Utilities/cmexpat/xmltok.c

   1 /*
   2 Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
   3 See the file COPYING for copying permission.
   4 */
   5
   6 #include <cmexpat/expatConfig.h>
   7
   8 #include "xmltok.h"
   9 #include "nametab.h"
  10
  11 #if defined(__BORLANDC__)
  12 #pragma warn -8008 // Disable "condition is always true" warning.
  13 #pragma warn -8066 // Disable "unreachable code" warning.
  14 #endif
  15
  16 #ifdef XML_DTD
  17 #define IGNORE_SECTION_TOK_VTABLE , PREFIX(ignoreSectionTok)
  18 #else
  19 #define IGNORE_SECTION_TOK_VTABLE /* as nothing */
  20 #endif
  21
  22 #define VTABLE1 \
  23   { PREFIX(prologTok), PREFIX(contentTok), \
  24     PREFIX(cdataSectionTok) IGNORE_SECTION_TOK_VTABLE }, \
  25   { PREFIX(attributeValueTok), PREFIX(entityValueTok) }, \
  26   PREFIX(sameName), \
  27   PREFIX(nameMatchesAscii), \
  28   PREFIX(nameLength), \
  29   PREFIX(skipS), \
  30   PREFIX(getAtts), \
  31   PREFIX(charRefNumber), \
  32   PREFIX(predefinedEntityName), \
  33   PREFIX(updatePosition), \
  34   PREFIX(isPublicId)
  35
  36 #define VTABLE VTABLE1, PREFIX(toUtf8), PREFIX(toUtf16)
  37
  38 #define UCS2_GET_NAMING(pages, hi, lo) \
  39    (namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1 << ((lo) & 0x1F)))
  40
  41 /* A 2 byte UTF-8 representation splits the characters 11 bits
  42 between the bottom 5 and 6 bits of the bytes.
  43 We need 8 bits to index into pages, 3 bits to add to that index and
  44 5 bits to generate the mask. */
  45 #define UTF8_GET_NAMING2(pages, byte) \
  46     (namingBitmap[((pages)[(((byte)[0]) >> 2) & 7] << 3) \
  47                       + ((((byte)[0]) & 3) << 1) \
  48                       + ((((byte)[1]) >> 5) & 1)] \
  49          & (1 << (((byte)[1]) & 0x1F)))
  50
  51 /* A 3 byte UTF-8 representation splits the characters 16 bits
  52 between the bottom 4, 6 and 6 bits of the bytes.
  53 We need 8 bits to index into pages, 3 bits to add to that index and
  54 5 bits to generate the mask. */
  55 #define UTF8_GET_NAMING3(pages, byte) \
  56   (namingBitmap[((pages)[((((byte)[0]) & 0xF) << 4) \
  57                              + ((((byte)[1]) >> 2) & 0xF)] \
  58                        << 3) \
  59                       + ((((byte)[1]) & 3) << 1) \
  60                       + ((((byte)[2]) >> 5) & 1)] \
  61          & (1 << (((byte)[2]) & 0x1F)))
  62
  63 #define UTF8_GET_NAMING(pages, p, n) \
  64   ((n) == 2 \
  65   ? UTF8_GET_NAMING2(pages, (const unsigned char *)(p)) \
  66   : ((n) == 3 \
  67      ? UTF8_GET_NAMING3(pages, (const unsigned char *)(p)) \
  68      : 0))
  69
  70 #define UTF8_INVALID3(p) \
  71   ((*p) == 0xED \
  72   ? (((p)[1] & 0x20) != 0) \
  73   : ((*p) == 0xEF \
  74      ? ((p)[1] == 0xBF && ((p)[2] == 0xBF || (p)[2] == 0xBE)) \
  75      : 0))
  76
  77 #define UTF8_INVALID4(p) ((*p) == 0xF4 && ((p)[1] & 0x30) != 0)
  78
  79 static
  80 int isNever(const ENCODING *enc, const char *p)
  81 {
  82   cmExpatUnused(enc);
  83   cmExpatUnused(p);
  84   return 0;
  85 }
  86
  87 static
  88 int utf8_isName2(const ENCODING *enc, const char *p)
  89 {
  90   cmExpatUnused(enc);
  91   return UTF8_GET_NAMING2(namePages, (const unsigned char *)p);
  92 }
  93
  94 static
  95 int utf8_isName3(const ENCODING *enc, const char *p)
  96 {
  97   cmExpatUnused(enc);
  98   return UTF8_GET_NAMING3(namePages, (const unsigned char *)p);
  99 }
 100
 101 #define utf8_isName4 isNever
 102
 103 static
 104 int utf8_isNmstrt2(const ENCODING *enc, const char *p)
 105 {
 106   cmExpatUnused(enc);
 107   return UTF8_GET_NAMING2(nmstrtPages, (const unsigned char *)p);
 108 }
 109
 110 static
 111 int utf8_isNmstrt3(const ENCODING *enc, const char *p)
 112 {
 113   cmExpatUnused(enc);
 114   return UTF8_GET_NAMING3(nmstrtPages, (const unsigned char *)p);
 115 }
 116
 117 #define utf8_isNmstrt4 isNever
 118
 119 #define utf8_isInvalid2 isNever
 120
 121 static
 122 int utf8_isInvalid3(const ENCODING *enc, const char *p)
 123 {
 124   cmExpatUnused(enc);
 125   return UTF8_INVALID3((const unsigned char *)p);
 126 }
 127
 128 static
 129 int utf8_isInvalid4(const ENCODING *enc, const char *p)
 130 {
 131   cmExpatUnused(enc);
 132   return UTF8_INVALID4((const unsigned char *)p);
 133 }
 134
 135 struct normal_encoding {
 136   ENCODING enc;
 137   unsigned char type[256];
 138 #ifdef XML_MIN_SIZE
 139   int (*byteType)(const ENCODING *, const char *);
 140   int (*isNameMin)(const ENCODING *, const char *);
 141   int (*isNmstrtMin)(const ENCODING *, const char *);
 142   int (*byteToAscii)(const ENCODING *, const char *);
 143   int (*charMatches)(const ENCODING *, const char *, int);
 144 #endif /* XML_MIN_SIZE */
 145   int (*isName2)(const ENCODING *, const char *);
 146   int (*isName3)(const ENCODING *, const char *);
 147   int (*isName4)(const ENCODING *, const char *);
 148   int (*isNmstrt2)(const ENCODING *, const char *);
 149   int (*isNmstrt3)(const ENCODING *, const char *);
 150   int (*isNmstrt4)(const ENCODING *, const char *);
 151   int (*isInvalid2)(const ENCODING *, const char *);
 152   int (*isInvalid3)(const ENCODING *, const char *);
 153   int (*isInvalid4)(const ENCODING *, const char *);
 154 };
 155
 156 #ifdef XML_MIN_SIZE
 157
 158 #define STANDARD_VTABLE(E) \
 159  E ## byteType, \
 160  E ## isNameMin, \
 161  E ## isNmstrtMin, \
 162  E ## byteToAscii, \
 163  E ## charMatches,
 164
 165 #else
 166
 167 #define STANDARD_VTABLE(E) /* as nothing */
 168
 169 #endif
 170
 171 #define NORMAL_VTABLE(E) \
 172  E ## isName2, \
 173  E ## isName3, \
 174  E ## isName4, \
 175  E ## isNmstrt2, \
 176  E ## isNmstrt3, \
 177  E ## isNmstrt4, \
 178  E ## isInvalid2, \
 179  E ## isInvalid3, \
 180  E ## isInvalid4
 181
 182 #define EMPTY_VTABLE(E) 0, 0, 0, 0, 0, 0, 0, 0, 0
 183
 184 static int checkCharRefNumber(int);
 185
 186 #include "xmltok_impl.h"
 187 #include "ascii.h"
 188
 189 #ifdef XML_MIN_SIZE
 190 #define sb_isNameMin isNever
 191 #define sb_isNmstrtMin isNever
 192 #endif
 193
 194 #ifdef XML_MIN_SIZE
