src/expat/gennmtab/gennmtab.c

   1 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
   2    See the file COPYING for copying permission.
   3 */
   4
   5 #include <string.h>
   6 #include <stdio.h>
   7 #include <stddef.h>
   8
   9 struct range {
  10   int start;
  11   int end;
  12 };
  13
  14 struct range nmstrt[] = {
  15   { '_' },
  16   { ':' },
  17   /* BaseChar */
  18   { 0x0041, 0x005a },
  19   { 0x0061, 0x007a },
  20   { 0x00c0, 0x00d6 },
  21   { 0x00d8, 0x00f6 },
  22   { 0x00f8, 0x00ff },
  23   { 0x0100, 0x0131 },
  24   { 0x0134, 0x013e },
  25   { 0x0141, 0x0148 },
  26   { 0x014a, 0x017e },
  27   { 0x0180, 0x01c3 },
  28   { 0x01cd, 0x01f0 },
  29   { 0x01f4, 0x01f5 },
  30   { 0x01fa, 0x0217 },
  31   { 0x0250, 0x02a8 },
  32   { 0x02bb, 0x02c1 },
  33   { 0x0386 },
  34   { 0x0388, 0x038a },
  35   { 0x038c },
  36   { 0x038e, 0x03a1 },
  37   { 0x03a3, 0x03ce },
  38   { 0x03d0, 0x03d6 },
  39   { 0x03da },
  40   { 0x03dc },
  41   { 0x03de },
  42   { 0x03e0 },
  43   { 0x03e2, 0x03f3 },
  44   { 0x0401, 0x040c },
  45   { 0x040e, 0x044f },
  46   { 0x0451, 0x045c },
  47   { 0x045e, 0x0481 },
  48   { 0x0490, 0x04c4 },
  49   { 0x04c7, 0x04c8 },
  50   { 0x04cb, 0x04cc },
  51   { 0x04d0, 0x04eb },
  52   { 0x04ee, 0x04f5 },
  53   { 0x04f8, 0x04f9 },
  54   { 0x0531, 0x0556 },
  55   { 0x0559 },
  56   { 0x0561, 0x0586 },
  57   { 0x05d0, 0x05ea },
  58   { 0x05f0, 0x05f2 },
  59   { 0x0621, 0x063a },
  60   { 0x0641, 0x064a },
  61   { 0x0671, 0x06b7 },
  62   { 0x06ba, 0x06be },
  63   { 0x06c0, 0x06ce },
  64   { 0x06d0, 0x06d3 },
  65   { 0x06d5 },
  66   { 0x06e5, 0x06e6 },
  67   { 0x0905, 0x0939 },
  68   { 0x093d },
  69   { 0x0958, 0x0961 },
  70   { 0x0985, 0x098c },
  71   { 0x098f, 0x0990 },
  72   { 0x0993, 0x09a8 },
  73   { 0x09aa, 0x09b0 },
  74   { 0x09b2 },
  75   { 0x09b6, 0x09b9 },
  76   { 0x09dc, 0x09dd },
  77   { 0x09df, 0x09e1 },
  78   { 0x09f0, 0x09f1 },
  79   { 0x0a05, 0x0a0a },
  80   { 0x0a0f, 0x0a10 },
  81   { 0x0a13, 0x0a28 },
  82   { 0x0a2a, 0x0a30 },
  83   { 0x0a32, 0x0a33 },
  84   { 0x0a35, 0x0a36 },
  85   { 0x0a38, 0x0a39 },
  86   { 0x0a59, 0x0a5c },
  87   { 0x0a5e },
  88   { 0x0a72, 0x0a74 },
  89   { 0x0a85, 0x0a8b },
  90   { 0x0a8d },
  91   { 0x0a8f, 0x0a91 },
  92   { 0x0a93, 0x0aa8 },
  93   { 0x0aaa, 0x0ab0 },
  94   { 0x0ab2, 0x0ab3 },
  95   { 0x0ab5, 0x0ab9 },
  96   { 0x0abd },
  97   { 0x0ae0 },
  98   { 0x0b05, 0x0b0c },
  99   { 0x0b0f, 0x0b10 },
 100   { 0x0b13, 0x0b28 },
 101   { 0x0b2a, 0x0b30 },
 102   { 0x0b32, 0x0b33 },
 103   { 0x0b36, 0x0b39 },
 104   { 0x0b3d },
 105   { 0x0b5c, 0x0b5d },
 106   { 0x0b5f, 0x0b61 },
 107   { 0x0b85, 0x0b8a },
 108   { 0x0b8e, 0x0b90 },
