src/xrc/expat/xmltok/xmltok_impl.c

   1 /*
   2 Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
   3 See the file copying.txt for copying permission.
   4 */
   5
   6 #undef INVALID_LEAD_CASE
   7
   8 #ifndef IS_INVALID_CHAR
   9 #define IS_INVALID_CHAR(enc, ptr, n) (0)
  10 #define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
  11     case BT_LEAD ## n: \
  12       if (end - ptr < n) \
  13         return XML_TOK_PARTIAL_CHAR; \
  14       ptr += n; \
  15       break;
  16 #else
  17 #define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
  18     case BT_LEAD ## n: \
  19       if (end - ptr < n) \
  20         return XML_TOK_PARTIAL_CHAR; \
  21       if (IS_INVALID_CHAR(enc, ptr, n)) { \
  22         *(nextTokPtr) = (ptr); \
  23         return XML_TOK_INVALID; \
  24       } \
  25       ptr += n; \
  26       break;
  27 #endif
  28
  29 #define INVALID_CASES(ptr, nextTokPtr) \
  30   INVALID_LEAD_CASE(2, ptr, nextTokPtr) \
  31   INVALID_LEAD_CASE(3, ptr, nextTokPtr) \
  32   INVALID_LEAD_CASE(4, ptr, nextTokPtr) \
  33   case BT_NONXML: \
  34   case BT_MALFORM: \
  35   case BT_TRAIL: \
  36     *(nextTokPtr) = (ptr); \
  37     return XML_TOK_INVALID;
  38
  39 #define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \
  40    case BT_LEAD ## n: \
  41      if (end - ptr < n) \
  42        return XML_TOK_PARTIAL_CHAR; \
  43      if (!IS_NAME_CHAR(enc, ptr, n)) { \
  44        *nextTokPtr = ptr; \
  45        return XML_TOK_INVALID; \
  46      } \
  47      ptr += n; \
  48      break;
  49
  50 #define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \
  51   case BT_NONASCII: \
  52     if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \
  53       *nextTokPtr = ptr; \
  54       return XML_TOK_INVALID; \
  55     } \
  56   case BT_NMSTRT: \
  57   case BT_HEX: \
  58   case BT_DIGIT: \
  59   case BT_NAME: \
  60   case BT_MINUS: \
  61     ptr += MINBPC(enc); \
  62     break; \
  63   CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \
  64   CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \
  65   CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr)
  66
  67 #define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \
  68    case BT_LEAD ## n: \
  69      if (end - ptr < n) \
  70        return XML_TOK_PARTIAL_CHAR; \
  71      if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \
  72        *nextTokPtr = ptr; \
  73        return XML_TOK_INVALID; \
  74      } \
  75      ptr += n; \
  76      break;
  77
  78 #define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \
  79   case BT_NONASCII: \
  80     if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \
  81       *nextTokPtr = ptr; \
  82       return XML_TOK_INVALID; \
  83     } \
  84   case BT_NMSTRT: \
  85   case BT_HEX: \
  86     ptr += MINBPC(enc); \
  87     break; \
  88   CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
  89   CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
  90   CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)
  91
  92 #ifndef PREFIX
  93 #define PREFIX(ident) ident
  94 #endif
  95
  96 /* ptr points to character following "<!-" */
  97
  98 static
  99 int PREFIX(scanComment)(const ENCODING *enc, const char *ptr, const char *end,
 100                         const char **nextTokPtr)
 101 {
 102   if (ptr != end) {
 103     if (!CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
 104       *nextTokPtr = ptr;
 105       return XML_TOK_INVALID;
 106     }
 107     ptr += MINBPC(enc);
 108     while (ptr != end) {
 109       switch (BYTE_TYPE(enc, ptr)) {
 110       INVALID_CASES(ptr, nextTokPtr)
 111       case BT_MINUS:
 112         if ((ptr += MINBPC(enc)) == end)
 113           return XML_TOK_PARTIAL;
 114         if (CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
 115           if ((ptr += MINBPC(enc)) == end)
 116             return XML_TOK_PARTIAL;
 117           if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 118             *nextTokPtr = ptr;
 119             return XML_TOK_INVALID;
 120           }
 121           *nextTokPtr = ptr + MINBPC(enc);
 122           return XML_TOK_COMMENT;
 123         }
 124         break;
 125       default:
 126         ptr += MINBPC(enc);
 127         break;
 128       }
 129     }
 130   }
 131   return XML_TOK_PARTIAL;
 132 }
 133
 134 /* ptr points to character following "<!" */
 135
 136 static
 137 int PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, const char *end,
 138                      const char **nextTokPtr)
 139 {
 140   if (ptr == end)
 141     return XML_TOK_PARTIAL;
 142   switch (BYTE_TYPE(enc, ptr)) {
 143   case BT_MINUS:
 144     return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 145   case BT_LSQB:
 146     *nextTokPtr = ptr + MINBPC(enc);
 147     return XML_TOK_COND_SECT_OPEN;
 148   case BT_NMSTRT:
 149   case BT_HEX:
 150     ptr += MINBPC(enc);
 151     break;
 152   default:
 153     *nextTokPtr = ptr;
 154     return XML_TOK_INVALID;
 155   }
 156   while (ptr != end) {
 157     switch (BYTE_TYPE(enc, ptr)) {
 158     case BT_PERCNT:
 159       if (ptr + MINBPC(enc) == end)
 160         return XML_TOK_PARTIAL;
 161       /* don't allow <!ENTITY% foo "whatever"> */
 162       switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) {
 163       case BT_S: case BT_CR: case BT_LF: case BT_PERCNT:
 164         *nextTokPtr = ptr;
 165         return XML_TOK_INVALID;
 166       }
 167       /* fall through */
 168     case BT_S: case BT_CR: case BT_LF:
 169       *nextTokPtr = ptr;
 170       return XML_TOK_DECL_OPEN;
 171     case BT_NMSTRT:
 172     case BT_HEX:
 173       ptr += MINBPC(enc);
 174       break;
 175     default:
 176       *nextTokPtr = ptr;
 177       return XML_TOK_INVALID;
 178     }
 179   }
 180   return XML_TOK_PARTIAL;
 181 }
 182
 183 static
 184 int PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, const char *end, int *tokPtr)
 185 {
 186   int upper = 0;
 187   *tokPtr = XML_TOK_PI;
 188   if (end - ptr != MINBPC(enc)*3)
 189     return 1;
 190   switch (BYTE_TO_ASCII(enc, ptr)) {
 191   case ASCII_x:
 192     break;
 193   case ASCII_X:
 194     upper = 1;
 195     break;
 196   default:
 197     return 1;
 198   }
 199   ptr += MINBPC(enc);
 200   switch (BYTE_TO_ASCII(enc, ptr)) {
 201   case ASCII_m:
 202     break;
 203   case ASCII_M:
 204     upper = 1;
 205     break;
 206   default:
 207     return 1;
 208   }
 209   ptr += MINBPC(enc);
 210   switch (BYTE_TO_ASCII(enc, ptr)) {
 211   case ASCII_l:
 212     break;
 213   case ASCII_L:
 214     upper = 1;
 215     break;
 216   default:
 217     return 1;
 218   }
 219   if (upper)
 220     return 0;