 195 #define MINBPC(enc) ((enc)->minBytesPerChar)
 196 #else
 197 /* minimum bytes per character */
 198 #define MINBPC(enc) 1
 199 #endif
 200
 201 #define SB_BYTE_TYPE(enc, p) \
 202   (((struct normal_encoding *)(enc))->type[(unsigned char)*(p)])
 203
 204 #ifdef XML_MIN_SIZE
 205 static
 206 int sb_byteType(const ENCODING *enc, const char *p)
 207 {
 208   return SB_BYTE_TYPE(enc, p);
 209 }
 210 #define BYTE_TYPE(enc, p) \
 211  (((const struct normal_encoding *)(enc))->byteType(enc, p))
 212 #else
 213 #define BYTE_TYPE(enc, p) SB_BYTE_TYPE(enc, p)
 214 #endif
 215
 216 #ifdef XML_MIN_SIZE
 217 #define BYTE_TO_ASCII(enc, p) \
 218  (((const struct normal_encoding *)(enc))->byteToAscii(enc, p))
 219 static
 220 int sb_byteToAscii(const ENCODING *enc, const char *p)
 221 {
 222   return *p;
 223 }
 224 #else
 225 #define BYTE_TO_ASCII(enc, p) (*(p))
 226 #endif
 227
 228 #define IS_NAME_CHAR(enc, p, n) \
 229  (((const struct normal_encoding *)(enc))->isName ## n(enc, p))
 230 #define IS_NMSTRT_CHAR(enc, p, n) \
 231  (((const struct normal_encoding *)(enc))->isNmstrt ## n(enc, p))
 232 #define IS_INVALID_CHAR(enc, p, n) \
 233  (((const struct normal_encoding *)(enc))->isInvalid ## n(enc, p))
 234
 235 #ifdef XML_MIN_SIZE
 236 #define IS_NAME_CHAR_MINBPC(enc, p) \
 237  (((const struct normal_encoding *)(enc))->isNameMin(enc, p))
 238 #define IS_NMSTRT_CHAR_MINBPC(enc, p) \
 239  (((const struct normal_encoding *)(enc))->isNmstrtMin(enc, p))
 240 #else
 241 #define IS_NAME_CHAR_MINBPC(enc, p) (0)
 242 #define IS_NMSTRT_CHAR_MINBPC(enc, p) (0)
 243 #endif
 244
 245 #ifdef XML_MIN_SIZE
 246 #define CHAR_MATCHES(enc, p, c) \
 247  (((const struct normal_encoding *)(enc))->charMatches(enc, p, c))
 248 static
 249 int sb_charMatches(const ENCODING *enc, const char *p, int c)
 250 {
 251   return *p == c;
 252 }
 253 #else
 254 /* c is an ASCII character */
 255 #define CHAR_MATCHES(enc, p, c) (*(p) == c)
 256 #endif
 257
 258 #define PREFIX(ident) normal_ ## ident
 259 #include "xmltok_impl.c"
 260
 261 #undef MINBPC
 262 #undef BYTE_TYPE
 263 #undef BYTE_TO_ASCII
 264 #undef CHAR_MATCHES
 265 #undef IS_NAME_CHAR
 266 #undef IS_NAME_CHAR_MINBPC
 267 #undef IS_NMSTRT_CHAR
 268 #undef IS_NMSTRT_CHAR_MINBPC
 269 #undef IS_INVALID_CHAR
 270
 271 enum {  /* UTF8_cvalN is value of masked first byte of N byte sequence */
 272   UTF8_cval1 = 0x00,
 273   UTF8_cval2 = 0xc0,
 274   UTF8_cval3 = 0xe0,
 275   UTF8_cval4 = 0xf0
 276 };
 277
 278 static
 279 void utf8_toUtf8(const ENCODING *enc,
 280                  const char **fromP, const char *fromLim,
 281                  char **toP, const char *toLim)
 282 {
 283   char *to;
 284   const char *from;
 285   cmExpatUnused(enc);
 286   if (fromLim - *fromP > toLim - *toP) {
 287     /* Avoid copying partial characters. */
 288     for (fromLim = *fromP + (toLim - *toP); fromLim > *fromP; fromLim--)
 289       if (((unsigned char)fromLim[-1] & 0xc0) != 0x80)
 290         break;
 291   }
 292   for (to = *toP, from = *fromP; from != fromLim; from++, to++)
 293     *to = *from;
 294   *fromP = from;
 295   *toP = to;
 296 }
 297
 298 static
 299 void utf8_toUtf16(const ENCODING *enc,
 300                   const char **fromP, const char *fromLim,
 301                   unsigned short **toP, const unsigned short *toLim)
 302 {
 303   unsigned short *to = *toP;
 304   const char *from = *fromP;
 305   while (from != fromLim && to != toLim) {
 306     switch (((struct normal_encoding *)enc)->type[(unsigned char)*from]) {
 307     case BT_LEAD2:
 308       *to++ = ((from[0] & 0x1f) << 6) | (from[1] & 0x3f);
 309       from += 2;
 310       break;
 311     case BT_LEAD3:
 312       *to++ = ((from[0] & 0xf) << 12) | ((from[1] & 0x3f) << 6) | (from[2] & 0x3f);
 313       from += 3;
 314       break;
 315     case BT_LEAD4:
 316       {
 317         unsigned long n;
 318         if (to + 1 == toLim)
 319           break;
 320         n = ((from[0] & 0x7) << 18) | ((from[1] & 0x3f) << 12) | ((from[2] & 0x3f) << 6) | (from[3] & 0x3f);
 321         n -= 0x10000;
 322         to[0] = (unsigned short)((n >> 10) | 0xD800);
 323         to[1] = (unsigned short)((n & 0x3FF) | 0xDC00);
 324         to += 2;
 325         from += 4;
 326       }
 327       break;
 328     default:
 329       *to++ = *from++;
 330       break;
 331     }
 332   }
 333   *fromP = from;
 334   *toP = to;
 335 }
 336
 337 #ifdef XML_NS
 338 static const struct normal_encoding utf8_encoding_ns = {
 339   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 340   {
 341 #include "asciitab.h"
 342 #include "utf8tab.h"
 343   },
 344   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 345 };
 346 #endif
 347
 348 static const struct normal_encoding utf8_encoding = {
 349   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 350   {
 351 #define BT_COLON BT_NMSTRT
 352 #include "asciitab.h"
 353 #undef BT_COLON
 354 #include "utf8tab.h"
 355   },
 356   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 357 };
 358
 359 #ifdef XML_NS
 360
 361 static const struct normal_encoding internal_utf8_encoding_ns = {
 362   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 363   {
 364 #include "iasciitab.h"
 365 #include "utf8tab.h"
 366   },
 367   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 368 };
 369
 370 #endif
 371
 372 static const struct normal_encoding internal_utf8_encoding = {
 373   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 374   {
 375 #define BT_COLON BT_NMSTRT
 376 #include "iasciitab.h"
 377 #undef BT_COLON
 378 #include "utf8tab.h"
 379   },
 380   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 381 };
 382
 383 static
 384 void latin1_toUtf8(const ENCODING *enc,
 385                    const char **fromP, const char *fromLim,
 386                    char **toP, const char *toLim)
 387 {
 388   cmExpatUnused(enc);
 389   for (;;) {
 390     unsigned char c;
 391     if (*fromP == fromLim)
 392       break;
 393     c = (unsigned char)**fromP;
 394     if (c & 0x80) {
 395       if (toLim - *toP < 2)
