workbench/libs/expat/xmlrole.c

   1 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
   2    See the file COPYING for copying permission.
   3 */
   4
   5 #include "amigaconfig.h"
   6
   7 #include <stddef.h>
   8
   9 #ifdef COMPILED_FROM_DSP
  10 #include "winconfig.h"
  11 #elif defined(MACOS_CLASSIC)
  12 #include "macconfig.h"
  13 #elif defined(__amigaos4__)
  14 #include "amigaconfig.h"
  15 #elif defined(__WATCOMC__)
  16 #include "watcomconfig.h"
  17 #else
  18 #ifdef HAVE_EXPAT_CONFIG_H
  19 #include <expat_config.h>
  20 #endif
  21 #endif /* ndef COMPILED_FROM_DSP */
  22
  23 #include "expat_external.h"
  24 #include "internal.h"
  25 #include "xmlrole.h"
  26 #include "ascii.h"
  27
  28 /* Doesn't check:
  29
  30  that ,| are not mixed in a model group
  31  content of literals
  32
  33 */
  34
  35 static const char KW_ANY[] = {
  36     ASCII_A, ASCII_N, ASCII_Y, '\0' };
  37 static const char KW_ATTLIST[] = {
  38     ASCII_A, ASCII_T, ASCII_T, ASCII_L, ASCII_I, ASCII_S, ASCII_T, '\0' };
  39 static const char KW_CDATA[] = {
  40     ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0' };
  41 static const char KW_DOCTYPE[] = {
  42     ASCII_D, ASCII_O, ASCII_C, ASCII_T, ASCII_Y, ASCII_P, ASCII_E, '\0' };
  43 static const char KW_ELEMENT[] = {
  44     ASCII_E, ASCII_L, ASCII_E, ASCII_M, ASCII_E, ASCII_N, ASCII_T, '\0' };
  45 static const char KW_EMPTY[] = {
  46     ASCII_E, ASCII_M, ASCII_P, ASCII_T, ASCII_Y, '\0' };
  47 static const char KW_ENTITIES[] = {
  48     ASCII_E, ASCII_N, ASCII_T, ASCII_I, ASCII_T, ASCII_I, ASCII_E, ASCII_S,
  49     '\0' };
  50 static const char KW_ENTITY[] = {
  51     ASCII_E, ASCII_N, ASCII_T, ASCII_I, ASCII_T, ASCII_Y, '\0' };
  52 static const char KW_FIXED[] = {
  53     ASCII_F, ASCII_I, ASCII_X, ASCII_E, ASCII_D, '\0' };
  54 static const char KW_ID[] = {
  55     ASCII_I, ASCII_D, '\0' };
  56 static const char KW_IDREF[] = {
  57     ASCII_I, ASCII_D, ASCII_R, ASCII_E, ASCII_F, '\0' };
  58 static const char KW_IDREFS[] = {
  59     ASCII_I, ASCII_D, ASCII_R, ASCII_E, ASCII_F, ASCII_S, '\0' };
  60 #ifdef XML_DTD
  61 static const char KW_IGNORE[] = {
  62     ASCII_I, ASCII_G, ASCII_N, ASCII_O, ASCII_R, ASCII_E, '\0' };
  63 #endif
  64 static const char KW_IMPLIED[] = {
  65     ASCII_I, ASCII_M, ASCII_P, ASCII_L, ASCII_I, ASCII_E, ASCII_D, '\0' };
  66 #ifdef XML_DTD
  67 static const char KW_INCLUDE[] = {
  68     ASCII_I, ASCII_N, ASCII_C, ASCII_L, ASCII_U, ASCII_D, ASCII_E, '\0' };
  69 #endif
  70 static const char KW_NDATA[] = {
  71     ASCII_N, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0' };
  72 static const char KW_NMTOKEN[] = {
  73     ASCII_N, ASCII_M, ASCII_T, ASCII_O, ASCII_K, ASCII_E, ASCII_N, '\0' };
  74 static const char KW_NMTOKENS[] = {
  75     ASCII_N, ASCII_M, ASCII_T, ASCII_O, ASCII_K, ASCII_E, ASCII_N, ASCII_S,
  76     '\0' };
  77 static const char KW_NOTATION[] =
  78     { ASCII_N, ASCII_O, ASCII_T, ASCII_A, ASCII_T, ASCII_I, ASCII_O, ASCII_N,
  79       '\0' };
  80 static const char KW_PCDATA[] = {
  81     ASCII_P, ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0' };
  82 static const char KW_PUBLIC[] = {
  83     ASCII_P, ASCII_U, ASCII_B, ASCII_L, ASCII_I, ASCII_C, '\0' };
  84 static const char KW_REQUIRED[] = {
  85     ASCII_R, ASCII_E, ASCII_Q, ASCII_U, ASCII_I, ASCII_R, ASCII_E, ASCII_D,
  86     '\0' };
  87 static const char KW_SYSTEM[] = {
  88     ASCII_S, ASCII_Y, ASCII_S, ASCII_T, ASCII_E, ASCII_M, '\0' };
  89
  90 #ifndef MIN_BYTES_PER_CHAR
  91 #define MIN_BYTES_PER_CHAR(enc) ((enc)->minBytesPerChar)
  92 #endif
  93
  94 #ifdef XML_DTD
  95 #define setTopLevel(state) \
  96   ((state)->handler = ((state)->documentEntity \
  97                        ? internalSubset \
  98                        : externalSubset1))
  99 #else /* not XML_DTD */
 100 #define setTopLevel(state) ((state)->handler = internalSubset)
 101 #endif /* not XML_DTD */
 102
 103 typedef int PTRCALL PROLOG_HANDLER(PROLOG_STATE *state,
 104                                    int tok,
 105                                    const char *ptr,
 106                                    const char *end,
 107                                    const ENCODING *enc);
 108
 109 static PROLOG_HANDLER
 110   prolog0, prolog1, prolog2,
 111   doctype0, doctype1, doctype2, doctype3, doctype4, doctype5,
 112   internalSubset,
 113   entity0, entity1, entity2, entity3, entity4, entity5, entity6,
 114   entity7, entity8, entity9, entity10,
 115   notation0, notation1, notation2, notation3, notation4,
 116   attlist0, attlist1, attlist2, attlist3, attlist4, attlist5, attlist6,
 117   attlist7, attlist8, attlist9,
 118   element0, element1, element2, element3, element4, element5, element6,
 119   element7,
 120 #ifdef XML_DTD
 121   externalSubset0, externalSubset1,
 122   condSect0, condSect1, condSect2,
