dmd/unialpha.c

   1
   2 // Copyright (c) 2003 by Digital Mars
   3 // All Rights Reserved
   4 // written by Walter Bright
   5 // http://www.digitalmars.com
   6 // License for redistribution is by either the Artistic License
   7 // in artistic.txt, or the GNU General Public License in gnu.txt.
   8 // See the included readme.txt for details.
   9
  10 #include <assert.h>
  11
  12 /*******************************
  13  * Return !=0 if unicode alpha.
  14  * Use table from C99 Appendix D.
  15  */
  16
  17 int isUniAlpha(unsigned u)
  18 {
  19     static unsigned short table[][2] =
  20     {
  21         { 0x00AA, 0x00AA },
  22         { 0x00B5, 0x00B5 },
  23         { 0x00B7, 0x00B7 },
  24         { 0x00BA, 0x00BA },
  25         { 0x00C0, 0x00D6 },
  26         { 0x00D8, 0x00F6 },
  27         { 0x00F8, 0x01F5 },
  28         { 0x01FA, 0x0217 },
  29         { 0x0250, 0x02A8 },
  30         { 0x02B0, 0x02B8 },
  31         { 0x02BB, 0x02BB },
  32         { 0x02BD, 0x02C1 },
  33         { 0x02D0, 0x02D1 },
  34         { 0x02E0, 0x02E4 },
  35         { 0x037A, 0x037A },
  36         { 0x0386, 0x0386 },
  37         { 0x0388, 0x038A },
  38         { 0x038C, 0x038C },
  39         { 0x038E, 0x03A1 },
  40         { 0x03A3, 0x03CE },
  41         { 0x03D0, 0x03D6 },
  42         { 0x03DA, 0x03DA },
  43         { 0x03DC, 0x03DC },
  44         { 0x03DE, 0x03DE },
  45         { 0x03E0, 0x03E0 },
  46         { 0x03E2, 0x03F3 },
  47         { 0x0401, 0x040C },
  48         { 0x040E, 0x044F },
  49         { 0x0451, 0x045C },
  50         { 0x045E, 0x0481 },
  51         { 0x0490, 0x04C4 },
  52         { 0x04C7, 0x04C8 },
  53         { 0x04CB, 0x04CC },
  54         { 0x04D0, 0x04EB },
  55         { 0x04EE, 0x04F5 },
  56         { 0x04F8, 0x04F9 },
  57         { 0x0531, 0x0556 },
  58         { 0x0559, 0x0559 },
  59         { 0x0561, 0x0587 },
  60         { 0x05B0, 0x05B9 },
  61         { 0x05BB, 0x05BD },
  62         { 0x05BF, 0x05BF },
  63         { 0x05C1, 0x05C2 },
  64         { 0x05D0, 0x05EA },
  65         { 0x05F0, 0x05F2 },
  66         { 0x0621, 0x063A },
  67         { 0x0640, 0x0652 },
  68         { 0x0660, 0x0669 },
  69         { 0x0670, 0x06B7 },
  70         { 0x06BA, 0x06BE },
  71         { 0x06C0, 0x06CE },
  72         { 0x06D0, 0x06DC },
  73         { 0x06E5, 0x06E8 },
  74         { 0x06EA, 0x06ED },
  75         { 0x06F0, 0x06F9 },
  76         { 0x0901, 0x0903 },
  77         { 0x0905, 0x0939 },
  78         { 0x093D, 0x093D },
  79         { 0x093E, 0x094D },
  80         { 0x0950, 0x0952 },
  81         { 0x0958, 0x0963 },
  82         { 0x0966, 0x096F },
  83         { 0x0981, 0x0983 },
  84         { 0x0985, 0x098C },
  85         { 0x098F, 0x0990 },
  86         { 0x0993, 0x09A8 },
  87         { 0x09AA, 0x09B0 },
  88         { 0x09B2, 0x09B2 },
  89         { 0x09B6, 0x09B9 },
  90         { 0x09BE, 0x09C4 },
  91         { 0x09C7, 0x09C8 },
  92         { 0x09CB, 0x09CD },
  93         { 0x09DC, 0x09DD },
  94         { 0x09DF, 0x09E3 },
  95         { 0x09E6, 0x09EF },
  96         { 0x09F0, 0x09F1 },
  97         { 0x0A02, 0x0A02 },
  98         { 0x0A05, 0x0A0A },
  99         { 0x0A0F, 0x0A10 },
 100         { 0x0A13, 0x0A28 },
 101         { 0x0A2A, 0x0A30 },
 102         { 0x0A32, 0x0A33 },
 103         { 0x0A35, 0x0A36 },
 104         { 0x0A38, 0x0A39 },