 396         break;
 397       *(*toP)++ = ((c >> 6) | UTF8_cval2);
 398       *(*toP)++ = ((c & 0x3f) | 0x80);
 399       (*fromP)++;
 400     }
 401     else {
 402       if (*toP == toLim)
 403         break;
 404       *(*toP)++ = *(*fromP)++;
 405     }
 406   }
 407 }
 408
 409 static
 410 void latin1_toUtf16(const ENCODING *enc,
 411                     const char **fromP, const char *fromLim,
 412                     unsigned short **toP, const unsigned short *toLim)
 413 {
 414   cmExpatUnused(enc);
 415   while (*fromP != fromLim && *toP != toLim)
 416     *(*toP)++ = (unsigned char)*(*fromP)++;
 417 }
 418
 419 #ifdef XML_NS
 420
 421 static const struct normal_encoding latin1_encoding_ns = {
 422   { VTABLE1, latin1_toUtf8, latin1_toUtf16, 1, 0, 0 },
 423   {
 424 #include "asciitab.h"
 425 #include "latin1tab.h"
 426   },
 427   STANDARD_VTABLE(sb_) EMPTY_VTABLE(sb_)
 428 };
 429
 430 #endif
 431
 432 static const struct normal_encoding latin1_encoding = {
 433   { VTABLE1, latin1_toUtf8, latin1_toUtf16, 1, 0, 0 },
 434   {
 435 #define BT_COLON BT_NMSTRT
 436 #include "asciitab.h"
 437 #undef BT_COLON
 438 #include "latin1tab.h"
 439   },
 440   STANDARD_VTABLE(sb_) EMPTY_VTABLE(sb_)
 441 };
 442
 443 static
 444 void ascii_toUtf8(const ENCODING *enc,
 445                   const char **fromP, const char *fromLim,
 446                   char **toP, const char *toLim)
 447 {
 448   cmExpatUnused(enc);
 449   while (*fromP != fromLim && *toP != toLim)
 450     *(*toP)++ = *(*fromP)++;
 451 }
 452
 453 #ifdef XML_NS
 454
 455 static const struct normal_encoding ascii_encoding_ns = {
 456   { VTABLE1, ascii_toUtf8, latin1_toUtf16, 1, 1, 0 },
 457   {
 458 #include "asciitab.h"
 459 /* BT_NONXML == 0 */
 460   },
 461   STANDARD_VTABLE(sb_) EMPTY_VTABLE(sb_)
 462 };
 463
 464 #endif
 465
 466 static const struct normal_encoding ascii_encoding = {
 467   { VTABLE1, ascii_toUtf8, latin1_toUtf16, 1, 1, 0 },
 468   {
 469 #define BT_COLON BT_NMSTRT
 470 #include "asciitab.h"
 471 #undef BT_COLON
 472 /* BT_NONXML == 0 */
 473   },
 474   STANDARD_VTABLE(sb_) EMPTY_VTABLE(sb_)
 475 };
 476
 477 static int unicode_byte_type(char hi, char lo)
 478 {
 479   switch ((unsigned char)hi) {
 480   case 0xD8: case 0xD9: case 0xDA: case 0xDB:
 481     return BT_LEAD4;
 482   case 0xDC: case 0xDD: case 0xDE: case 0xDF:
 483     return BT_TRAIL;
 484   case 0xFF:
 485     switch ((unsigned char)lo) {
 486     case 0xFF:
 487     case 0xFE:
 488       return BT_NONXML;
 489     }
 490     break;
 491   }
 492   return BT_NONASCII;
 493 }
 494
 495 #define DEFINE_UTF16_TO_UTF8(E) \
 496 static \
 497 void E ## toUtf8(const ENCODING *enc, \
 498                  const char **fromP, const char *fromLim, \
 499                  char **toP, const char *toLim) \
 500 { \
 501   const char *from; \
 502   cmExpatUnused(enc);\
 503   for (from = *fromP; from != fromLim; from += 2) { \
 504     int plane; \
 505     unsigned char lo2; \
 506     unsigned char lo = GET_LO(from); \
 507     unsigned char hi = GET_HI(from); \
 508     switch (hi) { \
 509     case 0: \
 510       if (lo < 0x80) { \
 511         if (*toP == toLim) { \
 512           *fromP = from; \
 513           return; \
 514         } \
 515         *(*toP)++ = lo; \
 516         break; \
 517       } \
 518       /* fall through */ \
 519     case 0x1: case 0x2: case 0x3: \
 520     case 0x4: case 0x5: case 0x6: case 0x7: \
 521       if (toLim -  *toP < 2) { \
 522         *fromP = from; \
 523         return; \
 524       } \
 525       *(*toP)++ = ((lo >> 6) | (hi << 2) |  UTF8_cval2); \
 526       *(*toP)++ = ((lo & 0x3f) | 0x80); \
 527       break; \
 528     default: \
 529       if (toLim -  *toP < 3)  { \
 530         *fromP = from; \
 531         return; \
 532       } \
 533       /* 16 bits divided 4, 6, 6 amongst 3 bytes */ \
 534       *(*toP)++ = ((hi >> 4) | UTF8_cval3); \
 535       *(*toP)++ = (((hi & 0xf) << 2) | (lo >> 6) | 0x80); \
 536       *(*toP)++ = ((lo & 0x3f) | 0x80); \
 537       break; \
 538     case 0xD8: case 0xD9: case 0xDA: case 0xDB: \
 539       if (toLim -  *toP < 4) { \
 540         *fromP = from; \
 541         return; \
 542       } \
 543       plane = (((hi & 0x3) << 2) | ((lo >> 6) & 0x3)) + 1; \
 544       *(*toP)++ = ((plane >> 2) | UTF8_cval4); \
 545       *(*toP)++ = (((lo >> 2) & 0xF) | ((plane & 0x3) << 4) | 0x80); \
 546       from += 2; \
 547       lo2 = GET_LO(from); \
 548       *(*toP)++ = (((lo & 0x3) << 4) \
 549                    | ((GET_HI(from) & 0x3) << 2) \
 550                    | (lo2 >> 6) \
 551                    | 0x80); \
 552       *(*toP)++ = ((lo2 & 0x3f) | 0x80); \
 553       break; \
 554     } \
 555   } \
 556   *fromP = from; \
 557 }
 558
 559 #define DEFINE_UTF16_TO_UTF16(E) \
 560 static \
 561 void E ## toUtf16(const ENCODING *enc, \
 562                   const char **fromP, const char *fromLim, \
 563                   unsigned short **toP, const unsigned short *toLim) \
 564 { \
 565   cmExpatUnused(enc);\
 566   /* Avoid copying first half only of surrogate */ \
 567   if (fromLim - *fromP > ((toLim - *toP) << 1) \
 568       && (GET_HI(fromLim - 2) & 0xF8) == 0xD8) \
 569     fromLim -= 2; \
 570   for (; *fromP != fromLim && *toP != toLim; *fromP += 2) \
 571     *(*toP)++ = (GET_HI(*fromP) << 8) | GET_LO(*fromP); \
 572 }
 573
 574 #define SET2(ptr, ch) \
 575   (((ptr)[0] = ((ch) & 0xff)), ((ptr)[1] = ((ch) >> 8)))
 576 #define GET_LO(ptr) ((unsigned char)(ptr)[0])
 577 #define GET_HI(ptr) ((unsigned char)(ptr)[1])
 578
 579 DEFINE_UTF16_TO_UTF8(little2_)
 580 DEFINE_UTF16_TO_UTF16(little2_)
 581
 582 #undef SET2
 583 #undef GET_LO
 584 #undef GET_HI
 585
 586 #define SET2(ptr, ch) \
 587   (((ptr)[0] = ((ch) >> 8)), ((ptr)[1] = ((ch) & 0xFF)))
 588 #define GET_LO(ptr) ((unsigned char)(ptr)[1])
 589 #define GET_HI(ptr) ((unsigned char)(ptr)[0])
 590
 591 DEFINE_UTF16_TO_UTF8(big2_)
 592 DEFINE_UTF16_TO_UTF16(big2_)
 593
 594 #undef SET2
 595 #undef GET_LO
 596 #undef GET_HI
 597