 123 #endif /* XML_DTD */
 124   declClose,
 125   error;
 126
 127 static int FASTCALL common(PROLOG_STATE *state, int tok);
 128
 129 static int PTRCALL
 130 prolog0(PROLOG_STATE *state,
 131         int tok,
 132         const char *ptr,
 133         const char *end,
 134         const ENCODING *enc)
 135 {
 136   switch (tok) {
 137   case XML_TOK_PROLOG_S:
 138     state->handler = prolog1;
 139     return XML_ROLE_NONE;
 140   case XML_TOK_XML_DECL:
 141     state->handler = prolog1;
 142     return XML_ROLE_XML_DECL;
 143   case XML_TOK_PI:
 144     state->handler = prolog1;
 145     return XML_ROLE_PI;
 146   case XML_TOK_COMMENT:
 147     state->handler = prolog1;
 148     return XML_ROLE_COMMENT;
 149   case XML_TOK_BOM:
 150     return XML_ROLE_NONE;
 151   case XML_TOK_DECL_OPEN:
 152     if (!XmlNameMatchesAscii(enc,
 153                              ptr + 2 * MIN_BYTES_PER_CHAR(enc),
 154                              end,
 155                              KW_DOCTYPE))
 156       break;
 157     state->handler = doctype0;
 158     return XML_ROLE_DOCTYPE_NONE;
 159   case XML_TOK_INSTANCE_START:
 160     state->handler = error;
 161     return XML_ROLE_INSTANCE_START;
 162   }
 163   return common(state, tok);
 164 }
 165
 166 static int PTRCALL
 167 prolog1(PROLOG_STATE *state,
 168         int tok,
 169         const char *ptr,
 170         const char *end,
 171         const ENCODING *enc)
 172 {
 173   switch (tok) {
 174   case XML_TOK_PROLOG_S:
 175     return XML_ROLE_NONE;
 176   case XML_TOK_PI:
 177     return XML_ROLE_PI;
 178   case XML_TOK_COMMENT:
 179     return XML_ROLE_COMMENT;
 180   case XML_TOK_BOM:
 181     return XML_ROLE_NONE;
 182   case XML_TOK_DECL_OPEN:
 183     if (!XmlNameMatchesAscii(enc,
 184                              ptr + 2 * MIN_BYTES_PER_CHAR(enc),
 185                              end,
 186                              KW_DOCTYPE))
 187       break;
 188     state->handler = doctype0;
 189     return XML_ROLE_DOCTYPE_NONE;
 190   case XML_TOK_INSTANCE_START:
 191     state->handler = error;
 192     return XML_ROLE_INSTANCE_START;
 193   }
 194   return common(state, tok);
 195 }
 196
 197 static int PTRCALL
 198 prolog2(PROLOG_STATE *state,
 199         int tok,
 200         const char *ptr,
 201         const char *end,
 202         const ENCODING *enc)
 203 {
 204   switch (tok) {
 205   case XML_TOK_PROLOG_S:
 206     return XML_ROLE_NONE;
 207   case XML_TOK_PI:
 208     return XML_ROLE_PI;
 209   case XML_TOK_COMMENT:
 210     return XML_ROLE_COMMENT;
 211   case XML_TOK_INSTANCE_START:
 212     state->handler = error;
 213     return XML_ROLE_INSTANCE_START;
 214   }
 215   return common(state, tok);
 216 }
 217
 218 static int PTRCALL
 219 doctype0(PROLOG_STATE *state,
 220          int tok,
 221          const char *ptr,
 222          const char *end,
 223          const ENCODING *enc)
 224 {
 225   switch (tok) {
 226   case XML_TOK_PROLOG_S:
 227     return XML_ROLE_DOCTYPE_NONE;
 228   case XML_TOK_NAME:
 229   case XML_TOK_PREFIXED_NAME:
 230     state->handler = doctype1;
 231     return XML_ROLE_DOCTYPE_NAME;
 232   }
 233   return common(state, tok);
 234 }
 235
 236 static int PTRCALL
 237 doctype1(PROLOG_STATE *state,
 238          int tok,
 239          const char *ptr,
 240          const char *end,
 241          const ENCODING *enc)
 242 {
 243   switch (tok) {
 244   case XML_TOK_PROLOG_S:
 245     return XML_ROLE_DOCTYPE_NONE;
 246   case XML_TOK_OPEN_BRACKET:
 247     state->handler = internalSubset;
 248     return XML_ROLE_DOCTYPE_INTERNAL_SUBSET;
 249   case XML_TOK_DECL_CLOSE:
 250     state->handler = prolog2;
 251     return XML_ROLE_DOCTYPE_CLOSE;
 252   case XML_TOK_NAME:
 253     if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
 254       state->handler = doctype3;
 255       return XML_ROLE_DOCTYPE_NONE;
 256     }
 257     if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
 258       state->handler = doctype2;
 259       return XML_ROLE_DOCTYPE_NONE;
 260     }
 261     break;
 262   }
 263   return common(state, tok);
 264 }
 265
 266 static int PTRCALL
 267 doctype2(PROLOG_STATE *state,
 268          int tok,
 269          const char *ptr,
 270          const char *end,
 271          const ENCODING *enc)
 272 {
 273   switch (tok) {
 274   case XML_TOK_PROLOG_S:
 275     return XML_ROLE_DOCTYPE_NONE;
 276   case XML_TOK_LITERAL:
 277     state->handler = doctype3;
 278     return XML_ROLE_DOCTYPE_PUBLIC_ID;
 279   }
 280   return common(state, tok);
 281 }
 282
 283 static int PTRCALL
 284 doctype3(PROLOG_STATE *state,
 285          int tok,
 286          const char *ptr,
 287          const char *end,
 288          const ENCODING *enc)
 289 {
 290   switch (tok) {
 291   case XML_TOK_PROLOG_S:
 292     return XML_ROLE_DOCTYPE_NONE;
 293   case XML_TOK_LITERAL:
 294     state->handler = doctype4;
 295     return XML_ROLE_DOCTYPE_SYSTEM_ID;
