1;
  41     }
  42
  43     return 0;
  44 }
  45
  46 int mk_wcwidth(wchar_t ucs)
  47 {
  48     /* sorted list of non-overlapping intervals of non-spacing characters */
  49     /* generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c" */
  50     static const struct interval combining[] = {
  51         { 0x0300, 0x0357 }, { 0x035D, 0x036F }, { 0x0483, 0x0486 },
  52         { 0x0488, 0x0489 }, { 0x0591, 0x05A1 }, { 0x05A3, 0x05B9 },
  53     { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
  54     { 0x05C4, 0x05C4 }, { 0x0600, 0x0603 }, { 0x0610, 0x0615 },
  55     { 0x064B, 0x0658 }, { 0x0670, 0x0670 }, { 0x06D6, 0x06E4 },
  56     { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, { 0x070F, 0x070F },
  57     { 0x0711, 0x0711 }, { 0x0730, 0x074A }, { 0x07A6, 0x07B0 },
  58     { 0x0901, 0x0902 }, { 0x093C, 0x093C }, { 0x0941, 0x0948 },
  59     { 0x094D, 0x094D }, { 0x0951, 0x0954 }, { 0x0962, 0x0963 },
  60     { 0x0981, 0x0981 }, { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 },
  61     { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 },
  62     { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 },
  63     { 0x0A4B, 0x0A4D }, { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 },
  64     { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 },
  65     { 0x0ACD, 0x0ACD }, { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 },
  66     { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 },
  67     { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 },
  68     { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 },
  69     { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 },
  70     { 0x0CBC, 0x0CBC }, { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 },
  71     { 0x0CCC, 0x0CCD }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
  72     { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
  73     { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
  74     { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
  75     { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
  76     { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
  77     { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
  78     { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
  79     { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
  80     { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x1712, 0x1714 },
  81     { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, { 0x1772, 0x1773 },
  82     { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, { 0x17C6, 0x17C6 },
  83     { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, { 0x180B, 0x180D },
  84     { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, { 0x1927, 0x1928 },
  85     { 0x1932, 0x1932 }, { 0x1939, 0x193B }, { 0x200B, 0x200F },
  86     { 0x202A, 0x202E }, { 0x2060, 0x2063 }, { 0x206A, 0x206F },
  87     { 0x20D0, 0x20EA }, { 0x302A, 0x302F }, { 0x3099, 0x309A },
  88     { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, { 0xFE20, 0xFE23 },
  89     { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, { 0x1D167, 0x1D169 },
  90     { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD },
  91     { 0xE0001, 0xE0001 }, { 0xE0020, 0xE007F }, { 0xE0100, 0xE01EF }
  92   };
  93
  94   /* test for 8-bit control characters */
  95   if (ucs == 0)
  96     return 0;
  97   if (ucs < 32 || (ucs >= 0x7f && ucs < 0xa0))
  98     return -1;
  99
 100   /* binary search in table of non-spacing characters */
 101   if (bisearch(ucs, combining,
 102                sizeof(combining) / sizeof(struct interval) - 1))
 103     return 0;
 104
 105   /* if we arrive here, ucs is not a combining or C0/C1 control character */
 106
 107   return 1 +
 108     (ucs >= 0x1100 &&
 109      (ucs <= 0x115f ||                    /* Hangul Jamo init. consonants */
 110       ucs == 0x2329 || ucs == 0x232a ||
 111       (ucs >= 0x2e80 && ucs <= 0xa4cf &&
 112        ucs != 0x303f) ||                  /* CJK ... Yi */
 113       (ucs >= 0xac00 && ucs <= 0xd7a3) || /* Hangul Syllables */
 114       (ucs >= 0xf900 && ucs <= 0xfaff) || /* CJK Compatibility Ideographs */
 115       (ucs >= 0xfe30 && ucs <= 0xfe6f) || /* CJK Compatibility Forms */
 116       (ucs >= 0xff00 && ucs <= 0xff60) || /* Fullwidth Forms */
 117       (ucs >= 0xffe0 && ucs <= 0xffe6) ||
 118       (ucs >= 0x20000 && ucs <= 0x2fffd) ||
 119       (ucs >= 0x30000 && ucs <= 0x3fffd)));
 120 }
 121
 122
 123 int mk_wcswidth(const wchar_t *pwcs, size_t n)
 124 {
 125   int w, width = 0;
 126
 127   for (;*pwcs && n-- > 0; pwcs++)
 128     if ((w = mk_wcwidth(*pwcs)) < 0)
 129       return -1;
 130     else
 131       width += w;
 132
 133   return width;
 134 }
 135
 136 #endif
 137
 138 int ustrwid(wchar_t const *s, int charset)
 139 {
 140     char buf[256];
 141     int wid, len = ustrlen(s);
 142     charset_state state = CHARSET_INIT_STATE;
 143
 144     wid = 0;
 145
 146     while (len > 0) {
 147         int err;
 148         wchar_t const *s_orig;
 149
 150         err = 0;
 151         s_orig = s;
 152         charset_from_unicode(&s, &len, buf, lenof(buf), charset, &state, &err);
 153         wid += wcswidth(s_orig, s - s_orig);
 154         if (err) {
 155             assert(len > 0 && *s);
 156             s++;
 157             len--;
 158         }
 159     }
 160
 161     return wid;
 162 }
 163
 164 int strwid(char const *s, int charset)
 165 {
 166     wchar_t buf[256];
 167     int wid, len = strlen(s);
 168     charset_state state = CHARSET_INIT_STATE;
 169
 170     wid = 0;
 171
 172     while (len > 0) {
 173         int ret;
 174
 175         ret = charset_to_unicode(&s, &len, buf, lenof(buf),
 176                                  charset, &state, NULL, 0);
 177
 178         if (ret > 0)
 179             wid += wcswidth(buf, ret);
 180     }
 181
 182     return wid;
 183 }