 105         { 0x0A3E, 0x0A42 },
 106         { 0x0A47, 0x0A48 },
 107         { 0x0A4B, 0x0A4D },
 108         { 0x0A59, 0x0A5C },
 109         { 0x0A5E, 0x0A5E },
 110         { 0x0A66, 0x0A6F },
 111         { 0x0A74, 0x0A74 },
 112         { 0x0A81, 0x0A83 },
 113         { 0x0A85, 0x0A8B },
 114         { 0x0A8D, 0x0A8D },
 115         { 0x0A8F, 0x0A91 },
 116         { 0x0A93, 0x0AA8 },
 117         { 0x0AAA, 0x0AB0 },
 118         { 0x0AB2, 0x0AB3 },
 119         { 0x0AB5, 0x0AB9 },
 120         { 0x0ABD, 0x0AC5 },
 121         { 0x0AC7, 0x0AC9 },
 122         { 0x0ACB, 0x0ACD },
 123         { 0x0AD0, 0x0AD0 },
 124         { 0x0AE0, 0x0AE0 },
 125         { 0x0AE6, 0x0AEF },
 126         { 0x0B01, 0x0B03 },
 127         { 0x0B05, 0x0B0C },
 128         { 0x0B0F, 0x0B10 },
 129         { 0x0B13, 0x0B28 },
 130         { 0x0B2A, 0x0B30 },
 131         { 0x0B32, 0x0B33 },
 132         { 0x0B36, 0x0B39 },
 133         { 0x0B3D, 0x0B3D },
 134         { 0x0B3E, 0x0B43 },
 135         { 0x0B47, 0x0B48 },
 136         { 0x0B4B, 0x0B4D },
 137         { 0x0B5C, 0x0B5D },
 138         { 0x0B5F, 0x0B61 },
 139         { 0x0B66, 0x0B6F },
 140         { 0x0B82, 0x0B83 },
 141         { 0x0B85, 0x0B8A },
 142         { 0x0B8E, 0x0B90 },
 143         { 0x0B92, 0x0B95 },
 144         { 0x0B99, 0x0B9A },
 145         { 0x0B9C, 0x0B9C },
 146         { 0x0B9E, 0x0B9F },
 147         { 0x0BA3, 0x0BA4 },
 148         { 0x0BA8, 0x0BAA },
 149         { 0x0BAE, 0x0BB5 },
 150         { 0x0BB7, 0x0BB9 },
 151         { 0x0BBE, 0x0BC2 },
 152         { 0x0BC6, 0x0BC8 },
 153         { 0x0BCA, 0x0BCD },
 154         { 0x0BE7, 0x0BEF },
 155         { 0x0C01, 0x0C03 },
 156         { 0x0C05, 0x0C0C },
 157         { 0x0C0E, 0x0C10 },
 158         { 0x0C12, 0x0C28 },
 159         { 0x0C2A, 0x0C33 },
 160         { 0x0C35, 0x0C39 },
 161         { 0x0C3E, 0x0C44 },
 162         { 0x0C46, 0x0C48 },
 163         { 0x0C4A, 0x0C4D },
 164         { 0x0C60, 0x0C61 },
 165         { 0x0C66, 0x0C6F },
 166         { 0x0C82, 0x0C83 },
 167         { 0x0C85, 0x0C8C },
 168         { 0x0C8E, 0x0C90 },
 169         { 0x0C92, 0x0CA8 },
 170         { 0x0CAA, 0x0CB3 },
 171         { 0x0CB5, 0x0CB9 },
 172         { 0x0CBE, 0x0CC4 },
 173         { 0x0CC6, 0x0CC8 },
 174         { 0x0CCA, 0x0CCD },
 175         { 0x0CDE, 0x0CDE },
 176         { 0x0CE0, 0x0CE1 },
 177         { 0x0CE6, 0x0CEF },
 178         { 0x0D02, 0x0D03 },
 179         { 0x0D05, 0x0D0C },
 180         { 0x0D0E, 0x0D10 },
 181         { 0x0D12, 0x0D28 },
 182         { 0x0D2A, 0x0D39 },
 183         { 0x0D3E, 0x0D43 },
 184         { 0x0D46, 0x0D48 },
 185         { 0x0D4A, 0x0D4D },
 186         { 0x0D60, 0x0D61 },
 187         { 0x0D66, 0x0D6F },
 188         { 0x0E01, 0x0E3A },
 189         { 0x0E40, 0x0E5B },
 190 //      { 0x0E50, 0x0E59 },
 191         { 0x0E81, 0x0E82 },
 192         { 0x0E84, 0x0E84 },
 193         { 0x0E87, 0x0E88 },
 194         { 0x0E8A, 0x0E8A },
 195         { 0x0E8D, 0x0E8D },
 196         { 0x0E94, 0x0E97 },
 197         { 0x0E99, 0x0E9F },
 198         { 0x0EA1, 0x0EA3 },
 199         { 0x0EA5, 0x0EA5 },
 200         { 0x0EA7, 0x0EA7 },
 201         { 0x0EAA, 0x0EAB },
 202         { 0x0EAD, 0x0EAE },
 203         { 0x0EB0, 0x0EB9 },
 204         { 0x0EBB, 0x0EBD },
 205         { 0x0EC0, 0x0EC4 },
 206         { 0x0EC6, 0x0EC6 },
 207         { 0x0EC8, 0x0ECD },