 221   *tokPtr = XML_TOK_XML_DECL;
 222   return 1;
 223 }
 224
 225 /* ptr points to character following "<?" */
 226
 227 static
 228 int PREFIX(scanPi)(const ENCODING *enc, const char *ptr, const char *end,
 229                    const char **nextTokPtr)
 230 {
 231   int tok;
 232   const char *target = ptr;
 233   if (ptr == end)
 234     return XML_TOK_PARTIAL;
 235   switch (BYTE_TYPE(enc, ptr)) {
 236   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 237   default:
 238     *nextTokPtr = ptr;
 239     return XML_TOK_INVALID;
 240   }
 241   while (ptr != end) {
 242     switch (BYTE_TYPE(enc, ptr)) {
 243     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 244     case BT_S: case BT_CR: case BT_LF:
 245       if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
 246         *nextTokPtr = ptr;
 247         return XML_TOK_INVALID;
 248       }
 249       ptr += MINBPC(enc);
 250       while (ptr != end) {
 251         switch (BYTE_TYPE(enc, ptr)) {
 252         INVALID_CASES(ptr, nextTokPtr)
 253         case BT_QUEST:
 254           ptr += MINBPC(enc);
 255           if (ptr == end)
 256             return XML_TOK_PARTIAL;
 257           if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 258             *nextTokPtr = ptr + MINBPC(enc);
 259             return tok;
 260           }
 261           break;
 262         default:
 263           ptr += MINBPC(enc);
 264           break;
 265         }
 266       }
 267       return XML_TOK_PARTIAL;
 268     case BT_QUEST:
 269       if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
 270         *nextTokPtr = ptr;
 271         return XML_TOK_INVALID;
 272       }
 273       ptr += MINBPC(enc);
 274       if (ptr == end)
 275         return XML_TOK_PARTIAL;
 276       if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 277         *nextTokPtr = ptr + MINBPC(enc);
 278         return tok;
 279       }
 280       /* fall through */
 281     default:
 282       *nextTokPtr = ptr;
 283       return XML_TOK_INVALID;
 284     }
 285   }
 286   return XML_TOK_PARTIAL;
 287 }
 288
 289
 290 static
 291 int PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, const char *end,
 292                              const char **nextTokPtr)
 293 {
 294   static const char CDATA_LSQB[] = { ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, ASCII_LSQB };
 295   int i;
 296   /* CDATA[ */
 297   if (end - ptr < 6 * MINBPC(enc))
 298     return XML_TOK_PARTIAL;
 299   for (i = 0; i < 6; i++, ptr += MINBPC(enc)) {
 300     if (!CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) {
 301       *nextTokPtr = ptr;
 302       return XML_TOK_INVALID;
 303     }
 304   }
 305   *nextTokPtr = ptr;
 306   return XML_TOK_CDATA_SECT_OPEN;
 307 }
 308
 309 static
 310 int PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
 311                             const char **nextTokPtr)
 312 {
 313   if (ptr == end)
 314     return XML_TOK_NONE;
 315 #if !(MINBPC(enc) == 1)
 316   if (MINBPC(enc) > 1) {
 317     size_t n = end - ptr;
 318     if (n & (MINBPC(enc) - 1)) {
 319       n &= ~(MINBPC(enc) - 1);
 320       if (n == 0)
 321         return XML_TOK_PARTIAL;
 322       end = ptr + n;
 323     }
 324  }
 325 #endif
 326   switch (BYTE_TYPE(enc, ptr)) {
 327   case BT_RSQB:
 328     ptr += MINBPC(enc);
 329     if (ptr == end)
 330       return XML_TOK_PARTIAL;
 331     if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
 332       break;
 333     ptr += MINBPC(enc);
 334     if (ptr == end)
 335       return XML_TOK_PARTIAL;
 336     if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 337       ptr -= MINBPC(enc);
 338       break;
 339     }
 340     *nextTokPtr = ptr + MINBPC(enc);
 341     return XML_TOK_CDATA_SECT_CLOSE;
 342   case BT_CR:
 343     ptr += MINBPC(enc);
 344     if (ptr == end)
 345       return XML_TOK_PARTIAL;
 346     if (BYTE_TYPE(enc, ptr) == BT_LF)
 347       ptr += MINBPC(enc);
 348     *nextTokPtr = ptr;
 349     return XML_TOK_DATA_NEWLINE;
 350   case BT_LF:
 351     *nextTokPtr = ptr + MINBPC(enc);
 352     return XML_TOK_DATA_NEWLINE;
 353   INVALID_CASES(ptr, nextTokPtr)
 354   default:
 355     ptr += MINBPC(enc);
 356     break;
 357   }
 358   while (ptr != end) {
 359     switch (BYTE_TYPE(enc, ptr)) {
 360 #define LEAD_CASE(n) \
 361     case BT_LEAD ## n: \
 362       if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
 363         *nextTokPtr = ptr; \
 364         return XML_TOK_DATA_CHARS; \
 365       } \
 366       ptr += n; \
 367       break;
 368     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
 369 #undef LEAD_CASE
 370     case BT_NONXML:
 371     case BT_MALFORM:
 372     case BT_TRAIL:
 373     case BT_CR:
 374     case BT_LF:
 375     case BT_RSQB:
 376       *nextTokPtr = ptr;
 377       return XML_TOK_DATA_CHARS;
 378     default:
 379       ptr += MINBPC(enc);
 380       break;
 381     }
 382   }
 383   *nextTokPtr = ptr;
 384   return XML_TOK_DATA_CHARS;
 385 }
 386
 387 /* ptr points to character following "</" */
 388
 389 static
 390 int PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, const char *end,
 391                        const char **nextTokPtr)
 392 {
 393   if (ptr == end)
 394     return XML_TOK_PARTIAL;
 395   switch (BYTE_TYPE(enc, ptr)) {
 396   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 397   default:
 398     *nextTokPtr = ptr;
 399     return XML_TOK_INVALID;
 400   }
 401   while (ptr != end) {
 402     switch (BYTE_TYPE(enc, ptr)) {
 403     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 404     case BT_S: case BT_CR: case BT_LF:
 405       for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
 406         switch (BYTE_TYPE(enc, ptr)) {
 407         case BT_S: case BT_CR: case BT_LF:
 408           break;
 409         case BT_GT:
 410           *nextTokPtr = ptr + MINBPC(enc);
 411           return XML_TOK_END_TAG;
 412         default:
 413           *nextTokPtr = ptr;
 414           return XML_TOK_INVALID;
 415         }
 416       }
 417       return XML_TOK_PARTIAL;
 418 #ifdef XML_NS
 419     case BT_COLON:
 420       /* no need to check qname syntax here, since end-tag must match exactly */
 421       ptr += MINBPC(enc);
 422       break;
 423 #endif
 424     case BT_GT:
 425       *nextTokPtr = ptr + MINBPC(enc);
 426       return XML_TOK_END_TAG;
 427     default:
 428       *nextTokPtr = ptr;
 429       return XML_TOK_INVALID;
 430     }
 431   }
 432   return XML_TOK_PARTIAL;
 433 }
 434
 435 /* ptr points to character following "&#X" */
 436
 437 static
 438 int PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, const char *end,
 439                            const char **nextTokPtr)
 440 {