 109   { 0x0b92, 0x0b95 },
 110   { 0x0b99, 0x0b9a },
 111   { 0x0b9c },
 112   { 0x0b9e, 0x0b9f },
 113   { 0x0ba3, 0x0ba4 },
 114   { 0x0ba8, 0x0baa },
 115   { 0x0bae, 0x0bb5 },
 116   { 0x0bb7, 0x0bb9 },
 117   { 0x0c05, 0x0c0c },
 118   { 0x0c0e, 0x0c10 },
 119   { 0x0c12, 0x0c28 },
 120   { 0x0c2a, 0x0c33 },
 121   { 0x0c35, 0x0c39 },
 122   { 0x0c60, 0x0c61 },
 123   { 0x0c85, 0x0c8c },
 124   { 0x0c8e, 0x0c90 },
 125   { 0x0c92, 0x0ca8 },
 126   { 0x0caa, 0x0cb3 },
 127   { 0x0cb5, 0x0cb9 },
 128   { 0x0cde },
 129   { 0x0ce0, 0x0ce1 },
 130   { 0x0d05, 0x0d0c },
 131   { 0x0d0e, 0x0d10 },
 132   { 0x0d12, 0x0d28 },
 133   { 0x0d2a, 0x0d39 },
 134   { 0x0d60, 0x0d61 },
 135   { 0x0e01, 0x0e2e },
 136   { 0x0e30 },
 137   { 0x0e32, 0x0e33 },
 138   { 0x0e40, 0x0e45 },
 139   { 0x0e81, 0x0e82 },
 140   { 0x0e84 },
 141   { 0x0e87, 0x0e88 },
 142   { 0x0e8a },
 143   { 0x0e8d },
 144   { 0x0e94, 0x0e97 },
 145   { 0x0e99, 0x0e9f },
 146   { 0x0ea1, 0x0ea3 },
 147   { 0x0ea5 },
 148   { 0x0ea7 },
 149   { 0x0eaa, 0x0eab },
 150   { 0x0ead, 0x0eae },
 151   { 0x0eb0 },
 152   { 0x0eb2, 0x0eb3 },
 153   { 0x0ebd },
 154   { 0x0ec0, 0x0ec4 },
 155   { 0x0f40, 0x0f47 },
 156   { 0x0f49, 0x0f69 },
 157   { 0x10a0, 0x10c5 },
 158   { 0x10d0, 0x10f6 },
 159   { 0x1100 },
 160   { 0x1102, 0x1103 },
 161   { 0x1105, 0x1107 },
 162   { 0x1109 },
 163   { 0x110b, 0x110c },
 164   { 0x110e, 0x1112 },
 165   { 0x113c },
 166   { 0x113e },
 167   { 0x1140 },
 168   { 0x114c },
 169   { 0x114e },
 170   { 0x1150 },
 171   { 0x1154, 0x1155 },
 172   { 0x1159 },
 173   { 0x115f, 0x1161 },
 174   { 0x1163 },
 175   { 0x1165 },
 176   { 0x1167 },
 177   { 0x1169 },
 178   { 0x116d, 0x116e },
 179   { 0x1172, 0x1173 },
 180   { 0x1175 },
 181   { 0x119e },
 182   { 0x11a8 },
 183   { 0x11ab },
 184   { 0x11ae, 0x11af },
 185   { 0x11b7, 0x11b8 },
 186   { 0x11ba },
 187   { 0x11bc, 0x11c2 },
 188   { 0x11eb },
 189   { 0x11f0 },
 190   { 0x11f9 },
 191   { 0x1e00, 0x1e9b },
 192   { 0x1ea0, 0x1ef9 },
 193   { 0x1f00, 0x1f15 },
 194   { 0x1f18, 0x1f1d },
 195   { 0x1f20, 0x1f45 },
 196   { 0x1f48, 0x1f4d },
 197   { 0x1f50, 0x1f57 },
 198   { 0x1f59 },
 199   { 0x1f5b },
 200   { 0x1f5d },
 201   { 0x1f5f, 0x1f7d },
 202   { 0x1f80, 0x1fb4 },
 203   { 0x1fb6, 0x1fbc },
 204   { 0x1fbe },
 205   { 0x1fc2, 0x1fc4 },
 206   { 0x1fc6, 0x1fcc },
 207   { 0x1fd0, 0x1fd3 },
 208   { 0x1fd6, 0x1fdb },
 209   { 0x1fe0, 0x1fec },
 210   { 0x1ff2, 0x1ff4 },
 211   { 0x1ff6, 0x1ffc },
 212   { 0x2126 },
 213   { 0x212a, 0x212b },
 214   { 0x212e },
 215   { 0x2180, 0x2182 },
 216   { 0x3041, 0x3094 },
 217   { 0x30a1, 0x30fa },
 218   { 0x3105, 0x312c },
 219   { 0xac00, 0xd7a3 },
 220   /* Ideographic */