 296   }
 297   return common(state, tok);
 298 }
 299
 300 static int PTRCALL
 301 doctype4(PROLOG_STATE *state,
 302          int tok,
 303          const char *ptr,
 304          const char *end,
 305          const ENCODING *enc)
 306 {
 307   switch (tok) {
 308   case XML_TOK_PROLOG_S:
 309     return XML_ROLE_DOCTYPE_NONE;
 310   case XML_TOK_OPEN_BRACKET:
 311     state->handler = internalSubset;
 312     return XML_ROLE_DOCTYPE_INTERNAL_SUBSET;
 313   case XML_TOK_DECL_CLOSE:
 314     state->handler = prolog2;
 315     return XML_ROLE_DOCTYPE_CLOSE;
 316   }
 317   return common(state, tok);
 318 }
 319
 320 static int PTRCALL
 321 doctype5(PROLOG_STATE *state,
 322          int tok,
 323          const char *ptr,
 324          const char *end,
 325          const ENCODING *enc)
 326 {
 327   switch (tok) {
 328   case XML_TOK_PROLOG_S:
 329     return XML_ROLE_DOCTYPE_NONE;
 330   case XML_TOK_DECL_CLOSE:
 331     state->handler = prolog2;
 332     return XML_ROLE_DOCTYPE_CLOSE;
 333   }
 334   return common(state, tok);
 335 }
 336
 337 static int PTRCALL
 338 internalSubset(PROLOG_STATE *state,
 339                int tok,
 340                const char *ptr,
 341                const char *end,
 342                const ENCODING *enc)
 343 {
 344   switch (tok) {
 345   case XML_TOK_PROLOG_S:
 346     return XML_ROLE_NONE;
 347   case XML_TOK_DECL_OPEN:
 348     if (XmlNameMatchesAscii(enc,
 349                             ptr + 2 * MIN_BYTES_PER_CHAR(enc),
 350                             end,
 351                             KW_ENTITY)) {
 352       state->handler = entity0;
 353       return XML_ROLE_ENTITY_NONE;
 354     }
 355     if (XmlNameMatchesAscii(enc,
 356                             ptr + 2 * MIN_BYTES_PER_CHAR(enc),
 357                             end,
 358                             KW_ATTLIST)) {
 359       state->handler = attlist0;
 360       return XML_ROLE_ATTLIST_NONE;
 361     }
 362     if (XmlNameMatchesAscii(enc,
 363                             ptr + 2 * MIN_BYTES_PER_CHAR(enc),
 364                             end,
 365                             KW_ELEMENT)) {
 366       state->handler = element0;
 367       return XML_ROLE_ELEMENT_NONE;
 368     }
 369     if (XmlNameMatchesAscii(enc,
 370                             ptr + 2 * MIN_BYTES_PER_CHAR(enc),
 371                             end,
 372                             KW_NOTATION)) {
 373       state->handler = notation0;
 374       return XML_ROLE_NOTATION_NONE;
 375     }
 376     break;
 377   case XML_TOK_PI:
 378     return XML_ROLE_PI;
 379   case XML_TOK_COMMENT:
 380     return XML_ROLE_COMMENT;
 381   case XML_TOK_PARAM_ENTITY_REF:
 382     return XML_ROLE_PARAM_ENTITY_REF;
 383   case XML_TOK_CLOSE_BRACKET:
 384     state->handler = doctype5;
 385     return XML_ROLE_DOCTYPE_NONE;
 386   case XML_TOK_NONE:
 387     return XML_ROLE_NONE;
 388   }
 389   return common(state, tok);
 390 }
 391
 392 #ifdef XML_DTD
 393
 394 static int PTRCALL
 395 externalSubset0(PROLOG_STATE *state,
 396                 int tok,
 397                 const char *ptr,
 398                 const char *end,
 399                 const ENCODING *enc)
 400 {
 401   state->handler = externalSubset1;
 402   if (tok == XML_TOK_XML_DECL)
 403     return XML_ROLE_TEXT_DECL;
 404   return externalSubset1(state, tok, ptr, end, enc);
 405 }
 406
 407 static int PTRCALL
 408 externalSubset1(PROLOG_STATE *state,
 409                 int tok,
 410                 const char *ptr,
 411                 const char *end,
 412                 const ENCODING *enc)
 413 {
 414   switch (tok) {
 415   case XML_TOK_COND_SECT_OPEN:
 416     state->handler = condSect0;
 417     return XML_ROLE_NONE;
 418   case XML_TOK_COND_SECT_CLOSE:
 419     if (state->includeLevel == 0)
 420       break;
 421     state->includeLevel -= 1;
 422     return XML_ROLE_NONE;
 423   case XML_TOK_PROLOG_S:
 424     return XML_ROLE_NONE;
 425   case XML_TOK_CLOSE_BRACKET:
 426     break;
 427   case XML_TOK_NONE:
 428     if (state->includeLevel)
 429       break;
 430     return XML_ROLE_NONE;
 431   default:
 432     return internalSubset(state, tok, ptr, end, enc);
 433   }
 434   return common(state, tok);
 435 }
 436
 437 #endif /* XML_DTD */
 438
 439 static int PTRCALL
 440 entity0(PROLOG_STATE *state,
 441         int tok,
 442         const char *ptr,
 443         const char *end,
 444         const ENCODING *enc)
 445 {
 446   switch (tok) {
 447   case XML_TOK_PROLOG_S:
 448     return XML_ROLE_ENTITY_NONE;
 449   case XML_TOK_PERCENT:
 450     state->handler = entity1;
 451     return XML_ROLE_ENTITY_NONE;
 452   case XML_TOK_NAME:
 453     state->handler = entity2;
 454     return XML_ROLE_GENERAL_ENTITY_NAME;
 455   }
 456   return common(state, tok);
 457 }
 458
 459 static int PTRCALL
 460 entity1(PROLOG_STATE *state,
 461         int tok,