 441   if (ptr != end) {
 442     switch (BYTE_TYPE(enc, ptr)) {
 443     case BT_DIGIT:
 444     case BT_HEX:
 445       break;
 446     default:
 447       *nextTokPtr = ptr;
 448       return XML_TOK_INVALID;
 449     }
 450     for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
 451       switch (BYTE_TYPE(enc, ptr)) {
 452       case BT_DIGIT:
 453       case BT_HEX:
 454         break;
 455       case BT_SEMI:
 456         *nextTokPtr = ptr + MINBPC(enc);
 457         return XML_TOK_CHAR_REF;
 458       default:
 459         *nextTokPtr = ptr;
 460         return XML_TOK_INVALID;
 461       }
 462     }
 463   }
 464   return XML_TOK_PARTIAL;
 465 }
 466
 467 /* ptr points to character following "&#" */
 468
 469 static
 470 int PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, const char *end,
 471                         const char **nextTokPtr)
 472 {
 473   if (ptr != end) {
 474     if (CHAR_MATCHES(enc, ptr, ASCII_x))
 475       return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 476     switch (BYTE_TYPE(enc, ptr)) {
 477     case BT_DIGIT:
 478       break;
 479     default:
 480       *nextTokPtr = ptr;
 481       return XML_TOK_INVALID;
 482     }
 483     for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
 484       switch (BYTE_TYPE(enc, ptr)) {
 485       case BT_DIGIT:
 486         break;
 487       case BT_SEMI:
 488         *nextTokPtr = ptr + MINBPC(enc);
 489         return XML_TOK_CHAR_REF;
 490       default:
 491         *nextTokPtr = ptr;
 492         return XML_TOK_INVALID;
 493       }
 494     }
 495   }
 496   return XML_TOK_PARTIAL;
 497 }
 498
 499 /* ptr points to character following "&" */
 500
 501 static
 502 int PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end,
 503                     const char **nextTokPtr)
 504 {
 505   if (ptr == end)
 506     return XML_TOK_PARTIAL;
 507   switch (BYTE_TYPE(enc, ptr)) {
 508   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 509   case BT_NUM:
 510     return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 511   default:
 512     *nextTokPtr = ptr;
 513     return XML_TOK_INVALID;
 514   }
 515   while (ptr != end) {
 516     switch (BYTE_TYPE(enc, ptr)) {
 517     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 518     case BT_SEMI:
 519       *nextTokPtr = ptr + MINBPC(enc);
 520       return XML_TOK_ENTITY_REF;
 521     default:
 522       *nextTokPtr = ptr;
 523       return XML_TOK_INVALID;
 524     }
 525   }
 526   return XML_TOK_PARTIAL;
 527 }
 528
 529 /* ptr points to character following first character of attribute name */
 530
 531 static
 532 int PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end,
 533                      const char **nextTokPtr)
 534 {
 535 #ifdef XML_NS
 536   int hadColon = 0;
 537 #endif
 538   while (ptr != end) {
 539     switch (BYTE_TYPE(enc, ptr)) {
 540     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 541 #ifdef XML_NS
 542     case BT_COLON:
 543       if (hadColon) {
 544         *nextTokPtr = ptr;
 545         return XML_TOK_INVALID;
 546       }
 547       hadColon = 1;
 548       ptr += MINBPC(enc);
 549       if (ptr == end)
 550         return XML_TOK_PARTIAL;
 551       switch (BYTE_TYPE(enc, ptr)) {
 552       CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 553       default:
 554         *nextTokPtr = ptr;
 555         return XML_TOK_INVALID;
 556       }
 557       break;
 558 #endif
 559     case BT_S: case BT_CR: case BT_LF:
 560       for (;;) {
 561         int t;
 562
 563         ptr += MINBPC(enc);
 564         if (ptr == end)
 565           return XML_TOK_PARTIAL;
 566         t = BYTE_TYPE(enc, ptr);
 567         if (t == BT_EQUALS)
 568           break;
 569         switch (t) {
 570         case BT_S:
 571         case BT_LF:
 572         case BT_CR:
 573           break;
 574         default:
 575           *nextTokPtr = ptr;
 576           return XML_TOK_INVALID;
 577         }
 578       }
 579     /* fall through */
 580     case BT_EQUALS:
 581       {
 582         int open;
 583 #ifdef XML_NS
 584         hadColon = 0;
 585 #endif
 586         for (;;) {
 587
 588           ptr += MINBPC(enc);
 589           if (ptr == end)
 590             return XML_TOK_PARTIAL;
 591           open = BYTE_TYPE(enc, ptr);
 592           if (open == BT_QUOT || open == BT_APOS)
 593             break;
 594           switch (open) {
 595           case BT_S:
 596           case BT_LF:
 597           case BT_CR:
 598             break;
 599           default:
 600             *nextTokPtr = ptr;
 601             return XML_TOK_INVALID;
 602           }
 603         }
 604         ptr += MINBPC(enc);
 605         /* in attribute value */
 606         for (;;) {
 607           int t;
 608           if (ptr == end)
 609             return XML_TOK_PARTIAL;
 610           t = BYTE_TYPE(enc, ptr);
 611           if (t == open)
 612             break;
 613           switch (t) {
 614           INVALID_CASES(ptr, nextTokPtr)
 615           case BT_AMP:
 616             {
 617               int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
 618               if (tok <= 0) {
 619                 if (tok == XML_TOK_INVALID)
 620                   *nextTokPtr = ptr;
 621                 return tok;
 622               }
 623               break;
 624             }
 625           case BT_LT:
 626             *nextTokPtr = ptr;
 627             return XML_TOK_INVALID;
 628           default:
 629             ptr += MINBPC(enc);
 630             break;
 631           }
 632         }
 633         ptr += MINBPC(enc);
 634         if (ptr == end)
 635           return XML_TOK_PARTIAL;
 636         switch (BYTE_TYPE(enc, ptr)) {
 637         case BT_S:
 638         case BT_CR:
 639         case BT_LF:
 640           break;
 641         case BT_SOL:
 642           goto sol;
 643         case BT_GT:
 644           goto gt;
 645         default:
 646           *nextTokPtr = ptr;
 647           return XML_TOK_INVALID;
 648         }
 649         /* ptr points to closing quote */
 650         for (;;) {
 651           ptr += MINBPC(enc);
 652           if (ptr == end)
 653             return XML_TOK_PARTIAL;
 654           switch (BYTE_TYPE(enc, ptr)) {
 655           CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 656           case BT_S: case BT_CR: case BT_LF:
 657             continue;
 658           case BT_GT:
 659           gt:
 660             *nextTokPtr = ptr + MINBPC(enc);
 661             return XML_TOK_START_TAG_WITH_ATTS;
 662           case BT_SOL:
 663           sol:
 664             ptr += MINBPC(enc);
 665             if (ptr == end)
 666               return XML_TOK_PARTIAL;