 221   { 0x4e00, 0x9fa5 },
 222   { 0x3007 },
 223   { 0x3021, 0x3029 },
 224 };
 225
 226 /* name chars that are not name start chars */
 227 struct range name[] = {
 228   { '.' },
 229   { '-' },
 230   /* CombiningChar */
 231   { 0x0300, 0x0345 },
 232   { 0x0360, 0x0361 },
 233   { 0x0483, 0x0486 },
 234   { 0x0591, 0x05a1 },
 235   { 0x05a3, 0x05b9 },
 236   { 0x05bb, 0x05bd },
 237   { 0x05bf },
 238   { 0x05c1, 0x05c2 },
 239   { 0x05c4 },
 240   { 0x064b, 0x0652 },
 241   { 0x0670 },
 242   { 0x06d6, 0x06dc },
 243   { 0x06dd, 0x06df },
 244   { 0x06e0, 0x06e4 },
 245   { 0x06e7, 0x06e8 },
 246   { 0x06ea, 0x06ed },
 247   { 0x0901, 0x0903 },
 248   { 0x093c },
 249   { 0x093e, 0x094c },
 250   { 0x094d },
 251   { 0x0951, 0x0954 },
 252   { 0x0962, 0x0963 },
 253   { 0x0981, 0x0983 },
 254   { 0x09bc },
 255   { 0x09be },
 256   { 0x09bf },
 257   { 0x09c0, 0x09c4 },
 258   { 0x09c7, 0x09c8 },
 259   { 0x09cb, 0x09cd },
 260   { 0x09d7 },
 261   { 0x09e2, 0x09e3 },
 262   { 0x0a02 },
 263   { 0x0a3c },
 264   { 0x0a3e },
 265   { 0x0a3f },
 266   { 0x0a40, 0x0a42 },
 267   { 0x0a47, 0x0a48 },
 268   { 0x0a4b, 0x0a4d },
 269   { 0x0a70, 0x0a71 },
 270   { 0x0a81, 0x0a83 },
 271   { 0x0abc },
 272   { 0x0abe, 0x0ac5 },
 273   { 0x0ac7, 0x0ac9 },
 274   { 0x0acb, 0x0acd },
 275   { 0x0b01, 0x0b03 },
 276   { 0x0b3c },
 277   { 0x0b3e, 0x0b43 },
 278   { 0x0b47, 0x0b48 },
 279   { 0x0b4b, 0x0b4d },
 280   { 0x0b56, 0x0b57 },
 281   { 0x0b82, 0x0b83 },
 282   { 0x0bbe, 0x0bc2 },
 283   { 0x0bc6, 0x0bc8 },
 284   { 0x0bca, 0x0bcd },
 285   { 0x0bd7 },
 286   { 0x0c01, 0x0c03 },
 287   { 0x0c3e, 0x0c44 },
 288   { 0x0c46, 0x0c48 },
 289   { 0x0c4a, 0x0c4d },
 290   { 0x0c55, 0x0c56 },
 291   { 0x0c82, 0x0c83 },
 292   { 0x0cbe, 0x0cc4 },
 293   { 0x0cc6, 0x0cc8 },
 294   { 0x0cca, 0x0ccd },
 295   { 0x0cd5, 0x0cd6 },
 296   { 0x0d02, 0x0d03 },
 297   { 0x0d3e, 0x0d43 },
 298   { 0x0d46, 0x0d48 },
 299   { 0x0d4a, 0x0d4d },
 300   { 0x0d57 },
 301   { 0x0e31 },
 302   { 0x0e34, 0x0e3a },
 303   { 0x0e47, 0x0e4e },
 304   { 0x0eb1 },
 305   { 0x0eb4, 0x0eb9 },
 306   { 0x0ebb, 0x0ebc },
 307   { 0x0ec8, 0x0ecd },
 308   { 0x0f18, 0x0f19 },
 309   { 0x0f35 },
 310   { 0x0f37 },
 311   { 0x0f39 },
 312   { 0x0f3e },
 313   { 0x0f3f },
 314   { 0x0f71, 0x0f84 },
 315   { 0x0f86, 0x0f8b },
 316   { 0x0f90, 0x0f95 },
 317   { 0x0f97 },
 318   { 0x0f99, 0x0fad },
 319   { 0x0fb1, 0x0fb7 },
 320   { 0x0fb9 },
 321   { 0x20d0, 0x20dc },
 322   { 0x20e1 },
 323   { 0x302a, 0x302f },
 324   { 0x3099 },
 325   { 0x309a },
 326   /* Digit */
 327   { 0x0030, 0x0039 },
 328   { 0x0660, 0x0669 },
 329   { 0x06f0, 0x06f9 },
 330   { 0x0966, 0x096f },
 331   { 0x09e6, 0x09ef },
 332   { 0x0a66, 0x0a6f },