 598 #define LITTLE2_BYTE_TYPE(enc, p) \
 599  ((p)[1] == 0 \
 600   ? ((struct normal_encoding *)(enc))->type[(unsigned char)*(p)] \
 601   : unicode_byte_type((p)[1], (p)[0]))
 602 #define LITTLE2_BYTE_TO_ASCII(enc, p) ((p)[1] == 0 ? (p)[0] : -1)
 603 #define LITTLE2_CHAR_MATCHES(enc, p, c) ((p)[1] == 0 && (p)[0] == c)
 604 #define LITTLE2_IS_NAME_CHAR_MINBPC(enc, p) \
 605   UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0])
 606 #define LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p) \
 607   UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0])
 608
 609 #ifdef XML_MIN_SIZE
 610
 611 static
 612 int little2_byteType(const ENCODING *enc, const char *p)
 613 {
 614   return LITTLE2_BYTE_TYPE(enc, p);
 615 }
 616
 617 static
 618 int little2_byteToAscii(const ENCODING *enc, const char *p)
 619 {
 620   return LITTLE2_BYTE_TO_ASCII(enc, p);
 621 }
 622
 623 static
 624 int little2_charMatches(const ENCODING *enc, const char *p, int c)
 625 {
 626   return LITTLE2_CHAR_MATCHES(enc, p, c);
 627 }
 628
 629 static
 630 int little2_isNameMin(const ENCODING *enc, const char *p)
 631 {
 632   return LITTLE2_IS_NAME_CHAR_MINBPC(enc, p);
 633 }
 634
 635 static
 636 int little2_isNmstrtMin(const ENCODING *enc, const char *p)
 637 {
 638   return LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p);
 639 }
 640
 641 #undef VTABLE
 642 #define VTABLE VTABLE1, little2_toUtf8, little2_toUtf16
 643
 644 #else /* not XML_MIN_SIZE */
 645
 646 #undef PREFIX
 647 #define PREFIX(ident) little2_ ## ident
 648 #define MINBPC(enc) 2
 649 /* CHAR_MATCHES is guaranteed to have MINBPC bytes available. */
 650 #define BYTE_TYPE(enc, p) LITTLE2_BYTE_TYPE(enc, p)
 651 #define BYTE_TO_ASCII(enc, p) LITTLE2_BYTE_TO_ASCII(enc, p)
 652 #define CHAR_MATCHES(enc, p, c) LITTLE2_CHAR_MATCHES(enc, p, c)
 653 #define IS_NAME_CHAR(enc, p, n) 0
 654 #define IS_NAME_CHAR_MINBPC(enc, p) LITTLE2_IS_NAME_CHAR_MINBPC(enc, p)
 655 #define IS_NMSTRT_CHAR(enc, p, n) (0)
 656 #define IS_NMSTRT_CHAR_MINBPC(enc, p) LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p)
 657
 658 #include "xmltok_impl.c"
 659
 660 #undef MINBPC
 661 #undef BYTE_TYPE
 662 #undef BYTE_TO_ASCII
 663 #undef CHAR_MATCHES
 664 #undef IS_NAME_CHAR
 665 #undef IS_NAME_CHAR_MINBPC
 666 #undef IS_NMSTRT_CHAR
 667 #undef IS_NMSTRT_CHAR_MINBPC
 668 #undef IS_INVALID_CHAR
 669
 670 #endif /* not XML_MIN_SIZE */
 671
 672 #ifdef XML_NS
 673
 674 static const struct normal_encoding little2_encoding_ns = {
 675   { VTABLE, 2, 0,
 676 #if XML_BYTE_ORDER == 12
 677     1
 678 #else
 679     0
 680 #endif
 681   },
 682   {
 683 #include "asciitab.h"
 684 #include "latin1tab.h"
 685   },
 686   STANDARD_VTABLE(little2_) EMPTY_VTABLE(little2_)
 687 };
 688
 689 #endif
 690
 691 static const struct normal_encoding little2_encoding = {
 692   { VTABLE, 2, 0,
 693 #if XML_BYTE_ORDER == 12
 694     1
 695 #else
 696     0
 697 #endif
 698   },
 699   {
 700 #define BT_COLON BT_NMSTRT
 701 #include "asciitab.h"
 702 #undef BT_COLON
 703 #include "latin1tab.h"
 704   },
 705   STANDARD_VTABLE(little2_) EMPTY_VTABLE(little2_)
 706 };
 707
 708 #if XML_BYTE_ORDER != 21
 709
 710 #ifdef XML_NS
 711
 712 static const struct normal_encoding internal_little2_encoding_ns = {
 713   { VTABLE, 2, 0, 1 },
 714   {
 715 #include "iasciitab.h"
 716 #include "latin1tab.h"
 717   },
 718   STANDARD_VTABLE(little2_) EMPTY_VTABLE(little2_)
 719 };
 720
 721 #endif
 722
 723 static const struct normal_encoding internal_little2_encoding = {
 724   { VTABLE, 2, 0, 1 },
 725   {
 726 #define BT_COLON BT_NMSTRT
 727 #include "iasciitab.h"
 728 #undef BT_COLON
 729 #include "latin1tab.h"
 730   },
 731   STANDARD_VTABLE(little2_) EMPTY_VTABLE(little2_)
 732 };
 733
 734 #endif
 735
 736
 737 #define BIG2_BYTE_TYPE(enc, p) \
 738  ((p)[0] == 0 \
 739   ? ((struct normal_encoding *)(enc))->type[(unsigned char)(p)[1]] \
 740   : unicode_byte_type((p)[0], (p)[1]))
 741 #define BIG2_BYTE_TO_ASCII(enc, p) ((p)[0] == 0 ? (p)[1] : -1)
 742 #define BIG2_CHAR_MATCHES(enc, p, c) ((p)[0] == 0 && (p)[1] == c)
 743 #define BIG2_IS_NAME_CHAR_MINBPC(enc, p) \
 744   UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1])
 745 #define BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p) \
 746   UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1])
 747
 748 #ifdef XML_MIN_SIZE
 749
 750 static
 751 int big2_byteType(const ENCODING *enc, const char *p)
 752 {
 753   return BIG2_BYTE_TYPE(enc, p);
 754 }
 755
 756 static
 757 int big2_byteToAscii(const ENCODING *enc, const char *p)
 758 {
 759   return BIG2_BYTE_TO_ASCII(enc, p);
 760 }
 761
 762 static
 763 int big2_charMatches(const ENCODING *enc, const char *p, int c)
 764 {
 765   return BIG2_CHAR_MATCHES(enc, p, c);
 766 }
 767
 768 static
 769 int big2_isNameMin(const ENCODING *enc, const char *p)
 770 {
 771   return BIG2_IS_NAME_CHAR_MINBPC(enc, p);
 772 }
 773
 774 static
 775 int big2_isNmstrtMin(const ENCODING *enc, const char *p)
 776 {
 777   return BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p);
 778 }
 779
 780 #undef VTABLE
 781 #define VTABLE VTABLE1, big2_toUtf8, big2_toUtf16
 782
 783 #else /* not XML_MIN_SIZE */
 784
 785 #undef PREFIX
 786 #define PREFIX(ident) big2_ ## ident
 787 #define MINBPC(enc) 2
 788 /* CHAR_MATCHES is guaranteed to have MINBPC bytes available. */
 789 #define BYTE_TYPE(enc, p) BIG2_BYTE_TYPE(enc, p)
 790 #define BYTE_TO_ASCII(enc, p) BIG2_BYTE_TO_ASCII(enc, p)
 791 #define CHAR_MATCHES(enc, p, c) BIG2_CHAR_MATCHES(enc, p, c)
 792 #define IS_NAME_CHAR(enc, p, n) 0
 793 #define IS_NAME_CHAR_MINBPC(enc, p) BIG2_IS_NAME_CHAR_MINBPC(enc, p)
 794 #define IS_NMSTRT_CHAR(enc, p, n) (0)
 795 #define IS_NMSTRT_CHAR_MINBPC(enc, p) BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p)
 796
 797 #include "xmltok_impl.c"
 798
 799 #undef MINBPC
 800 #undef BYTE_TYPE
 801 #undef BYTE_TO_ASCII
 802 #undef CHAR_MATCHES
 803 #undef IS_NAME_CHAR
 804 #undef IS_NAME_CHAR_MINBPC
 805 #undef IS_NMSTRT_CHAR
 806 #undef IS_NMSTRT_CHAR_MINBPC
 807 #undef IS_INVALID_CHAR