 667             if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 668               *nextTokPtr = ptr;
 669               return XML_TOK_INVALID;
 670             }
 671             *nextTokPtr = ptr + MINBPC(enc);
 672             return XML_TOK_EMPTY_ELEMENT_WITH_ATTS;
 673           default:
 674             *nextTokPtr = ptr;
 675             return XML_TOK_INVALID;
 676           }
 677           break;
 678         }
 679         break;
 680       }
 681     default:
 682       *nextTokPtr = ptr;
 683       return XML_TOK_INVALID;
 684     }
 685   }
 686   return XML_TOK_PARTIAL;
 687 }
 688
 689 /* ptr points to character following "<" */
 690
 691 static
 692 int PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end,
 693                    const char **nextTokPtr)
 694 {
 695 #ifdef XML_NS
 696   int hadColon;
 697 #endif
 698   if (ptr == end)
 699     return XML_TOK_PARTIAL;
 700   switch (BYTE_TYPE(enc, ptr)) {
 701   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 702   case BT_EXCL:
 703     if ((ptr += MINBPC(enc)) == end)
 704       return XML_TOK_PARTIAL;
 705     switch (BYTE_TYPE(enc, ptr)) {
 706     case BT_MINUS:
 707       return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 708     case BT_LSQB:
 709       return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 710     }
 711     *nextTokPtr = ptr;
 712     return XML_TOK_INVALID;
 713   case BT_QUEST:
 714     return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 715   case BT_SOL:
 716     return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 717   default:
 718     *nextTokPtr = ptr;
 719     return XML_TOK_INVALID;
 720   }
 721 #ifdef XML_NS
 722   hadColon = 0;
 723 #endif
 724   /* we have a start-tag */
 725   while (ptr != end) {
 726     switch (BYTE_TYPE(enc, ptr)) {
 727     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 728 #ifdef XML_NS
 729     case BT_COLON:
 730       if (hadColon) {
 731         *nextTokPtr = ptr;
 732         return XML_TOK_INVALID;
 733       }
 734       hadColon = 1;
 735       ptr += MINBPC(enc);
 736       if (ptr == end)
 737         return XML_TOK_PARTIAL;
 738       switch (BYTE_TYPE(enc, ptr)) {
 739       CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 740       default:
 741         *nextTokPtr = ptr;
 742         return XML_TOK_INVALID;
 743       }
 744       break;
 745 #endif
 746     case BT_S: case BT_CR: case BT_LF:
 747       {
 748         ptr += MINBPC(enc);
 749         while (ptr != end) {
 750           switch (BYTE_TYPE(enc, ptr)) {
 751           CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 752           case BT_GT:
 753             goto gt;
 754           case BT_SOL:
 755             goto sol;
 756           case BT_S: case BT_CR: case BT_LF:
 757             ptr += MINBPC(enc);
 758             continue;
 759           default:
 760             *nextTokPtr = ptr;
 761             return XML_TOK_INVALID;
 762           }
 763           return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
 764         }
 765         return XML_TOK_PARTIAL;
 766       }
 767     case BT_GT:
 768     gt:
 769       *nextTokPtr = ptr + MINBPC(enc);
 770       return XML_TOK_START_TAG_NO_ATTS;
 771     case BT_SOL:
 772     sol:
 773       ptr += MINBPC(enc);
 774       if (ptr == end)
 775         return XML_TOK_PARTIAL;
 776       if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 777         *nextTokPtr = ptr;
 778         return XML_TOK_INVALID;
 779       }
 780       *nextTokPtr = ptr + MINBPC(enc);
 781       return XML_TOK_EMPTY_ELEMENT_NO_ATTS;
 782     default:
 783       *nextTokPtr = ptr;
 784       return XML_TOK_INVALID;
 785     }
 786   }
 787   return XML_TOK_PARTIAL;
 788 }
 789
 790 static
 791 int PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end,
 792                        const char **nextTokPtr)
 793 {
 794   if (ptr == end)
 795     return XML_TOK_NONE;
 796 #if !(MINBPC(enc) == 1)
 797   if (MINBPC(enc) > 1) {
 798     size_t n = end - ptr;
 799     if (n & (MINBPC(enc) - 1)) {
 800       n &= ~(MINBPC(enc) - 1);
 801       if (n == 0)
 802         return XML_TOK_PARTIAL;
 803       end = ptr + n;
 804     }
 805   }
 806 #endif
 807   switch (BYTE_TYPE(enc, ptr)) {
 808   case BT_LT:
 809     return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 810   case BT_AMP:
 811     return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
 812   case BT_CR:
 813     ptr += MINBPC(enc);
 814     if (ptr == end)
 815       return XML_TOK_TRAILING_CR;
 816     if (BYTE_TYPE(enc, ptr) == BT_LF)
 817       ptr += MINBPC(enc);
 818     *nextTokPtr = ptr;
 819     return XML_TOK_DATA_NEWLINE;
 820   case BT_LF:
 821     *nextTokPtr = ptr + MINBPC(enc);
 822     return XML_TOK_DATA_NEWLINE;
 823   case BT_RSQB:
 824     ptr += MINBPC(enc);
 825     if (ptr == end)
 826       return XML_TOK_TRAILING_RSQB;
 827     if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
 828       break;
 829     ptr += MINBPC(enc);
 830     if (ptr == end)
 831       return XML_TOK_TRAILING_RSQB;
 832     if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
 833       ptr -= MINBPC(enc);
 834       break;
 835     }
 836     *nextTokPtr = ptr;
 837     return XML_TOK_INVALID;
 838   INVALID_CASES(ptr, nextTokPtr)
 839   default:
 840     ptr += MINBPC(enc);
 841     break;
 842   }
 843   while (ptr != end) {
 844     switch (BYTE_TYPE(enc, ptr)) {
 845 #define LEAD_CASE(n) \
 846     case BT_LEAD ## n: \
 847       if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
 848         *nextTokPtr = ptr; \
 849         return XML_TOK_DATA_CHARS; \
 850       } \
 851       ptr += n; \
 852       break;
 853     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
 854 #undef LEAD_CASE
 855     case BT_RSQB:
 856       if (ptr + MINBPC(enc) != end) {
 857          if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) {
 858            ptr += MINBPC(enc);
 859            break;
 860          }
 861          if (ptr + 2*MINBPC(enc) != end) {
 862            if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), ASCII_GT)) {
 863              ptr += MINBPC(enc);
 864              break;
 865            }
 866            *nextTokPtr = ptr + 2*MINBPC(enc);
 867            return XML_TOK_INVALID;
 868          }
 869       }
 870       /* fall through */
 871     case BT_AMP:
 872     case BT_LT:
 873     case BT_NONXML:
 874     case BT_MALFORM:
 875     case BT_TRAIL:
 876     case BT_CR:
 877     case BT_LF:
 878       *nextTokPtr = ptr;
 879       return XML_TOK_DATA_CHARS;
 880     default:
 881       ptr += MINBPC(enc);
 882       break;
 883     }
 884   }
 885   *nextTokPtr = ptr;
 886   return XML_TOK_DATA_CHARS;