 462         const char *ptr,
 463         const char *end,
 464         const ENCODING *enc)
 465 {
 466   switch (tok) {
 467   case XML_TOK_PROLOG_S:
 468     return XML_ROLE_ENTITY_NONE;
 469   case XML_TOK_NAME:
 470     state->handler = entity7;
 471     return XML_ROLE_PARAM_ENTITY_NAME;
 472   }
 473   return common(state, tok);
 474 }
 475
 476 static int PTRCALL
 477 entity2(PROLOG_STATE *state,
 478         int tok,
 479         const char *ptr,
 480         const char *end,
 481         const ENCODING *enc)
 482 {
 483   switch (tok) {
 484   case XML_TOK_PROLOG_S:
 485     return XML_ROLE_ENTITY_NONE;
 486   case XML_TOK_NAME:
 487     if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
 488       state->handler = entity4;
 489       return XML_ROLE_ENTITY_NONE;
 490     }
 491     if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
 492       state->handler = entity3;
 493       return XML_ROLE_ENTITY_NONE;
 494     }
 495     break;
 496   case XML_TOK_LITERAL:
 497     state->handler = declClose;
 498     state->role_none = XML_ROLE_ENTITY_NONE;
 499     return XML_ROLE_ENTITY_VALUE;
 500   }
 501   return common(state, tok);
 502 }
 503
 504 static int PTRCALL
 505 entity3(PROLOG_STATE *state,
 506         int tok,
 507         const char *ptr,
 508         const char *end,
 509         const ENCODING *enc)
 510 {
 511   switch (tok) {
 512   case XML_TOK_PROLOG_S:
 513     return XML_ROLE_ENTITY_NONE;
 514   case XML_TOK_LITERAL:
 515     state->handler = entity4;
 516     return XML_ROLE_ENTITY_PUBLIC_ID;
 517   }
 518   return common(state, tok);
 519 }
 520
 521 static int PTRCALL
 522 entity4(PROLOG_STATE *state,
 523         int tok,
 524         const char *ptr,
 525         const char *end,
 526         const ENCODING *enc)
 527 {
 528   switch (tok) {
 529   case XML_TOK_PROLOG_S:
 530     return XML_ROLE_ENTITY_NONE;
 531   case XML_TOK_LITERAL:
 532     state->handler = entity5;
 533     return XML_ROLE_ENTITY_SYSTEM_ID;
 534   }
 535   return common(state, tok);
 536 }
 537
 538 static int PTRCALL
 539 entity5(PROLOG_STATE *state,
 540         int tok,
 541         const char *ptr,
 542         const char *end,
 543         const ENCODING *enc)
 544 {
 545   switch (tok) {
 546   case XML_TOK_PROLOG_S:
 547     return XML_ROLE_ENTITY_NONE;
 548   case XML_TOK_DECL_CLOSE:
 549     setTopLevel(state);
 550     return XML_ROLE_ENTITY_COMPLETE;
 551   case XML_TOK_NAME:
 552     if (XmlNameMatchesAscii(enc, ptr, end, KW_NDATA)) {
 553       state->handler = entity6;
 554       return XML_ROLE_ENTITY_NONE;
 555     }
 556     break;
 557   }
 558   return common(state, tok);
 559 }
 560
 561 static int PTRCALL
 562 entity6(PROLOG_STATE *state,
 563         int tok,
 564         const char *ptr,
 565         const char *end,
 566         const ENCODING *enc)
 567 {
 568   switch (tok) {
 569   case XML_TOK_PROLOG_S:
 570     return XML_ROLE_ENTITY_NONE;
 571   case XML_TOK_NAME:
 572     state->handler = declClose;
 573     state->role_none = XML_ROLE_ENTITY_NONE;
 574     return XML_ROLE_ENTITY_NOTATION_NAME;
 575   }
 576   return common(state, tok);
 577 }
 578
 579 static int PTRCALL
 580 entity7(PROLOG_STATE *state,
 581         int tok,
 582         const char *ptr,
 583         const char *end,
 584         const ENCODING *enc)
 585 {
 586   switch (tok) {
 587   case XML_TOK_PROLOG_S:
 588     return XML_ROLE_ENTITY_NONE;
 589   case XML_TOK_NAME:
 590     if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
 591       state->handler = entity9;
 592       return XML_ROLE_ENTITY_NONE;
 593     }
 594     if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
 595       state->handler = entity8;
 596       return XML_ROLE_ENTITY_NONE;
 597     }
 598     break;
 599   case XML_TOK_LITERAL:
 600     state->handler = declClose;
 601     state->role_none = XML_ROLE_ENTITY_NONE;
 602     return XML_ROLE_ENTITY_VALUE;
 603   }
 604   return common(state, tok);
 605 }
 606
 607 static int PTRCALL
 608 entity8(PROLOG_STATE *state,
 609         int tok,
 610         const char *ptr,
 611         const char *end,
 612         const ENCODING *enc)
 613 {
 614   switch (tok) {
 615   case XML_TOK_PROLOG_S:
 616     return XML_ROLE_ENTITY_NONE;
 617   case XML_TOK_LITERAL:
 618     state->handler = entity9;
 619     return XML_ROLE_ENTITY_PUBLIC_ID;
 620   }
 621   return common(state, tok);
 622 }
 623
 624 static int PTRCALL
 625 entity9(PROLOG_STATE *state,
 626         int tok,
 627         const char *ptr,
 628         const char *end,
 629         const ENCODING *enc)
 630 {
 631   switch (tok) {
 632   case XML_TOK_PROLOG_S:
 633     return XML_ROLE_ENTITY_NONE;
 634   case XML_TOK_LITERAL:
 635     state->handler = entity10;
 636     return XML_ROLE_ENTITY_SYSTEM_ID;
 637   }
 638   return common(state, tok);
 639 }
 640
 641 static int PTRCALL
 642 entity10(PROLOG_STATE *state,