 808
 809 #endif /* not XML_MIN_SIZE */
 810
 811 #ifdef XML_NS
 812
 813 static const struct normal_encoding big2_encoding_ns = {
 814   { VTABLE, 2, 0,
 815 #if XML_BYTE_ORDER == 21
 816   1
 817 #else
 818   0
 819 #endif
 820   },
 821   {
 822 #include "asciitab.h"
 823 #include "latin1tab.h"
 824   },
 825   STANDARD_VTABLE(big2_) EMPTY_VTABLE(big2_)
 826 };
 827
 828 #endif
 829
 830 static const struct normal_encoding big2_encoding = {
 831   { VTABLE, 2, 0,
 832 #if XML_BYTE_ORDER == 21
 833   1
 834 #else
 835   0
 836 #endif
 837   },
 838   {
 839 #define BT_COLON BT_NMSTRT
 840 #include "asciitab.h"
 841 #undef BT_COLON
 842 #include "latin1tab.h"
 843   },
 844   STANDARD_VTABLE(big2_) EMPTY_VTABLE(big2_)
 845 };
 846
 847 #if XML_BYTE_ORDER != 12
 848
 849 #ifdef XML_NS
 850
 851 static const struct normal_encoding internal_big2_encoding_ns = {
 852   { VTABLE, 2, 0, 1 },
 853   {
 854 #include "iasciitab.h"
 855 #include "latin1tab.h"
 856   },
 857   STANDARD_VTABLE(big2_) EMPTY_VTABLE(big2_)
 858 };
 859
 860 #endif
 861
 862 static const struct normal_encoding internal_big2_encoding = {
 863   { VTABLE, 2, 0, 1 },
 864   {
 865 #define BT_COLON BT_NMSTRT
 866 #include "iasciitab.h"
 867 #undef BT_COLON
 868 #include "latin1tab.h"
 869   },
 870   STANDARD_VTABLE(big2_) EMPTY_VTABLE(big2_)
 871 };
 872
 873 #endif
 874
 875 #undef PREFIX
 876
 877 static
 878 int streqci(const char *s1, const char *s2)
 879 {
 880   for (;;) {
 881     char c1 = *s1++;
 882     char c2 = *s2++;
 883     if (ASCII_a <= c1 && c1 <= ASCII_z)
 884       c1 += ASCII_A - ASCII_a;
 885     if (ASCII_a <= c2 && c2 <= ASCII_z)
 886       c2 += ASCII_A - ASCII_a;
 887     if (c1 != c2)
 888       return 0;
 889     if (!c1)
 890       break;
 891   }
 892   return 1;
 893 }
 894
 895 static
 896 void initUpdatePosition(const ENCODING *enc, const char *ptr,
 897                         const char *end, POSITION *pos)
 898 {
 899   cmExpatUnused(enc);
 900   normal_updatePosition(&utf8_encoding.enc, ptr, end, pos);
 901 }
 902
 903 static
 904 int toAscii(const ENCODING *enc, const char *ptr, const char *end)
 905 {
 906   char buf[1];
 907   char *p = buf;
 908   XmlUtf8Convert(enc, &ptr, end, &p, p + 1);
 909   if (p == buf)
 910     return -1;
 911   else
 912     return buf[0];
 913 }
 914
 915 static
 916 int isSpace(int c)
 917 {
 918   switch (c) {
 919   case 0x20:
 920   case 0xD:
 921   case 0xA:
 922   case 0x9:
 923     return 1;
 924   }
 925   return 0;
 926 }
 927
 928 /* Return 1 if there's just optional white space
 929 or there's an S followed by name=val. */
 930 static
 931 int parsePseudoAttribute(const ENCODING *enc,
 932                          const char *ptr,
 933                          const char *end,
 934                          const char **namePtr,
 935                          const char **nameEndPtr,
 936                          const char **valPtr,
 937                          const char **nextTokPtr)
 938 {
 939   int c;
 940   char open;
 941   if (ptr == end) {
 942     *namePtr = 0;
 943     return 1;
 944   }
 945   if (!isSpace(toAscii(enc, ptr, end))) {
 946     *nextTokPtr = ptr;
 947     return 0;
 948   }
 949   do {
 950     ptr += enc->minBytesPerChar;
 951   } while (isSpace(toAscii(enc, ptr, end)));
 952   if (ptr == end) {
 953     *namePtr = 0;
 954     return 1;
 955   }
 956   *namePtr = ptr;
 957   for (;;) {
 958     c = toAscii(enc, ptr, end);
 959     if (c == -1) {
 960       *nextTokPtr = ptr;
 961       return 0;
 962     }
 963     if (c == ASCII_EQUALS) {
 964       *nameEndPtr = ptr;
 965       break;
 966     }
 967     if (isSpace(c)) {
 968       *nameEndPtr = ptr;
 969       do {
 970         ptr += enc->minBytesPerChar;
 971       } while (isSpace(c = toAscii(enc, ptr, end)));
 972       if (c != ASCII_EQUALS) {
 973         *nextTokPtr = ptr;
 974         return 0;
 975       }
 976       break;
 977     }
 978     ptr += enc->minBytesPerChar;
 979   }
 980   if (ptr == *namePtr) {
 981     *nextTokPtr = ptr;
 982     return 0;
 983   }
 984   ptr += enc->minBytesPerChar;
 985   c = toAscii(enc, ptr, end);
 986   while (isSpace(c)) {
 987     ptr += enc->minBytesPerChar;
 988     c = toAscii(enc, ptr, end);
 989   }
 990   if (c != ASCII_QUOT && c != ASCII_APOS) {
 991     *nextTokPtr = ptr;
 992     return 0;
 993   }
 994   open = c;
 995   ptr += enc->minBytesPerChar;
 996   *valPtr = ptr;
 997   for (;; ptr += enc->minBytesPerChar) {
 998     c = toAscii(enc, ptr, end);
 999     if (c == open)
1000       break;
1001     if (!(ASCII_a <= c && c <= ASCII_z)
1002         && !(ASCII_A <= c && c <= ASCII_Z)
1003         && !(ASCII_0 <= c && c <= ASCII_9)
1004         && c != ASCII_PERIOD
1005         && c != ASCII_MINUS
1006         && c != ASCII_UNDERSCORE) {
1007       *nextTokPtr = ptr;
1008       return 0;
1009     }
1010   }
1011   *nextTokPtr = ptr + enc->minBytesPerChar;
1012   return 1;
1013 }
1014
1015 static const char KW_version[] = {
1016   ASCII_v, ASCII_e, ASCII_r, ASCII_s, ASCII_i, ASCII_o, ASCII_n, '\0'
1017 };
1018
1019 static const char KW_encoding[] = {
1020   ASCII_e, ASCII_n, ASCII_c, ASCII_o, ASCII_d, ASCII_i, ASCII_n, ASCII_g, '\0'
1021 };
1022
1023 static const char KW_standalone[] = {
1024   ASCII_s, ASCII_t, ASCII_a, ASCII_n, ASCII_d, ASCII_a, ASCII_l, ASCII_o, ASCII_n, ASCII_e, '\0'
1025 };
1026
1027 static const char KW_yes[] = {
1028   ASCII_y, ASCII_e, ASCII_s,  '\0'
1029 };
1030
1031 static const char KW_no[] = {
1032   ASCII_n, ASCII_o,  '\0'
1033 };
1034
1035 static
1036 int doParseXmlDecl(const ENCODING *(*encodingFinder)(const ENCODING *,
1037                                                      const char *,
1038                                                      const char *),
1039                    int isGeneralTextEntity,
1040                    const ENCODING *enc,
1041                    const char *ptr,
1042                    const char *end,
1043                    const char **badPtr,
1044                    const char **versionPtr,