 887 }
 888
 889 /* ptr points to character following "%" */
 890
 891 static
 892 int PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end,
 893                         const char **nextTokPtr)
 894 {
 895   if (ptr == end)
 896     return XML_TOK_PARTIAL;
 897   switch (BYTE_TYPE(enc, ptr)) {
 898   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 899   case BT_S: case BT_LF: case BT_CR: case BT_PERCNT:
 900     *nextTokPtr = ptr;
 901     return XML_TOK_PERCENT;
 902   default:
 903     *nextTokPtr = ptr;
 904     return XML_TOK_INVALID;
 905   }
 906   while (ptr != end) {
 907     switch (BYTE_TYPE(enc, ptr)) {
 908     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 909     case BT_SEMI:
 910       *nextTokPtr = ptr + MINBPC(enc);
 911       return XML_TOK_PARAM_ENTITY_REF;
 912     default:
 913       *nextTokPtr = ptr;
 914       return XML_TOK_INVALID;
 915     }
 916   }
 917   return XML_TOK_PARTIAL;
 918 }
 919
 920 static
 921 int PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end,
 922                           const char **nextTokPtr)
 923 {
 924   if (ptr == end)
 925     return XML_TOK_PARTIAL;
 926   switch (BYTE_TYPE(enc, ptr)) {
 927   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
 928   default:
 929     *nextTokPtr = ptr;
 930     return XML_TOK_INVALID;
 931   }
 932   while (ptr != end) {
 933     switch (BYTE_TYPE(enc, ptr)) {
 934     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 935     case BT_CR: case BT_LF: case BT_S:
 936     case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR:
 937       *nextTokPtr = ptr;
 938       return XML_TOK_POUND_NAME;
 939     default:
 940       *nextTokPtr = ptr;
 941       return XML_TOK_INVALID;
 942     }
 943   }
 944   return -XML_TOK_POUND_NAME;
 945 }
 946
 947 static
 948 int PREFIX(scanLit)(int open, const ENCODING *enc,
 949                     const char *ptr, const char *end,
 950                     const char **nextTokPtr)
 951 {
 952   while (ptr != end) {
 953     int t = BYTE_TYPE(enc, ptr);
 954     switch (t) {
 955     INVALID_CASES(ptr, nextTokPtr)
 956     case BT_QUOT:
 957     case BT_APOS:
 958       ptr += MINBPC(enc);
 959       if (t != open)
 960         break;
 961       if (ptr == end)
 962         return -XML_TOK_LITERAL;
 963       *nextTokPtr = ptr;
 964       switch (BYTE_TYPE(enc, ptr)) {
 965       case BT_S: case BT_CR: case BT_LF:
 966       case BT_GT: case BT_PERCNT: case BT_LSQB:
 967         return XML_TOK_LITERAL;
 968       default:
 969         return XML_TOK_INVALID;
 970       }
 971     default:
 972       ptr += MINBPC(enc);
 973       break;
 974     }
 975   }
 976   return XML_TOK_PARTIAL;
 977 }
 978
 979 static
 980 int PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end,
 981                       const char **nextTokPtr)
 982 {
 983   int tok;
 984   if (ptr == end)
 985     return XML_TOK_NONE;
 986 #if !(MINBPC(enc) == 1)
 987   if (MINBPC(enc) > 1) {
 988     size_t n = end - ptr;
 989     if (n & (MINBPC(enc) - 1)) {
 990       n &= ~(MINBPC(enc) - 1);
 991       if (n == 0)
 992         return XML_TOK_PARTIAL;
 993       end = ptr + n;
 994     }
 995   }
 996 #endif
 997   switch (BYTE_TYPE(enc, ptr)) {
 998   case BT_QUOT:
 999     return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr);
1000   case BT_APOS:
1001     return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr);
1002   case BT_LT:
1003     {
1004       ptr += MINBPC(enc);
1005       if (ptr == end)
1006         return XML_TOK_PARTIAL;
1007       switch (BYTE_TYPE(enc, ptr)) {
1008       case BT_EXCL:
1009         return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1010       case BT_QUEST:
1011         return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1012       case BT_NMSTRT:
1013       case BT_HEX:
1014       case BT_NONASCII:
1015       case BT_LEAD2:
1016       case BT_LEAD3:
1017       case BT_LEAD4:
1018         *nextTokPtr = ptr - MINBPC(enc);
1019         return XML_TOK_INSTANCE_START;
1020       }
1021       *nextTokPtr = ptr;
1022       return XML_TOK_INVALID;
1023     }
1024   case BT_CR:
1025     if (ptr + MINBPC(enc) == end)
1026       return -XML_TOK_PROLOG_S;
1027     /* fall through */
1028   case BT_S: case BT_LF:
1029     for (;;) {
1030       ptr += MINBPC(enc);
1031       if (ptr == end)
1032         break;
1033       switch (BYTE_TYPE(enc, ptr)) {
1034       case BT_S: case BT_LF:
1035         break;
1036       case BT_CR:
1037         /* don't split CR/LF pair */
1038         if (ptr + MINBPC(enc) != end)
1039           break;
1040         /* fall through */
1041       default:
1042         *nextTokPtr = ptr;
1043         return XML_TOK_PROLOG_S;
1044       }
1045     }
1046     *nextTokPtr = ptr;
1047     return XML_TOK_PROLOG_S;
1048   case BT_PERCNT:
1049     return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1050   case BT_COMMA:
1051     *nextTokPtr = ptr + MINBPC(enc);
1052     return XML_TOK_COMMA;
1053   case BT_LSQB:
1054     *nextTokPtr = ptr + MINBPC(enc);
1055     return XML_TOK_OPEN_BRACKET;
1056   case BT_RSQB:
1057     ptr += MINBPC(enc);
1058     if (ptr == end)
1059       return -XML_TOK_CLOSE_BRACKET;
1060     if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
1061       if (ptr + MINBPC(enc) == end)
1062         return XML_TOK_PARTIAL;
1063       if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_GT)) {
1064         *nextTokPtr = ptr + 2*MINBPC(enc);
1065         return XML_TOK_COND_SECT_CLOSE;
1066       }
1067     }
1068     *nextTokPtr = ptr;
1069     return XML_TOK_CLOSE_BRACKET;
1070   case BT_LPAR:
1071     *nextTokPtr = ptr + MINBPC(enc);
1072     return XML_TOK_OPEN_PAREN;
1073   case BT_RPAR:
1074     ptr += MINBPC(enc);
1075     if (ptr == end)
1076       return -XML_TOK_CLOSE_PAREN;
1077     switch (BYTE_TYPE(enc, ptr)) {
1078     case BT_AST:
1079       *nextTokPtr = ptr + MINBPC(enc);
1080       return XML_TOK_CLOSE_PAREN_ASTERISK;
1081     case BT_QUEST:
1082       *nextTokPtr = ptr + MINBPC(enc);
1083       return XML_TOK_CLOSE_PAREN_QUESTION;
1084     case BT_PLUS:
1085       *nextTokPtr = ptr + MINBPC(enc);
1086       return XML_TOK_CLOSE_PAREN_PLUS;
1087     case BT_CR: case BT_LF: case BT_S:
1088     case BT_GT: case BT_COMMA: case BT_VERBAR:
1089     case BT_RPAR:
1090       *nextTokPtr = ptr;
1091       return XML_TOK_CLOSE_PAREN;
1092     }
1093     *nextTokPtr = ptr;
1094     return XML_TOK_INVALID;
1095   case BT_VERBAR:
1096     *nextTokPtr = ptr + MINBPC(enc);
1097     return XML_TOK_OR;
1098   case BT_GT:
1099     *nextTokPtr = ptr + MINBPC(enc);
1100     return XML_TOK_DECL_CLOSE;
1101   case BT_NUM:
1102     return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1103 #ifdef XML_MIN_SIZE
1104 #define LEAD_CASE(n) \
1105   case BT_LEAD ## n: \
1106     if (end - ptr < n) \
1107       return XML_TOK_PARTIAL_CHAR; \
1108     if (IS_NMSTRT_CHAR(enc, ptr, n)) { \
1109       ptr += n; \
1110       tok = XML_TOK_NAME; \
1111       break; \
1112     } \
1113     if (IS_NAME_CHAR(enc, ptr, n)) { \
1114       ptr += n; \
1115       tok = XML_TOK_NMTOKEN; \
1116       break; \
1117     } \
1118     *nextTokPtr = ptr; \
1119     return XML_TOK_INVALID;
1120 #else
1121 #define LEAD_CASE(n) \
1122     case BT_LEAD ## n: \
1123     if (end - ptr < n) \
1124       return XML_TOK_PARTIAL_CHAR; \
1125     *nextTokPtr = ptr; \
1126     return XML_TOK_INVALID;
1127 #endif
1128     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
1129 #undef LEAD_CASE
1130   case BT_NMSTRT:
1131   case BT_HEX:
1132     tok = XML_TOK_NAME;
1133     ptr += MINBPC(enc);
1134     break;
1135   case BT_DIGIT:
1136   case BT_NAME:
1137   case BT_MINUS:
1138 #ifdef XML_NS
1139   case BT_COLON:
1140 #endif
1141     tok = XML_TOK_NMTOKEN;
1142     ptr += MINBPC(enc);
1143     break;
1144   case BT_NONASCII:
1145 #ifdef XML_MIN_SIZE
1146     if (IS_NMSTRT_CHAR_MINBPC(enc, ptr)) {
1147       ptr += MINBPC(enc);
1148       tok = XML_TOK_NAME;
1149       break;
1150     }
1151     if (IS_NAME_CHAR_MINBPC(enc, ptr)) {
1152       ptr += MINBPC(enc);
1153       tok = XML_TOK_NMTOKEN;
1154       break;
1155     }
1156 #endif
1157     /* fall through */
1158   default:
1159     *nextTokPtr = ptr;
1160     return XML_TOK_INVALID;
1161   }
1162   while (ptr != end) {
1163     switch (BYTE_TYPE(enc, ptr)) {
1164     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
1165     case BT_GT: case BT_RPAR: case BT_COMMA:
1166     case BT_VERBAR: case BT_LSQB: case BT_PERCNT:
1167     case BT_S: case BT_CR: case BT_LF:
1168       *nextTokPtr = ptr;
1169       return tok;
1170 #ifdef XML_NS
1171     case BT_COLON:
1172       ptr += MINBPC(enc);
1173       switch (tok) {
1174       case XML_TOK_NAME:
1175         if (ptr == end)
1176           return XML_TOK_PARTIAL;
1177         tok = XML_TOK_PREFIXED_NAME;
1178         switch (BYTE_TYPE(enc, ptr)) {
1179         CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
1180         default:
1181           tok = XML_TOK_NMTOKEN;
1182           break;
1183         }
1184         break;
1185       case XML_TOK_PREFIXED_NAME:
1186         tok = XML_TOK_NMTOKEN;
1187         break;
1188       }
1189       break;
1190 #endif
1191     case BT_PLUS:
1192       if (tok == XML_TOK_NMTOKEN)  {
1193         *nextTokPtr = ptr;
1194         return XML_TOK_INVALID;
1195       }
1196       *nextTokPtr = ptr + MINBPC(enc);
1197       return XML_TOK_NAME_PLUS;
1198     case BT_AST:
1199       if (tok == XML_TOK_NMTOKEN)  {
1200         *nextTokPtr = ptr;
1201         return XML_TOK_INVALID;
1202       }
1203       *nextTokPtr = ptr + MINBPC(enc);
1204       return XML_TOK_NAME_ASTERISK;
1205     case BT_QUEST:
1206       if (tok == XML_TOK_NMTOKEN)  {
1207         *nextTokPtr = ptr;
1208         return XML_TOK_INVALID;
1209       }
1210       *nextTokPtr = ptr + MINBPC(enc);
1211       return XML_TOK_NAME_QUESTION;
1212     default:
1213       *nextTokPtr = ptr;
1214       return XML_TOK_INVALID;
1215     }
1216   }
1217   return -tok;
1218 }
1219
1220 static
1221 int PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr, const char *end,
1222                               const char **nextTokPtr)
1223 {
1224   const char *start;
1225   if (ptr == end)
1226     return XML_TOK_NONE;
1227   start = ptr;
1228   while (ptr != end) {
1229     switch (BYTE_TYPE(enc, ptr)) {
1230 #define LEAD_CASE(n) \
1231     case BT_LEAD ## n: ptr += n; break;
1232     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
1233 #undef LEAD_CASE
1234     case BT_AMP:
1235       if (ptr == start)
1236         return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1237       *nextTokPtr = ptr;
1238       return XML_TOK_DATA_CHARS;
1239     case BT_LT:
1240       /* this is for inside entity references */
1241       *nextTokPtr = ptr;
1242       return XML_TOK_INVALID;
1243     case BT_LF:
1244       if (ptr == start) {
1245         *nextTokPtr = ptr + MINBPC(enc);
1246         return XML_TOK_DATA_NEWLINE;
1247       }
1248       *nextTokPtr = ptr;
1249       return XML_TOK_DATA_CHARS;
1250     case BT_CR:
1251       if (ptr == start) {
1252         ptr += MINBPC(enc);
1253         if (ptr == end)
1254           return XML_TOK_TRAILING_CR;
1255         if (BYTE_TYPE(enc, ptr) == BT_LF)
1256           ptr += MINBPC(enc);
1257         *nextTokPtr = ptr;
1258         return XML_TOK_DATA_NEWLINE;
1259       }
1260       *nextTokPtr = ptr;
1261       return XML_TOK_DATA_CHARS;
1262     case BT_S:
1263       if (ptr == start) {
1264         *nextTokPtr = ptr + MINBPC(enc);
1265         return XML_TOK_ATTRIBUTE_VALUE_S;
1266       }
1267       *nextTokPtr = ptr;
1268       return XML_TOK_DATA_CHARS;
1269     default:
1270       ptr += MINBPC(enc);
1271       break;
1272     }
1273   }
1274   *nextTokPtr = ptr;
1275   return XML_TOK_DATA_CHARS;
1276 }
1277
1278 static
1279 int PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr, const char *end,
1280                            const char **nextTokPtr)
1281 {
1282   const char *start;
1283   if (ptr == end)
1284     return XML_TOK_NONE;
1285   start = ptr;
1286   while (ptr != end) {
1287     switch (BYTE_TYPE(enc, ptr)) {
1288 #define LEAD_CASE(n) \
1289     case BT_LEAD ## n: ptr += n; break;
1290     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
1291 #undef LEAD_CASE
1292     case BT_AMP:
1293       if (ptr == start)
1294         return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1295       *nextTokPtr = ptr;
1296       return XML_TOK_DATA_CHARS;
1297     case BT_PERCNT:
1298       if (ptr == start)
1299         return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
1300       *nextTokPtr = ptr;
1301       return XML_TOK_DATA_CHARS;
1302     case BT_LF:
1303       if (ptr == start) {
1304         *nextTokPtr = ptr + MINBPC(enc);
1305         return XML_TOK_DATA_NEWLINE;
1306       }
1307       *nextTokPtr = ptr;
1308       return XML_TOK_DATA_CHARS;
1309     case BT_CR:
1310       if (ptr == start) {
1311         ptr += MINBPC(enc);
1312         if (ptr == end)
1313           return XML_TOK_TRAILING_CR;
1314         if (BYTE_TYPE(enc, ptr) == BT_LF)
1315           ptr += MINBPC(enc);
1316         *nextTokPtr = ptr;
1317         return XML_TOK_DATA_NEWLINE;
1318       }
1319       *nextTokPtr = ptr;
1320       return XML_TOK_DATA_CHARS;
1321     default:
1322       ptr += MINBPC(enc);