 643          int tok,
 644          const char *ptr,
 645          const char *end,
 646          const ENCODING *enc)
 647 {
 648   switch (tok) {
 649   case XML_TOK_PROLOG_S:
 650     return XML_ROLE_ENTITY_NONE;
 651   case XML_TOK_DECL_CLOSE:
 652     setTopLevel(state);
 653     return XML_ROLE_ENTITY_COMPLETE;
 654   }
 655   return common(state, tok);
 656 }
 657
 658 static int PTRCALL
 659 notation0(PROLOG_STATE *state,
 660           int tok,
 661           const char *ptr,
 662           const char *end,
 663           const ENCODING *enc)
 664 {
 665   switch (tok) {
 666   case XML_TOK_PROLOG_S:
 667     return XML_ROLE_NOTATION_NONE;
 668   case XML_TOK_NAME:
 669     state->handler = notation1;
 670     return XML_ROLE_NOTATION_NAME;
 671   }
 672   return common(state, tok);
 673 }
 674
 675 static int PTRCALL
 676 notation1(PROLOG_STATE *state,
 677           int tok,
 678           const char *ptr,
 679           const char *end,
 680           const ENCODING *enc)
 681 {
 682   switch (tok) {
 683   case XML_TOK_PROLOG_S:
 684     return XML_ROLE_NOTATION_NONE;
 685   case XML_TOK_NAME:
 686     if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
 687       state->handler = notation3;
 688       return XML_ROLE_NOTATION_NONE;
 689     }
 690     if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
 691       state->handler = notation2;
 692       return XML_ROLE_NOTATION_NONE;
 693     }
 694     break;
 695   }
 696   return common(state, tok);
 697 }
 698
 699 static int PTRCALL
 700 notation2(PROLOG_STATE *state,
 701           int tok,
 702           const char *ptr,
 703           const char *end,
 704           const ENCODING *enc)
 705 {
 706   switch (tok) {
 707   case XML_TOK_PROLOG_S:
 708     return XML_ROLE_NOTATION_NONE;
 709   case XML_TOK_LITERAL:
 710     state->handler = notation4;
 711     return XML_ROLE_NOTATION_PUBLIC_ID;
 712   }
 713   return common(state, tok);
 714 }
 715
 716 static int PTRCALL
 717 notation3(PROLOG_STATE *state,
 718           int tok,
 719           const char *ptr,
 720           const char *end,
 721           const ENCODING *enc)
 722 {
 723   switch (tok) {
 724   case XML_TOK_PROLOG_S:
 725     return XML_ROLE_NOTATION_NONE;
 726   case XML_TOK_LITERAL:
 727     state->handler = declClose;
 728     state->role_none = XML_ROLE_NOTATION_NONE;
 729     return XML_ROLE_NOTATION_SYSTEM_ID;
 730   }
 731   return common(state, tok);
 732 }
 733
 734 static int PTRCALL
 735 notation4(PROLOG_STATE *state,
 736           int tok,
 737           const char *ptr,
 738           const char *end,
 739           const ENCODING *enc)
 740 {
 741   switch (tok) {
 742   case XML_TOK_PROLOG_S:
 743     return XML_ROLE_NOTATION_NONE;
 744   case XML_TOK_LITERAL:
 745     state->handler = declClose;
 746     state->role_none = XML_ROLE_NOTATION_NONE;
 747     return XML_ROLE_NOTATION_SYSTEM_ID;
 748   case XML_TOK_DECL_CLOSE:
 749     setTopLevel(state);
 750     return XML_ROLE_NOTATION_NO_SYSTEM_ID;
 751   }
 752   return common(state, tok);
 753 }
 754
 755 static int PTRCALL
 756 attlist0(PROLOG_STATE *state,
 757          int tok,
 758          const char *ptr,
 759          const char *end,
 760          const ENCODING *enc)
 761 {
 762   switch (tok) {
 763   case XML_TOK_PROLOG_S:
 764     return XML_ROLE_ATTLIST_NONE;
 765   case XML_TOK_NAME:
 766   case XML_TOK_PREFIXED_NAME:
 767     state->handler = attlist1;
 768     return XML_ROLE_ATTLIST_ELEMENT_NAME;
 769   }
 770   return common(state, tok);
 771 }
 772
 773 static int PTRCALL
 774 attlist1(PROLOG_STATE *state,
 775          int tok,
 776          const char *ptr,
 777          const char *end,
 778          const ENCODING *enc)
 779 {
 780   switch (tok) {
 781   case XML_TOK_PROLOG_S:
 782     return XML_ROLE_ATTLIST_NONE;
 783   case XML_TOK_DECL_CLOSE:
 784     setTopLevel(state);
 785     return XML_ROLE_ATTLIST_NONE;
 786   case XML_TOK_NAME:
 787   case XML_TOK_PREFIXED_NAME:
 788     state->handler = attlist2;
 789     return XML_ROLE_ATTRIBUTE_NAME;
 790   }
 791   return common(state, tok);
 792 }
 793
 794 static int PTRCALL
 795 attlist2(PROLOG_STATE *state,
 796          int tok,
 797          const char *ptr,
 798          const char *end,
 799          const ENCODING *enc)
 800 {
 801   switch (tok) {
 802   case XML_TOK_PROLOG_S:
 803     return XML_ROLE_ATTLIST_NONE;
 804   case XML_TOK_NAME:
 805     {
 806       static const char * const types[] = {
 807         KW_CDATA,
 808         KW_ID,
 809         KW_IDREF,
 810         KW_IDREFS,
 811         KW_ENTITY,
 812         KW_ENTITIES,
 813         KW_NMTOKEN,
 814         KW_NMTOKENS,
 815       };
 816       int i;
 817       for (i = 0; i < (int)(sizeof(types)/sizeof(types[0])); i++)
 818         if (XmlNameMatchesAscii(enc, ptr, end, types[i])) {
 819           state->handler = attlist8;