1045                    const char **versionEndPtr,
1046                    const char **encodingName,
1047                    const ENCODING **encoding,
1048                    int *standalone)
1049 {
1050   const char *val = 0;
1051   const char *name = 0;
1052   const char *nameEnd = 0;
1053   ptr += 5 * enc->minBytesPerChar;
1054   end -= 2 * enc->minBytesPerChar;
1055   if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr) || !name) {
1056     *badPtr = ptr;
1057     return 0;
1058   }
1059   if (!XmlNameMatchesAscii(enc, name, nameEnd, KW_version)) {
1060     if (!isGeneralTextEntity) {
1061       *badPtr = name;
1062       return 0;
1063     }
1064   }
1065   else {
1066     if (versionPtr)
1067       *versionPtr = val;
1068     if (versionEndPtr)
1069       *versionEndPtr = ptr;
1070     if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)) {
1071       *badPtr = ptr;
1072       return 0;
1073     }
1074     if (!name) {
1075       if (isGeneralTextEntity) {
1076         /* a TextDecl must have an EncodingDecl */
1077         *badPtr = ptr;
1078         return 0;
1079       }
1080       return 1;
1081     }
1082   }
1083   if (XmlNameMatchesAscii(enc, name, nameEnd, KW_encoding)) {
1084     int c = toAscii(enc, val, end);
1085     if (!(ASCII_a <= c && c <= ASCII_z) && !(ASCII_A <= c && c <= ASCII_Z)) {
1086       *badPtr = val;
1087       return 0;
1088     }
1089     if (encodingName)
1090       *encodingName = val;
1091     if (encoding)
1092       *encoding = encodingFinder(enc, val, ptr - enc->minBytesPerChar);
1093     if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)) {
1094       *badPtr = ptr;
1095       return 0;
1096     }
1097     if (!name)
1098       return 1;
1099   }
1100   if (!XmlNameMatchesAscii(enc, name, nameEnd, KW_standalone) || isGeneralTextEntity) {
1101     *badPtr = name;
1102     return 0;
1103   }
1104   if (XmlNameMatchesAscii(enc, val, ptr - enc->minBytesPerChar, KW_yes)) {
1105     if (standalone)
1106       *standalone = 1;
1107   }
1108   else if (XmlNameMatchesAscii(enc, val, ptr - enc->minBytesPerChar, KW_no)) {
1109     if (standalone)
1110       *standalone = 0;
1111   }
1112   else {
1113     *badPtr = val;
1114     return 0;
1115   }
1116   while (isSpace(toAscii(enc, ptr, end)))
1117     ptr += enc->minBytesPerChar;
1118   if (ptr != end) {
1119     *badPtr = ptr;
1120     return 0;
1121   }
1122   return 1;
1123 }
1124
1125 static
1126 int checkCharRefNumber(int result)
1127 {
1128   switch (result >> 8) {
1129   case 0xD8: case 0xD9: case 0xDA: case 0xDB:
1130   case 0xDC: case 0xDD: case 0xDE: case 0xDF:
1131     return -1;
1132   case 0:
1133     if (latin1_encoding.type[result] == BT_NONXML)
1134       return -1;
1135     break;
1136   case 0xFF:
1137     if (result == 0xFFFE || result == 0xFFFF)
1138       return -1;
1139     break;
1140   }
1141   return result;
1142 }
1143
1144 int XmlUtf8Encode(int c, char *buf)
1145 {
1146   enum {
1147     /* minN is minimum legal resulting value for N byte sequence */
1148     min2 = 0x80,
1149     min3 = 0x800,
1150     min4 = 0x10000
1151   };
1152
1153   if (c < 0)
1154     return 0;
1155   if (c < min2) {
1156     buf[0] = (c | UTF8_cval1);
1157     return 1;
1158   }
1159   if (c < min3) {
1160     buf[0] = ((c >> 6) | UTF8_cval2);
1161     buf[1] = ((c & 0x3f) | 0x80);
1162     return 2;
1163   }
1164   if (c < min4) {
1165     buf[0] = ((c >> 12) | UTF8_cval3);
1166     buf[1] = (((c >> 6) & 0x3f) | 0x80);
1167     buf[2] = ((c & 0x3f) | 0x80);
1168     return 3;
1169   }
1170   if (c < 0x110000) {
1171     buf[0] = ((c >> 18) | UTF8_cval4);
1172     buf[1] = (((c >> 12) & 0x3f) | 0x80);
1173     buf[2] = (((c >> 6) & 0x3f) | 0x80);
1174     buf[3] = ((c & 0x3f) | 0x80);
1175     return 4;
1176   }
1177   return 0;
1178 }
1179
1180 int XmlUtf16Encode(int charNum, unsigned short *buf)
1181 {
1182   if (charNum < 0)
1183     return 0;
1184   if (charNum < 0x10000) {
1185     buf[0] = charNum;
1186     return 1;
1187   }
1188   if (charNum < 0x110000) {
1189     charNum -= 0x10000;
1190     buf[0] = (charNum >> 10) + 0xD800;
1191     buf[1] = (charNum & 0x3FF) + 0xDC00;
1192     return 2;
1193   }
1194   return 0;
1195 }
1196
1197 struct unknown_encoding {
1198   struct normal_encoding normal;
1199   int (*convert)(void *userData, const char *p);
1200   void *userData;
1201   unsigned short utf16[256];
1202   char utf8[256][4];
1203 };
1204
1205 int XmlSizeOfUnknownEncoding(void)
1206 {
1207   return sizeof(struct unknown_encoding);
1208 }
1209
1210 static
1211 int unknown_isName(const ENCODING *enc, const char *p)
1212 {
1213   int c = ((const struct unknown_encoding *)enc)
1214           ->convert(((const struct unknown_encoding *)enc)->userData, p);
1215   if (c & ~0xFFFF)
1216     return 0;
1217   return UCS2_GET_NAMING(namePages, c >> 8, c & 0xFF);
1218 }
1219
1220 static
1221 int unknown_isNmstrt(const ENCODING *enc, const char *p)
1222 {
1223   int c = ((const struct unknown_encoding *)enc)
1224           ->convert(((const struct unknown_encoding *)enc)->userData, p);
1225   if (c & ~0xFFFF)
1226     return 0;
1227   return UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xFF);
1228 }
1229
1230 static
1231 int unknown_isInvalid(const ENCODING *enc, const char *p)
1232 {
1233   int c = ((const struct unknown_encoding *)enc)
1234            ->convert(((const struct unknown_encoding *)enc)->userData, p);
1235   return (c & ~0xFFFF) || checkCharRefNumber(c) < 0;
1236 }
1237
1238 static
1239 void unknown_toUtf8(const ENCODING *enc,
1240                     const char **fromP, const char *fromLim,
1241                     char **toP, const char *toLim)
1242 {
1243   char buf[XML_UTF8_ENCODE_MAX];
1244   for (;;) {
1245     const char *utf8;
1246     int n;
1247     if (*fromP == fromLim)
1248       break;
1249     utf8 = ((const struct unknown_encoding *)enc)->utf8[(unsigned char)**fromP];
1250     n = *utf8++;
1251     if (n == 0) {
1252       int c = ((const struct unknown_encoding *)enc)