 333   { 0x0ae6, 0x0aef },
 334   { 0x0b66, 0x0b6f },
 335   { 0x0be7, 0x0bef },
 336   { 0x0c66, 0x0c6f },
 337   { 0x0ce6, 0x0cef },
 338   { 0x0d66, 0x0d6f },
 339   { 0x0e50, 0x0e59 },
 340   { 0x0ed0, 0x0ed9 },
 341   { 0x0f20, 0x0f29 },
 342   /* Extender */
 343   { 0xb7 },
 344   { 0x02d0 },
 345   { 0x02d1 },
 346   { 0x0387 },
 347   { 0x0640 },
 348   { 0x0e46 },
 349   { 0x0ec6 },
 350   { 0x3005 },
 351   { 0x3031, 0x3035 },
 352   { 0x309d, 0x309e },
 353   { 0x30fc, 0x30fe },
 354 };
 355
 356 static void
 357 setTab(char *tab, struct range *ranges, size_t nRanges)
 358 {
 359   size_t i;
 360   int j;
 361   for (i = 0; i < nRanges; i++) {
 362     if (ranges[i].end) {
 363       for (j = ranges[i].start; j <= ranges[i].end; j++)
 364         tab[j] = 1;
 365     }
 366     else
 367       tab[ranges[i].start] = 1;
 368   }
 369 }
 370
 371 static void
 372 printTabs(char *tab)
 373 {
 374   int nBitmaps = 2;
 375   int i, j, k;
 376   unsigned char pageIndex[512];
 377
 378   printf(
 379 "static const unsigned namingBitmap[] = {\n\
 380 0x00000000, 0x00000000, 0x00000000, 0x00000000,\n\
 381 0x00000000, 0x00000000, 0x00000000, 0x00000000,\n\
 382 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,\n\
 383 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,\n");
 384   for (i = 0; i < 512; i++) {
 385     int kind = tab[i*256];
 386     for (j = 1; j < 256; j++)
 387       if (tab[i*256 +j] != kind) {
 388         kind = -1;
 389         break;
 390       }
 391     if (i >= 256 &&  memcmp(tab + (i - 256)*256, tab + i*256, 256) == 0)
 392       pageIndex[i] = pageIndex[i - 256];
 393     else if (kind == -1) {
 394       pageIndex[i] = nBitmaps++;
 395       for (j = 0; j < 8; j++) {
 396         unsigned val = 0;
 397         for (k = 0; k < 32; k++) {
 398           if (tab[i*256 + j*32 +k])
 399             val |= (1 << k);
 400         }
 401         printf("0x%08X,", val);
 402         putchar((((j + 1) & 3) == 0) ? '\n' : ' ');
 403       }
 404     }
 405     else
 406       pageIndex[i] = kind;
 407   }
 408   printf("};\n");
 409   printf("static const unsigned char nmstrtPages[] = {\n");
 410   for (i = 0; i < 512; i++) {
 411     if (i == 256)
 412       printf("};\nstatic const unsigned char namePages[] = {\n");
 413     printf("0x%02X,", pageIndex[i]);
 414     putchar((((i + 1) & 7) == 0) ? '\n' : ' ');
 415   }
 416   printf("};\n");
 417 }
 418
 419 int
 420 main()
 421 {
 422   char tab[2*65536];
 423   memset(tab, 0, 65536);
 424   setTab(tab, nmstrt, sizeof(nmstrt)/sizeof(nmstrt[0]));
 425   memcpy(tab + 65536, tab, 65536);
 426   setTab(tab + 65536, name, sizeof(name)/sizeof(name[0]));
 427   printTabs(tab);
 428   return 0;
 429 }