 820           return XML_ROLE_ATTRIBUTE_TYPE_CDATA + i;
 821         }
 822     }
 823     if (XmlNameMatchesAscii(enc, ptr, end, KW_NOTATION)) {
 824       state->handler = attlist5;
 825       return XML_ROLE_ATTLIST_NONE;
 826     }
 827     break;
 828   case XML_TOK_OPEN_PAREN:
 829     state->handler = attlist3;
 830     return XML_ROLE_ATTLIST_NONE;
 831   }
 832   return common(state, tok);
 833 }
 834
 835 static int PTRCALL
 836 attlist3(PROLOG_STATE *state,
 837          int tok,
 838          const char *ptr,
 839          const char *end,
 840          const ENCODING *enc)
 841 {
 842   switch (tok) {
 843   case XML_TOK_PROLOG_S:
 844     return XML_ROLE_ATTLIST_NONE;
 845   case XML_TOK_NMTOKEN:
 846   case XML_TOK_NAME:
 847   case XML_TOK_PREFIXED_NAME:
 848     state->handler = attlist4;
 849     return XML_ROLE_ATTRIBUTE_ENUM_VALUE;
 850   }
 851   return common(state, tok);
 852 }
 853
 854 static int PTRCALL
 855 attlist4(PROLOG_STATE *state,
 856          int tok,
 857          const char *ptr,
 858          const char *end,
 859          const ENCODING *enc)
 860 {
 861   switch (tok) {
 862   case XML_TOK_PROLOG_S:
 863     return XML_ROLE_ATTLIST_NONE;
 864   case XML_TOK_CLOSE_PAREN:
 865     state->handler = attlist8;
 866     return XML_ROLE_ATTLIST_NONE;
 867   case XML_TOK_OR:
 868     state->handler = attlist3;
 869     return XML_ROLE_ATTLIST_NONE;
 870   }
 871   return common(state, tok);
 872 }
 873
 874 static int PTRCALL
 875 attlist5(PROLOG_STATE *state,
 876          int tok,
 877          const char *ptr,
 878          const char *end,
 879          const ENCODING *enc)
 880 {
 881   switch (tok) {
 882   case XML_TOK_PROLOG_S:
 883     return XML_ROLE_ATTLIST_NONE;
 884   case XML_TOK_OPEN_PAREN:
 885     state->handler = attlist6;
 886     return XML_ROLE_ATTLIST_NONE;
 887   }
 888   return common(state, tok);
 889 }
 890
 891 static int PTRCALL
 892 attlist6(PROLOG_STATE *state,
 893          int tok,
 894          const char *ptr,
 895          const char *end,
 896          const ENCODING *enc)
 897 {
 898   switch (tok) {
 899   case XML_TOK_PROLOG_S:
 900     return XML_ROLE_ATTLIST_NONE;
 901   case XML_TOK_NAME:
 902     state->handler = attlist7;
 903     return XML_ROLE_ATTRIBUTE_NOTATION_VALUE;
 904   }
 905   return common(state, tok);
 906 }
 907
 908 static int PTRCALL
 909 attlist7(PROLOG_STATE *state,
 910          int tok,
 911          const char *ptr,
 912          const char *end,
 913          const ENCODING *enc)
 914 {
 915   switch (tok) {
 916   case XML_TOK_PROLOG_S:
 917     return XML_ROLE_ATTLIST_NONE;
 918   case XML_TOK_CLOSE_PAREN:
 919     state->handler = attlist8;
 920     return XML_ROLE_ATTLIST_NONE;
 921   case XML_TOK_OR:
 922     state->handler = attlist6;
 923     return XML_ROLE_ATTLIST_NONE;
 924   }
 925   return common(state, tok);
 926 }
 927
 928 /* default value */
 929 static int PTRCALL
 930 attlist8(PROLOG_STATE *state,
 931          int tok,
 932          const char *ptr,
 933          const char *end,
 934          const ENCODING *enc)
 935 {
 936   switch (tok) {
 937   case XML_TOK_PROLOG_S:
 938     return XML_ROLE_ATTLIST_NONE;
 939   case XML_TOK_POUND_NAME:
 940     if (XmlNameMatchesAscii(enc,
 941                             ptr + MIN_BYTES_PER_CHAR(enc),
 942                             end,
 943                             KW_IMPLIED)) {
 944       state->handler = attlist1;
 945       return XML_ROLE_IMPLIED_ATTRIBUTE_VALUE;
 946     }
 947     if (XmlNameMatchesAscii(enc,
 948                             ptr + MIN_BYTES_PER_CHAR(enc),
 949                             end,
 950                             KW_REQUIRED)) {
 951       state->handler = attlist1;
 952       return XML_ROLE_REQUIRED_ATTRIBUTE_VALUE;
 953     }
 954     if (XmlNameMatchesAscii(enc,
 955                             ptr + MIN_BYTES_PER_CHAR(enc),
 956                             end,
 957                             KW_FIXED)) {
 958       state->handler = attlist9;
 959       return XML_ROLE_ATTLIST_NONE;
 960     }
 961     break;
 962   case XML_TOK_LITERAL:
 963     state->handler = attlist1;
 964     return XML_ROLE_DEFAULT_ATTRIBUTE_VALUE;
 965   }
 966   return common(state, tok);
 967 }
 968
 969 static int PTRCALL
 970 attlist9(PROLOG_STATE *state,
 971          int tok,
 972          const char *ptr,
 973          const char *end,
 974          const ENCODING *enc)
 975 {
 976   switch (tok) {
 977   case XML_TOK_PROLOG_S:
 978     return XML_ROLE_ATTLIST_NONE;
 979   case XML_TOK_LITERAL:
 980     state->handler = attlist1;
 981     return XML_ROLE_FIXED_ATTRIBUTE_VALUE;
 982   }
 983   return common(state, tok);
 984 }
 985
 986 static int PTRCALL
 987 element0(PROLOG_STATE *state,
 988          int tok,
 989          const char *ptr,
 990          const char *end,
 991          const ENCODING *enc)
 992 {
 993   switch (tok) {