1253               ->convert(((const struct unknown_encoding *)enc)->userData, *fromP);
1254       n = XmlUtf8Encode(c, buf);
1255       if (n > toLim - *toP)
1256         break;
1257       utf8 = buf;
1258       *fromP += ((const struct normal_encoding *)enc)->type[(unsigned char)**fromP]
1259                  - (BT_LEAD2 - 2);
1260     }
1261     else {
1262       if (n > toLim - *toP)
1263         break;
1264       (*fromP)++;
1265     }
1266     do {
1267       *(*toP)++ = *utf8++;
1268     } while (--n != 0);
1269   }
1270 }
1271
1272 static
1273 void unknown_toUtf16(const ENCODING *enc,
1274                      const char **fromP, const char *fromLim,
1275                      unsigned short **toP, const unsigned short *toLim)
1276 {
1277   while (*fromP != fromLim && *toP != toLim) {
1278     unsigned short c
1279       = ((const struct unknown_encoding *)enc)->utf16[(unsigned char)**fromP];
1280     if (c == 0) {
1281       c = (unsigned short)((const struct unknown_encoding *)enc)
1282            ->convert(((const struct unknown_encoding *)enc)->userData, *fromP);
1283       *fromP += ((const struct normal_encoding *)enc)->type[(unsigned char)**fromP]
1284                  - (BT_LEAD2 - 2);
1285     }
1286     else
1287       (*fromP)++;
1288     *(*toP)++ = c;
1289   }
1290 }
1291
1292 ENCODING *
1293 XmlInitUnknownEncoding(void *mem,
1294                        int *table,
1295                        int (*convert)(void *userData, const char *p),
1296                        void *userData)
1297 {
1298   int i;
1299   struct unknown_encoding *e = mem;
1300   for (i = 0; i < (int)sizeof(struct normal_encoding); i++)
1301     ((char *)mem)[i] = ((char *)&latin1_encoding)[i];
1302   for (i = 0; i < 128; i++)
1303     if (latin1_encoding.type[i] != BT_OTHER
1304         && latin1_encoding.type[i] != BT_NONXML
1305         && table[i] != i)
1306       return 0;
1307   for (i = 0; i < 256; i++) {
1308     int c = table[i];
1309     if (c == -1) {
1310       e->normal.type[i] = BT_MALFORM;
1311       /* This shouldn't really get used. */
1312       e->utf16[i] = 0xFFFF;
1313       e->utf8[i][0] = 1;
1314       e->utf8[i][1] = 0;
1315     }
1316     else if (c < 0) {
1317       if (c < -4)
1318         return 0;
1319       e->normal.type[i] = BT_LEAD2 - (c + 2);
1320       e->utf8[i][0] = 0;
1321       e->utf16[i] = 0;
1322     }
1323     else if (c < 0x80) {
1324       if (latin1_encoding.type[c] != BT_OTHER
1325           && latin1_encoding.type[c] != BT_NONXML
1326           && c != i)
1327         return 0;
1328       e->normal.type[i] = latin1_encoding.type[c];
1329       e->utf8[i][0] = 1;
1330       e->utf8[i][1] = (char)c;
1331       e->utf16[i] = c == 0 ? 0xFFFF : c;
1332     }
1333     else if (checkCharRefNumber(c) < 0) {
1334       e->normal.type[i] = BT_NONXML;
1335       /* This shouldn't really get used. */
1336       e->utf16[i] = 0xFFFF;
1337       e->utf8[i][0] = 1;
1338       e->utf8[i][1] = 0;
1339     }
1340     else {
1341       if (c > 0xFFFF)
1342         return 0;
1343       if (UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xff))
1344         e->normal.type[i] = BT_NMSTRT;
1345       else if (UCS2_GET_NAMING(namePages, c >> 8, c & 0xff))
1346         e->normal.type[i] = BT_NAME;
1347       else
1348         e->normal.type[i] = BT_OTHER;
1349       e->utf8[i][0] = (char)XmlUtf8Encode(c, e->utf8[i] + 1);
1350       e->utf16[i] = c;
1351     }
1352   }
1353   e->userData = userData;
1354   e->convert = convert;
1355   if (convert) {
1356     e->normal.isName2 = unknown_isName;
1357     e->normal.isName3 = unknown_isName;
1358     e->normal.isName4 = unknown_isName;
1359     e->normal.isNmstrt2 = unknown_isNmstrt;
1360     e->normal.isNmstrt3 = unknown_isNmstrt;
1361     e->normal.isNmstrt4 = unknown_isNmstrt;
1362     e->normal.isInvalid2 = unknown_isInvalid;
1363     e->normal.isInvalid3 = unknown_isInvalid;
1364     e->normal.isInvalid4 = unknown_isInvalid;
1365   }
1366   e->normal.enc.utf8Convert = unknown_toUtf8;
1367   e->normal.enc.utf16Convert = unknown_toUtf16;
1368   return &(e->normal.enc);
1369 }
1370
1371 /* If this enumeration is changed, getEncodingIndex and encodings
1372 must also be changed. */
1373 enum {
1374   UNKNOWN_ENC = -1,
1375   ISO_8859_1_ENC = 0,
1376   US_ASCII_ENC,
1377   UTF_8_ENC,
1378   UTF_16_ENC,
1379   UTF_16BE_ENC,
1380   UTF_16LE_ENC,
1381   /* must match encodingNames up to here */
1382   NO_ENC
1383 };
1384
1385 static const char KW_ISO_8859_1[] = {
1386   ASCII_I, ASCII_S, ASCII_O, ASCII_MINUS, ASCII_8, ASCII_8, ASCII_5, ASCII_9, ASCII_MINUS, ASCII_1, '\0'
1387 };
1388 static const char KW_US_ASCII[] = {
1389   ASCII_U, ASCII_S, ASCII_MINUS, ASCII_A, ASCII_S, ASCII_C, ASCII_I, ASCII_I, '\0'
1390 };
1391 static const char KW_UTF_8[] =  {
1392   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_8, '\0'
1393 };
1394 static const char KW_UTF_16[] = {
1395   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, '\0'
1396 };
1397 static const char KW_UTF_16BE[] = {
1398   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, ASCII_B, ASCII_E, '\0'
1399 };
1400 static const char KW_UTF_16LE[] = {
1401   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, ASCII_L, ASCII_E, '\0'
1402 };
1403
1404 static
1405 int getEncodingIndex(const char *name)
1406 {
1407   static const char *encodingNames[] = {
1408     KW_ISO_8859_1,
1409     KW_US_ASCII,
1410     KW_UTF_8,
1411     KW_UTF_16,
1412     KW_UTF_16BE,
1413     KW_UTF_16LE,
1414   };
1415   int i;
1416   if (name == 0)
1417     return NO_ENC;
1418   for (i = 0; i < (int)(sizeof(encodingNames)/sizeof(encodingNames[0])); i++)
1419     if (streqci(name, encodingNames[i]))
1420       return i;
1421   return UNKNOWN_ENC;
1422 }
1423
1424 /* For binary compatibility, we store the index of the encoding specified
1425 at initialization in the isUtf16 member. */
1426
1427 #define INIT_ENC_INDEX(enc) ((int)(enc)->initEnc.isUtf16)
1428 #define SET_INIT_ENC_INDEX(enc, i) ((enc)->initEnc.isUtf16 = (char)i)
1429
1430 /* This is what detects the encoding.