 208         { 0x0ED0, 0x0ED9 },
 209         { 0x0EDC, 0x0EDD },
 210         { 0x0F00, 0x0F00 },
 211         { 0x0F18, 0x0F19 },
 212         { 0x0F20, 0x0F33 },
 213         { 0x0F35, 0x0F35 },
 214         { 0x0F37, 0x0F37 },
 215         { 0x0F39, 0x0F39 },
 216         { 0x0F3E, 0x0F47 },
 217         { 0x0F49, 0x0F69 },
 218         { 0x0F71, 0x0F84 },
 219         { 0x0F86, 0x0F8B },
 220         { 0x0F90, 0x0F95 },
 221         { 0x0F97, 0x0F97 },
 222         { 0x0F99, 0x0FAD },
 223         { 0x0FB1, 0x0FB7 },
 224         { 0x0FB9, 0x0FB9 },
 225         { 0x10A0, 0x10C5 },
 226         { 0x10D0, 0x10F6 },
 227         { 0x1E00, 0x1E9B },
 228         { 0x1EA0, 0x1EF9 },
 229         { 0x1F00, 0x1F15 },
 230         { 0x1F18, 0x1F1D },
 231         { 0x1F20, 0x1F45 },
 232         { 0x1F48, 0x1F4D },
 233         { 0x1F50, 0x1F57 },
 234         { 0x1F59, 0x1F59 },
 235         { 0x1F5B, 0x1F5B },
 236         { 0x1F5D, 0x1F5D },
 237         { 0x1F5F, 0x1F7D },
 238         { 0x1F80, 0x1FB4 },
 239         { 0x1FB6, 0x1FBC },
 240         { 0x1FBE, 0x1FBE },
 241         { 0x1FC2, 0x1FC4 },
 242         { 0x1FC6, 0x1FCC },
 243         { 0x1FD0, 0x1FD3 },
 244         { 0x1FD6, 0x1FDB },
 245         { 0x1FE0, 0x1FEC },
 246         { 0x1FF2, 0x1FF4 },
 247         { 0x1FF6, 0x1FFC },
 248         { 0x203F, 0x2040 },
 249         { 0x207F, 0x207F },
 250         { 0x2102, 0x2102 },
 251         { 0x2107, 0x2107 },
 252         { 0x210A, 0x2113 },
 253         { 0x2115, 0x2115 },
 254         { 0x2118, 0x211D },
 255         { 0x2124, 0x2124 },
 256         { 0x2126, 0x2126 },
 257         { 0x2128, 0x2128 },
 258         { 0x212A, 0x2131 },
 259         { 0x2133, 0x2138 },
 260         { 0x2160, 0x2182 },
 261         { 0x3005, 0x3007 },
 262         { 0x3021, 0x3029 },
 263         { 0x3041, 0x3093 },
 264         { 0x309B, 0x309C },
 265         { 0x30A1, 0x30F6 },
 266         { 0x30FB, 0x30FC },
 267         { 0x3105, 0x312C },
 268         { 0x4E00, 0x9FA5 },
 269         { 0xAC00, 0xD7A3 },
 270     };
 271
 272 #ifdef DEBUG
 273     for (int i = 0; i < sizeof(table) / sizeof(table[0]); i++)
 274     {
 275         //printf("%x\n", table[i][0]);
 276         assert(table[i][0] <= table[i][1]);
 277         if (i < sizeof(table) / sizeof(table[0]) - 1)
 278             assert(table[i][1] < table[i + 1][0]);
 279     }
 280 #endif
 281
 282     if (u > 0xD7A3)
 283         goto Lisnot;
 284
 285     // Binary search
 286     int mid;
 287     int low;
 288     int high;
 289
 290     low = 0;
 291     high = sizeof(table) / sizeof(table[0]) - 1;
 292     while (low <= high)
 293     {
 294         mid = (low + high) >> 1;
 295         if (u < table[mid][0])
 296             high = mid - 1;
 297         else if (u > table[mid][1])
 298             low = mid + 1;
 299         else
 300             goto Lis;
 301     }
 302
 303 Lisnot:
 304 #ifdef DEBUG
 305     for (int i = 0; i < sizeof(table) / sizeof(table[0]); i++)
 306     {
 307         assert(u < table[i][0] || u > table[i][1]);
 308     }
 309 #endif
 310     return 0;
 311
 312 Lis:
 313 #ifdef DEBUG
 314     for (int i = 0; i < sizeof(table) / sizeof(table[0]); i++)
 315     {
 316         if (u >= table[i][0] && u <= table[i][1])
 317             return 1;
 318     }
 319     assert(0);          // should have been in table
 320 #endif
 321     return 1;
 322 }
 323