1323       break;
1324     }
1325   }
1326   *nextTokPtr = ptr;
1327   return XML_TOK_DATA_CHARS;
1328 }
1329
1330 #ifdef XML_DTD
1331
1332 static
1333 int PREFIX(ignoreSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
1334                              const char **nextTokPtr)
1335 {
1336   int level = 0;
1337   if (MINBPC(enc) > 1) {
1338     size_t n = end - ptr;
1339     if (n & (MINBPC(enc) - 1)) {
1340       n &= ~(MINBPC(enc) - 1);
1341       end = ptr + n;
1342     }
1343   }
1344   while (ptr != end) {
1345     switch (BYTE_TYPE(enc, ptr)) {
1346     INVALID_CASES(ptr, nextTokPtr)
1347     case BT_LT:
1348       if ((ptr += MINBPC(enc)) == end)
1349         return XML_TOK_PARTIAL;
1350       if (CHAR_MATCHES(enc, ptr, ASCII_EXCL)) {
1351         if ((ptr += MINBPC(enc)) == end)
1352           return XML_TOK_PARTIAL;
1353         if (CHAR_MATCHES(enc, ptr, ASCII_LSQB)) {
1354           ++level;
1355           ptr += MINBPC(enc);
1356         }
1357       }
1358       break;
1359     case BT_RSQB:
1360       if ((ptr += MINBPC(enc)) == end)
1361         return XML_TOK_PARTIAL;
1362       if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
1363         if ((ptr += MINBPC(enc)) == end)
1364           return XML_TOK_PARTIAL;
1365         if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
1366           ptr += MINBPC(enc);
1367           if (level == 0) {
1368             *nextTokPtr = ptr;
1369             return XML_TOK_IGNORE_SECT;
1370           }
1371           --level;
1372         }
1373       }
1374       break;
1375     default:
1376       ptr += MINBPC(enc);
1377       break;
1378     }
1379   }
1380   return XML_TOK_PARTIAL;
1381 }
1382
1383 #endif /* XML_DTD */
1384
1385 static
1386 int PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
1387                        const char **badPtr)
1388 {
1389   ptr += MINBPC(enc);
1390   end -= MINBPC(enc);
1391   for (; ptr != end; ptr += MINBPC(enc)) {
1392     switch (BYTE_TYPE(enc, ptr)) {
1393     case BT_DIGIT:
1394     case BT_HEX:
1395     case BT_MINUS:
1396     case BT_APOS:
1397     case BT_LPAR:
1398     case BT_RPAR:
1399     case BT_PLUS:
1400     case BT_COMMA:
1401     case BT_SOL:
1402     case BT_EQUALS:
1403     case BT_QUEST:
1404     case BT_CR:
1405     case BT_LF:
1406     case BT_SEMI:
1407     case BT_EXCL:
1408     case BT_AST:
1409     case BT_PERCNT:
1410     case BT_NUM:
1411 #ifdef XML_NS
1412     case BT_COLON:
1413 #endif
1414       break;
1415     case BT_S:
1416       if (CHAR_MATCHES(enc, ptr, ASCII_TAB)) {
1417         *badPtr = ptr;
1418         return 0;
1419       }
1420       break;
1421     case BT_NAME:
1422     case BT_NMSTRT:
1423       if (!(BYTE_TO_ASCII(enc, ptr) & ~0x7f))
1424         break;
1425     default:
1426       switch (BYTE_TO_ASCII(enc, ptr)) {
1427       case 0x24: /* $ */
1428       case 0x40: /* @ */
1429         break;
1430       default:
1431         *badPtr = ptr;
1432         return 0;
1433       }
1434       break;
1435     }
1436   }
1437   return 1;
1438 }
1439
1440 /* This must only be called for a well-formed start-tag or empty element tag.
1441 Returns the number of attributes.  Pointers to the first attsMax attributes
1442 are stored in atts. */
1443
1444 static
1445 int PREFIX(getAtts)(const ENCODING *enc, const char *ptr,
1446                     int attsMax, ATTRIBUTE *atts)
1447 {
1448   enum { other, inName, inValue } state = inName;
1449   int nAtts = 0;
1450   int open = 0; /* defined when state == inValue;
1451                    initialization just to shut up compilers */
1452
1453   for (ptr += MINBPC(enc);; ptr += MINBPC(enc)) {
1454     switch (BYTE_TYPE(enc, ptr)) {
1455 #define START_NAME \
1456       if (state == other) { \
1457         if (nAtts < attsMax) { \
1458           atts[nAtts].name = ptr; \
1459           atts[nAtts].normalized = 1; \
1460         } \
1461         state = inName; \
1462       }
1463 #define LEAD_CASE(n) \
1464     case BT_LEAD ## n: START_NAME ptr += (n - MINBPC(enc)); break;
1465     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
1466 #undef LEAD_CASE
1467     case BT_NONASCII:
1468     case BT_NMSTRT:
1469     case BT_HEX:
1470       START_NAME
1471       break;
1472 #undef START_NAME
1473     case BT_QUOT:
1474       if (state != inValue) {
1475         if (nAtts < attsMax)
1476           atts[nAtts].valuePtr = ptr + MINBPC(enc);
1477         state = inValue;
1478         open = BT_QUOT;
1479       }
1480       else if (open == BT_QUOT) {
1481         state = other;
1482         if (nAtts < attsMax)
1483           atts[nAtts].valueEnd = ptr;
1484         nAtts++;
1485       }
1486       break;
1487     case BT_APOS:
1488       if (state != inValue) {
1489         if (nAtts < attsMax)
1490           atts[nAtts].valuePtr = ptr + MINBPC(enc);
1491         state = inValue;
1492         open = BT_APOS;
1493       }
1494       else if (open == BT_APOS) {
1495         state = other;
1496         if (nAtts < attsMax)
1497           atts[nAtts].valueEnd = ptr;
1498         nAtts++;
1499       }
1500       break;
1501     case BT_AMP:
1502       if (nAtts < attsMax)
1503         atts[nAtts].normalized = 0;
1504       break;
1505     case BT_S:
1506       if (state == inName)
1507         state = other;
1508       else if (state == inValue
1509                && nAtts < attsMax
1510                && atts[nAtts].normalized
1511                && (ptr == atts[nAtts].valuePtr
1512                    || BYTE_TO_ASCII(enc, ptr) != ASCII_SPACE
1513                    || BYTE_TO_ASCII(enc, ptr + MINBPC(enc)) == ASCII_SPACE
1514                    || BYTE_TYPE(enc, ptr + MINBPC(enc)) == open))
1515         atts[nAtts].normalized = 0;
1516       break;
1517     case BT_CR: case BT_LF:
1518       /* This case ensures that the first attribute name is counted
1519          Apart from that we could just change state on the quote. */
1520       if (state == inName)
1521         state = other;
1522       else if (state == inValue && nAtts < attsMax)
1523         atts[nAtts].normalized = 0;
1524       break;
1525     case BT_GT:
1526     case BT_SOL:
1527       if (state != inValue)
1528         return nAtts;
1529       break;
1530     default:
1531       break;
1532     }
1533   }
1534   /* not reached */
1535 }
1536
1537 static
1538 int PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr)
1539 {
1540   int result = 0;
1541   /* skip &# */
1542   ptr += 2*MINBPC(enc);
1543   if (CHAR_MATCHES(enc, ptr, ASCII_x)) {
1544     for (ptr += MINBPC(enc); !CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) {
1545       int c = BYTE_TO_ASCII(enc, ptr);
1546       switch (c) {
1547       case ASCII_0: case ASCII_1: case ASCII_2: case ASCII_3: case ASCII_4:
1548       case ASCII_5: case ASCII_6: case ASCII_7: case ASCII_8: case ASCII_9:
1549         result <<= 4;
1550         result |= (c - ASCII_0);
1551         break;
1552       case ASCII_A: case ASCII_B: case ASCII_C: case ASCII_D: case ASCII_E: case ASCII_F:
1553         result <<= 4;
1554         result += 10 + (c - ASCII_A);
1555         break;
1556       case ASCII_a: case ASCII_b: case ASCII_c: case ASCII_d: case ASCII_e: case ASCII_f:
1557         result <<= 4;
1558         result += 10 + (c - ASCII_a);
1559         break;
1560       }
1561       if (result >= 0x110000)
1562         return -1;
1563     }
1564   }
1565   else {
1566     for (; !CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) {
1567       int c = BYTE_TO_ASCII(enc, ptr);
1568       result *= 10;
1569       result += (c - ASCII_0);
1570       if (result >= 0x110000)
1571         return -1;
1572     }
1573   }
1574   return checkCharRefNumber(result);
1575 }
1576
1577 static
1578 int PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr, const char *end)
1579 {
1580   switch ((end - ptr)/MINBPC(enc)) {
1581   case 2:
1582     if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_t)) {
1583       switch (BYTE_TO_ASCII(enc, ptr)) {
1584       case ASCII_l:
1585         return ASCII_LT;
1586       case ASCII_g:
1587         return ASCII_GT;
1588       }
1589     }
1590     break;
1591   case 3:
1592     if (CHAR_MATCHES(enc, ptr, ASCII_a)) {
1593       ptr += MINBPC(enc);
1594       if (CHAR_MATCHES(enc, ptr, ASCII_m)) {
1595         ptr += MINBPC(enc);
1596         if (CHAR_MATCHES(enc, ptr, ASCII_p))
1597           return ASCII_AMP;
1598       }
1599     }
1600     break;
1601   case 4:
1602     switch (BYTE_TO_ASCII(enc, ptr)) {
1603     case ASCII_q:
1604       ptr += MINBPC(enc);
1605       if (CHAR_MATCHES(enc, ptr, ASCII_u)) {
1606         ptr += MINBPC(enc);
1607         if (CHAR_MATCHES(enc, ptr, ASCII_o)) {
1608           ptr += MINBPC(enc);
1609           if (CHAR_MATCHES(enc, ptr, ASCII_t))
1610             return ASCII_QUOT;
1611         }
1612       }
1613       break;
1614     case ASCII_a:
1615       ptr += MINBPC(enc);
1616       if (CHAR_MATCHES(enc, ptr, ASCII_p)) {
1617         ptr += MINBPC(enc);
1618         if (CHAR_MATCHES(enc, ptr, ASCII_o)) {
1619           ptr += MINBPC(enc);
1620           if (CHAR_MATCHES(enc, ptr, ASCII_s))
1621             return ASCII_APOS;
1622         }
1623       }
1624       break;
1625     }
1626   }
1627   return 0;
1628 }
1629
1630 static
1631 int PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2)
1632 {
1633   for (;;) {
1634     switch (BYTE_TYPE(enc, ptr1)) {
1635 #define LEAD_CASE(n) \
1636     case BT_LEAD ## n: \
1637       if (*ptr1++ != *ptr2++) \
1638         return 0;
1639     LEAD_CASE(4) LEAD_CASE(3) LEAD_CASE(2)
1640 #undef LEAD_CASE
1641       /* fall through */
1642       if (*ptr1++ != *ptr2++)
1643         return 0;
1644       break;
1645     case BT_NONASCII:
1646     case BT_NMSTRT:
1647 #ifdef XML_NS
1648     case BT_COLON:
1649 #endif
1650     case BT_HEX:
1651     case BT_DIGIT:
1652     case BT_NAME:
1653     case BT_MINUS:
1654       if (*ptr2++ != *ptr1++)
1655         return 0;
1656       if (MINBPC(enc) > 1) {
1657         if (*ptr2++ != *ptr1++)
1658           return 0;
1659         if (MINBPC(enc) > 2) {
1660           if (*ptr2++ != *ptr1++)
1661             return 0;
1662           if (MINBPC(enc) > 3) {
1663             if (*ptr2++ != *ptr1++)
1664               return 0;
1665           }
1666         }
1667       }
1668       break;
1669     default:
1670       if (MINBPC(enc) == 1 && *ptr1 == *ptr2)
1671         return 1;
1672       switch (BYTE_TYPE(enc, ptr2)) {
1673       case BT_LEAD2:
1674       case BT_LEAD3:
1675       case BT_LEAD4:
1676       case BT_NONASCII:
1677       case BT_NMSTRT:
1678 #ifdef XML_NS
1679       case BT_COLON:
1680 #endif
1681       case BT_HEX:
1682       case BT_DIGIT:
1683       case BT_NAME:
1684       case BT_MINUS:
1685         return 0;
1686       default:
1687         return 1;
1688       }
1689     }
1690   }
1691   /* not reached */
1692 }
1693
1694 static
1695 int PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1,
1696                              const char *end1, const char *ptr2)
1697 {
1698   for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) {
1699     if (ptr1 == end1)
1700       return 0;
1701     if (!CHAR_MATCHES(enc, ptr1, *ptr2))
1702       return 0;
1703   }
1704   return ptr1 == end1;
1705 }
1706
1707 static
1708 int PREFIX(nameLength)(const ENCODING *enc, const char *ptr)
1709 {
1710   const char *start = ptr;
1711   for (;;) {
1712     switch (BYTE_TYPE(enc, ptr)) {
1713 #define LEAD_CASE(n) \
1714     case BT_LEAD ## n: ptr += n; break;
1715     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
1716 #undef LEAD_CASE
1717     case BT_NONASCII:
1718     case BT_NMSTRT:
1719 #ifdef XML_NS
1720     case BT_COLON:
1721 #endif
1722     case BT_HEX:
1723     case BT_DIGIT:
1724     case BT_NAME:
1725     case BT_MINUS:
1726       ptr += MINBPC(enc);
1727       break;
1728     default:
1729       return ptr - start;
1730     }
1731   }
1732 }
1733
1734 static
1735 const char *PREFIX(skipS)(const ENCODING *enc, const char *ptr)
1736 {
1737   for (;;) {
1738     switch (BYTE_TYPE(enc, ptr)) {
1739     case BT_LF:
1740     case BT_CR:
1741     case BT_S:
1742       ptr += MINBPC(enc);
1743       break;
1744     default:
1745       return ptr;
1746     }
1747   }
1748 }
1749
1750 static
1751 void PREFIX(updatePosition)(const ENCODING *enc,
1752                             const char *ptr,
1753                             const char *end,
1754                             POSITION *pos)
1755 {
1756   while (ptr != end) {
1757     switch (BYTE_TYPE(enc, ptr)) {
1758 #define LEAD_CASE(n) \
1759     case BT_LEAD ## n: \
1760       ptr += n; \
1761       break;
1762     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
1763 #undef LEAD_CASE
1764     case BT_LF:
1765       pos->columnNumber = (unsigned)-1;
1766       pos->lineNumber++;
1767       ptr += MINBPC(enc);
1768       break;
1769     case BT_CR:
1770       pos->lineNumber++;
1771       ptr += MINBPC(enc);
1772       if (ptr != end && BYTE_TYPE(enc, ptr) == BT_LF)
1773         ptr += MINBPC(enc);
1774       pos->columnNumber = (unsigned)-1;
1775       break;
1776     default:
1777       ptr += MINBPC(enc);
1778       break;
1779     }
1780     pos->columnNumber++;
1781   }
1782 }
1783
1784 #undef DO_LEAD_CASE
1785 #undef MULTIBYTE_CASES
1786 #undef INVALID_CASES
1787 #undef CHECK_NAME_CASE
1788 #undef CHECK_NAME_CASES
1789 #undef CHECK_NMSTRT_CASE
1790 #undef CHECK_NMSTRT_CASES