1431 encodingTable maps from encoding indices to encodings;
1432 INIT_ENC_INDEX(enc) is the index of the external (protocol) specified encoding;
1433 state is XML_CONTENT_STATE if we're parsing an external text entity,
1434 and XML_PROLOG_STATE otherwise.
1435 */
1436
1437
1438 static
1439 int initScan(const ENCODING **encodingTable,
1440              const INIT_ENCODING *enc,
1441              int state,
1442              const char *ptr,
1443              const char *end,
1444              const char **nextTokPtr)
1445 {
1446   const ENCODING **encPtr;
1447
1448   if (ptr == end)
1449     return XML_TOK_NONE;
1450   encPtr = enc->encPtr;
1451   if (ptr + 1 == end) {
1452     /* only a single byte available for auto-detection */
1453 #ifndef XML_DTD /* FIXME */
1454     /* a well-formed document entity must have more than one byte */
1455     if (state != XML_CONTENT_STATE)
1456       return XML_TOK_PARTIAL;
1457 #endif
1458     /* so we're parsing an external text entity... */
1459     /* if UTF-16 was externally specified, then we need at least 2 bytes */
1460     switch (INIT_ENC_INDEX(enc)) {
1461     case UTF_16_ENC:
1462     case UTF_16LE_ENC:
1463     case UTF_16BE_ENC:
1464       return XML_TOK_PARTIAL;
1465     }
1466     switch ((unsigned char)*ptr) {
1467     case 0xFE:
1468     case 0xFF:
1469     case 0xEF: /* possibly first byte of UTF-8 BOM */
1470       if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC
1471           && state == XML_CONTENT_STATE)
1472         break;
1473       /* fall through */
1474     case 0x00:
1475     case 0x3C:
1476       return XML_TOK_PARTIAL;
1477     }
1478   }
1479   else {
1480     switch (((unsigned char)ptr[0] << 8) | (unsigned char)ptr[1]) {
1481     case 0xFEFF:
1482       if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC
1483           && state == XML_CONTENT_STATE)
1484         break;
1485       *nextTokPtr = ptr + 2;
1486       *encPtr = encodingTable[UTF_16BE_ENC];
1487       return XML_TOK_BOM;
1488     /* 00 3C is handled in the default case */
1489     case 0x3C00:
1490       if ((INIT_ENC_INDEX(enc) == UTF_16BE_ENC
1491            || INIT_ENC_INDEX(enc) == UTF_16_ENC)
1492           && state == XML_CONTENT_STATE)
1493         break;
1494       *encPtr = encodingTable[UTF_16LE_ENC];
1495       return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1496     case 0xFFFE:
1497       if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC
1498           && state == XML_CONTENT_STATE)
1499         break;
1500       *nextTokPtr = ptr + 2;
1501       *encPtr = encodingTable[UTF_16LE_ENC];
1502       return XML_TOK_BOM;
1503     case 0xEFBB:
1504       /* Maybe a UTF-8 BOM (EF BB BF) */
1505       /* If there's an explicitly specified (external) encoding
1506          of ISO-8859-1 or some flavour of UTF-16
1507          and this is an external text entity,
1508          don't look for the BOM,
1509          because it might be a legal data. */
1510       if (state == XML_CONTENT_STATE) {
1511         int e = INIT_ENC_INDEX(enc);
1512         if (e == ISO_8859_1_ENC || e == UTF_16BE_ENC || e == UTF_16LE_ENC || e == UTF_16_ENC)
1513           break;
1514       }
1515       if (ptr + 2 == end)
1516         return XML_TOK_PARTIAL;
1517       if ((unsigned char)ptr[2] == 0xBF) {
1518         *nextTokPtr = ptr + 3;
1519         *encPtr = encodingTable[UTF_8_ENC];
1520         return XML_TOK_BOM;
1521       }
1522       break;
1523     default:
1524       if (ptr[0] == '\0') {
1525         /* 0 isn't a legal data character. Furthermore a document entity can only
1526            start with ASCII characters.  So the only way this can fail to be big-endian
1527            UTF-16 if it it's an external parsed general entity that's labelled as
1528            UTF-16LE. */
1529         if (state == XML_CONTENT_STATE && INIT_ENC_INDEX(enc) == UTF_16LE_ENC)
1530           break;
1531         *encPtr = encodingTable[UTF_16BE_ENC];
1532         return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1533       }
1534       else if (ptr[1] == '\0') {
1535         /* We could recover here in the case:
1536             - parsing an external entity
1537             - second byte is 0
1538             - no externally specified encoding
1539             - no encoding declaration
1540            by assuming UTF-16LE.  But we don't, because this would mean when
1541            presented just with a single byte, we couldn't reliably determine
1542            whether we needed further bytes. */
1543         if (state == XML_CONTENT_STATE)
1544           break;
1545         *encPtr = encodingTable[UTF_16LE_ENC];
1546         return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1547       }
1548       break;
1549     }
1550   }
1551   *encPtr = encodingTable[INIT_ENC_INDEX(enc)];
1552   return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1553 }
1554
1555
1556 #define NS(x) x
1557 #define ns(x) x
1558 #include "xmltok_ns.c"
1559 #undef NS
1560 #undef ns
1561
1562 #ifdef XML_NS
1563
1564 #define NS(x) x ## NS
1565 #define ns(x) x ## _ns
1566
1567 #include "xmltok_ns.c"
1568
1569 #undef NS
1570 #undef ns
1571
1572 ENCODING *
1573 XmlInitUnknownEncodingNS(void *mem,
1574                          int *table,
1575                          int (*convert)(void *userData, const char *p),
1576                          void *userData)
1577 {
1578   ENCODING *enc = XmlInitUnknownEncoding(mem, table, convert, userData);
1579   if (enc)
1580     ((struct normal_encoding *)enc)->type[ASCII_COLON] = BT_COLON;
1581   return enc;
1582 }
1583
1584 #endif /* XML_NS */