 994   case XML_TOK_PROLOG_S:
 995     return XML_ROLE_ELEMENT_NONE;
 996   case XML_TOK_NAME:
 997   case XML_TOK_PREFIXED_NAME:
 998     state->handler = element1;
 999     return XML_ROLE_ELEMENT_NAME;
1000   }
1001   return common(state, tok);
1002 }
1003
1004 static int PTRCALL
1005 element1(PROLOG_STATE *state,
1006          int tok,
1007          const char *ptr,
1008          const char *end,
1009          const ENCODING *enc)
1010 {
1011   switch (tok) {
1012   case XML_TOK_PROLOG_S:
1013     return XML_ROLE_ELEMENT_NONE;
1014   case XML_TOK_NAME:
1015     if (XmlNameMatchesAscii(enc, ptr, end, KW_EMPTY)) {
1016       state->handler = declClose;
1017       state->role_none = XML_ROLE_ELEMENT_NONE;
1018       return XML_ROLE_CONTENT_EMPTY;
1019     }
1020     if (XmlNameMatchesAscii(enc, ptr, end, KW_ANY)) {
1021       state->handler = declClose;
1022       state->role_none = XML_ROLE_ELEMENT_NONE;
1023       return XML_ROLE_CONTENT_ANY;
1024     }
1025     break;
1026   case XML_TOK_OPEN_PAREN:
1027     state->handler = element2;
1028     state->level = 1;
1029     return XML_ROLE_GROUP_OPEN;
1030   }
1031   return common(state, tok);
1032 }
1033
1034 static int PTRCALL
1035 element2(PROLOG_STATE *state,
1036          int tok,
1037          const char *ptr,
1038          const char *end,
1039          const ENCODING *enc)
1040 {
1041   switch (tok) {
1042   case XML_TOK_PROLOG_S:
1043     return XML_ROLE_ELEMENT_NONE;
1044   case XML_TOK_POUND_NAME:
1045     if (XmlNameMatchesAscii(enc,
1046                             ptr + MIN_BYTES_PER_CHAR(enc),
1047                             end,
1048                             KW_PCDATA)) {
1049       state->handler = element3;
1050       return XML_ROLE_CONTENT_PCDATA;
1051     }
1052     break;
1053   case XML_TOK_OPEN_PAREN:
1054     state->level = 2;
1055     state->handler = element6;
1056     return XML_ROLE_GROUP_OPEN;
1057   case XML_TOK_NAME:
1058   case XML_TOK_PREFIXED_NAME:
1059     state->handler = element7;
1060     return XML_ROLE_CONTENT_ELEMENT;
1061   case XML_TOK_NAME_QUESTION:
1062     state->handler = element7;
1063     return XML_ROLE_CONTENT_ELEMENT_OPT;
1064   case XML_TOK_NAME_ASTERISK:
1065     state->handler = element7;
1066     return XML_ROLE_CONTENT_ELEMENT_REP;
1067   case XML_TOK_NAME_PLUS:
1068     state->handler = element7;
1069     return XML_ROLE_CONTENT_ELEMENT_PLUS;
1070   }
1071   return common(state, tok);
1072 }
1073
1074 static int PTRCALL
1075 element3(PROLOG_STATE *state,
1076          int tok,
1077          const char *ptr,
1078          const char *end,
1079          const ENCODING *enc)
1080 {
1081   switch (tok) {
1082   case XML_TOK_PROLOG_S:
1083     return XML_ROLE_ELEMENT_NONE;
1084   case XML_TOK_CLOSE_PAREN:
1085     state->handler = declClose;
1086     state->role_none = XML_ROLE_ELEMENT_NONE;
1087     return XML_ROLE_GROUP_CLOSE;
1088   case XML_TOK_CLOSE_PAREN_ASTERISK:
1089     state->handler = declClose;
1090     state->role_none = XML_ROLE_ELEMENT_NONE;
1091     return XML_ROLE_GROUP_CLOSE_REP;
1092   case XML_TOK_OR:
1093     state->handler = element4;
1094     return XML_ROLE_ELEMENT_NONE;
1095   }
1096   return common(state, tok);
1097 }
1098
1099 static int PTRCALL
1100 element4(PROLOG_STATE *state,
1101          int tok,
1102          const char *ptr,
1103          const char *end,
1104          const ENCODING *enc)
1105 {
1106   switch (tok) {
1107   case XML_TOK_PROLOG_S:
1108     return XML_ROLE_ELEMENT_NONE;
1109   case XML_TOK_NAME:
1110   case XML_TOK_PREFIXED_NAME:
1111     state->handler = element5;
1112     return XML_ROLE_CONTENT_ELEMENT;
1113   }
1114   return common(state, tok);
1115 }
1116
1117 static int PTRCALL
1118 element5(PROLOG_STATE *state,
1119          int tok,
1120          const char *ptr,
1121          const char *end,
1122          const ENCODING *enc)
1123 {
1124   switch (tok) {
1125   case XML_TOK_PROLOG_S:
1126     return XML_ROLE_ELEMENT_NONE;
1127   case XML_TOK_CLOSE_PAREN_ASTERISK:
1128     state->handler = declClose;
1129     state->role_none = XML_ROLE_ELEMENT_NONE;
1130     return XML_ROLE_GROUP_CLOSE_REP;
1131   case XML_TOK_OR:
1132     state->handler = element4;
1133     return XML_ROLE_ELEMENT_NONE;
1134   }
1135   return common(state, tok);
1136 }
1137
1138 static int PTRCALL
1139 element6(PROLOG_STATE *state,
1140          int tok,
1141          const char *ptr,
1142          const char *end,
1143          const ENCODING *enc)
1144 {
1145   switch (tok) {
1146   case XML_TOK_PROLOG_S:
1147     return XML_ROLE_ELEMENT_NONE;
1148   case XML_TOK_OPEN_PAREN:
1149     state->level += 1;
1150     return XML_ROLE_GROUP_OPEN;
1151   case XML_TOK_NAME:
1152   case XML_TOK_PREFIXED_NAME:
1153     state->handler = element7;
1154     return XML_ROLE_CONTENT_ELEMENT;
1155   case XML_TOK_NAME_QUESTION:
1156     state->handler = element7;
1157     return XML_ROLE_CONTENT_ELEMENT_OPT;
1158   case XML_TOK_NAME_ASTERISK:
1159     state->handler = element7;
1160     return XML_ROLE_CONTENT_ELEMENT_REP;
1161   case XML_TOK_NAME_PLUS:
1162     state->handler = element7;
1163     return XML_ROLE_CONTENT_ELEMENT_PLUS;
1164   }
1165   return common(state, tok);
1166 }
1167
1168 static int PTRCALL
1169 element7(PROLOG_STATE *state,
1170          int tok,
1171          const char *ptr,
1172          const char *end,
1173          const ENCODING *enc)
1174 {
1175   switch (tok) {
1176   case XML_TOK_PROLOG_S:
1177     return XML_ROLE_ELEMENT_NONE;
1178   case XML_TOK_CLOSE_PAREN:
1179     state->level -= 1;
1180     if (state->level == 0) {
1181       state->handler = declClose;
1182       state->role_none = XML_ROLE_ELEMENT_NONE;
1183     }
1184     return XML_ROLE_GROUP_CLOSE;
1185   case XML_TOK_CLOSE_PAREN_ASTERISK:
1186     state->level -= 1;
1187     if (state->level == 0) {
1188       state->handler = declClose;
1189       state->role_none = XML_ROLE_ELEMENT_NONE;
1190     }
1191     return XML_ROLE_GROUP_CLOSE_REP;
1192   case XML_TOK_CLOSE_PAREN_QUESTION:
1193     state->level -= 1;
1194     if (state->level == 0) {
1195       state->handler = declClose;
1196       state->role_none = XML_ROLE_ELEMENT_NONE;
1197     }
1198     return XML_ROLE_GROUP_CLOSE_OPT;
1199   case XML_TOK_CLOSE_PAREN_PLUS:
1200     state->level -= 1;
1201     if (state->level == 0) {
1202       state->handler = declClose;
1203       state->role_none = XML_ROLE_ELEMENT_NONE;
1204     }
1205     return XML_ROLE_GROUP_CLOSE_PLUS;
1206   case XML_TOK_COMMA:
1207     state->handler = element6;
1208     return XML_ROLE_GROUP_SEQUENCE;
1209   case XML_TOK_OR:
1210     state->handler = element6;
1211     return XML_ROLE_GROUP_CHOICE;
1212   }
1213   return common(state, tok);
1214 }
1215
1216 #ifdef XML_DTD
1217
1218 static int PTRCALL
1219 condSect0(PROLOG_STATE *state,
1220           int tok,
1221           const char *ptr,
1222           const char *end,
1223           const ENCODING *enc)
1224 {
1225   switch (tok) {
1226   case XML_TOK_PROLOG_S:
1227     return XML_ROLE_NONE;
1228   case XML_TOK_NAME:
1229     if (XmlNameMatchesAscii(enc, ptr, end, KW_INCLUDE)) {
1230       state->handler = condSect1;
1231       return XML_ROLE_NONE;
1232     }
1233     if (XmlNameMatchesAscii(enc, ptr, end, KW_IGNORE)) {
1234       state->handler = condSect2;
1235       return XML_ROLE_NONE;
1236     }
1237     break;
1238   }
1239   return common(state, tok);
1240 }
1241
1242 static int PTRCALL
1243 condSect1(PROLOG_STATE *state,
1244           int tok,
1245           const char *ptr,
1246           const char *end,
1247           const ENCODING *enc)
1248 {
1249   switch (tok) {
1250   case XML_TOK_PROLOG_S:
1251     return XML_ROLE_NONE;
1252   case XML_TOK_OPEN_BRACKET:
1253     state->handler = externalSubset1;
1254     state->includeLevel += 1;
1255     return XML_ROLE_NONE;
1256   }
1257   return common(state, tok);
1258 }
1259
1260 static int PTRCALL
1261 condSect2(PROLOG_STATE *state,
1262           int tok,
1263           const char *ptr,
1264           const char *end,
1265           const ENCODING *enc)
1266 {
1267   switch (tok) {
1268   case XML_TOK_PROLOG_S:
1269     return XML_ROLE_NONE;
1270   case XML_TOK_OPEN_BRACKET:
1271     state->handler = externalSubset1;
1272     return XML_ROLE_IGNORE_SECT;
1273   }
1274   return common(state, tok);
1275 }
1276
1277 #endif /* XML_DTD */
1278
1279 static int PTRCALL
1280 declClose(PROLOG_STATE *state,
1281           int tok,
1282           const char *ptr,
1283           const char *end,
1284           const ENCODING *enc)
1285 {
1286   switch (tok) {
1287   case XML_TOK_PROLOG_S:
1288     return state->role_none;
1289   case XML_TOK_DECL_CLOSE:
1290     setTopLevel(state);
1291     return state->role_none;
1292   }
1293   return common(state, tok);
1294 }
1295
1296 static int PTRCALL
1297 error(PROLOG_STATE *state,
1298       int tok,
1299       const char *ptr,
1300       const char *end,
1301       const ENCODING *enc)
1302 {
1303   return XML_ROLE_NONE;
1304 }
1305
1306 static int FASTCALL
1307 common(PROLOG_STATE *state, int tok)
1308 {
1309 #ifdef XML_DTD
1310   if (!state->documentEntity && tok == XML_TOK_PARAM_ENTITY_REF)
1311     return XML_ROLE_INNER_PARAM_ENTITY_REF;
1312 #endif
1313   state->handler = error;
1314   return XML_ROLE_ERROR;
1315 }
1316
1317 void
1318 XmlPrologStateInit(PROLOG_STATE *state)
1319 {
1320   state->handler = prolog0;
1321 #ifdef XML_DTD
1322   state->documentEntity = 1;
1323   state->includeLevel = 0;
1324   state->inEntityValue = 0;
1325 #endif /* XML_DTD */
1326 }
1327
1328 #ifdef XML_DTD
1329
1330 void
1331 XmlPrologStateInitExternalEntity(PROLOG_STATE *state)
1332 {
1333   state->handler = externalSubset0;
1334   state->documentEntity = 0;
1335   state->includeLevel = 0;
1336 }
1337
1338 #endif /* XML_DTD */