src/expat/lib/xmltok.c

   1 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
   2    See the file COPYING for copying permission.
   3 */
   4
   5 #ifdef COMPILED_FROM_DSP
   6 #include "winconfig.h"
   7 #elif defined(OS2_32)
   8 #include "os2config.h"
   9 #elif defined(__MSDOS__)
  10 #include "dosconfig.h"
  11 #elif defined(MACOS_CLASSIC)
  12 #include "macconfig.h"
  13 #else
  14 #include "expat_config.h"
  15 #endif /* ndef COMPILED_FROM_DSP */
  16
  17 #include "internal.h"
  18 #include "xmltok.h"
  19 #include "nametab.h"
  20
  21 #ifdef XML_DTD
  22 #define IGNORE_SECTION_TOK_VTABLE , PREFIX(ignoreSectionTok)
  23 #else
  24 #define IGNORE_SECTION_TOK_VTABLE /* as nothing */
  25 #endif
  26
  27 #define VTABLE1 \
  28   { PREFIX(prologTok), PREFIX(contentTok), \
  29     PREFIX(cdataSectionTok) IGNORE_SECTION_TOK_VTABLE }, \
  30   { PREFIX(attributeValueTok), PREFIX(entityValueTok) }, \
  31   PREFIX(sameName), \
  32   PREFIX(nameMatchesAscii), \
  33   PREFIX(nameLength), \
  34   PREFIX(skipS), \
  35   PREFIX(getAtts), \
  36   PREFIX(charRefNumber), \
  37   PREFIX(predefinedEntityName), \
  38   PREFIX(updatePosition), \
  39   PREFIX(isPublicId)
  40
  41 #define VTABLE VTABLE1, PREFIX(toUtf8), PREFIX(toUtf16)
  42
  43 #define UCS2_GET_NAMING(pages, hi, lo) \
  44    (namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1 << ((lo) & 0x1F)))
  45
  46 /* A 2 byte UTF-8 representation splits the characters 11 bits between
  47    the bottom 5 and 6 bits of the bytes.  We need 8 bits to index into
  48    pages, 3 bits to add to that index and 5 bits to generate the mask.
  49 */
  50 #define UTF8_GET_NAMING2(pages, byte) \
  51     (namingBitmap[((pages)[(((byte)[0]) >> 2) & 7] << 3) \
  52                       + ((((byte)[0]) & 3) << 1) \
  53                       + ((((byte)[1]) >> 5) & 1)] \
  54          & (1 << (((byte)[1]) & 0x1F)))
  55
  56 /* A 3 byte UTF-8 representation splits the characters 16 bits between
  57    the bottom 4, 6 and 6 bits of the bytes.  We need 8 bits to index
  58    into pages, 3 bits to add to that index and 5 bits to generate the
  59    mask.
  60 */
  61 #define UTF8_GET_NAMING3(pages, byte) \
  62   (namingBitmap[((pages)[((((byte)[0]) & 0xF) << 4) \
  63                              + ((((byte)[1]) >> 2) & 0xF)] \
  64                        << 3) \
  65                       + ((((byte)[1]) & 3) << 1) \
  66                       + ((((byte)[2]) >> 5) & 1)] \
  67          & (1 << (((byte)[2]) & 0x1F)))
  68
  69 #define UTF8_GET_NAMING(pages, p, n) \
  70   ((n) == 2 \
  71   ? UTF8_GET_NAMING2(pages, (const unsigned char *)(p)) \
  72   : ((n) == 3 \
  73      ? UTF8_GET_NAMING3(pages, (const unsigned char *)(p)) \
  74      : 0))
  75
  76 /* Detection of invalid UTF-8 sequences is based on Table 3.1B
  77    of Unicode 3.2: http://www.unicode.org/unicode/reports/tr28/
  78    with the additional restriction of not allowing the Unicode
  79    code points 0xFFFF and 0xFFFE (sequences EF,BF,BF and EF,BF,BE).
  80    Implementation details:
  81      (A & 0x80) == 0     means A < 0x80
  82    and
  83      (A & 0xC0) == 0xC0  means A > 0xBF
  84 */
  85
  86 #define UTF8_INVALID2(p) \
  87   ((*p) < 0xC2 || ((p)[1] & 0x80) == 0 || ((p)[1] & 0xC0) == 0xC0)
  88
  89 #define UTF8_INVALID3(p) \
  90   (((p)[2] & 0x80) == 0 \
  91   || \
  92   ((*p) == 0xEF && (p)[1] == 0xBF \
  93     ? \
  94     (p)[2] > 0xBD \
  95     : \
  96     ((p)[2] & 0xC0) == 0xC0) \
  97   || \
  98   ((*p) == 0xE0 \
  99     ? \
 100     (p)[1] < 0xA0 || ((p)[1] & 0xC0) == 0xC0 \
 101     : \
 102     ((p)[1] & 0x80) == 0 \
 103     || \
 104     ((*p) == 0xED ? (p)[1] > 0x9F : ((p)[1] & 0xC0) == 0xC0)))
 105
 106 #define UTF8_INVALID4(p) \
 107   (((p)[3] & 0x80) == 0 || ((p)[3] & 0xC0) == 0xC0 \
 108   || \
 109   ((p)[2] & 0x80) == 0 || ((p)[2] & 0xC0) == 0xC0 \
 110   || \
 111   ((*p) == 0xF0 \
 112     ? \
 113     (p)[1] < 0x90 || ((p)[1] & 0xC0) == 0xC0 \
 114     : \
 115     ((p)[1] & 0x80) == 0 \
 116     || \
 117     ((*p) == 0xF4 ? (p)[1] > 0x8F : ((p)[1] & 0xC0) == 0xC0)))
 118
 119 static int PTRFASTCALL
 120 isNever(const ENCODING *enc, const char *p)
 121 {
 122   return 0;
 123 }
 124
 125 static int PTRFASTCALL
 126 utf8_isName2(const ENCODING *enc, const char *p)
 127 {
 128   return UTF8_GET_NAMING2(namePages, (const unsigned char *)p);
 129 }
 130
 131 static int PTRFASTCALL
 132 utf8_isName3(const ENCODING *enc, const char *p)
 133 {
 134   return UTF8_GET_NAMING3(namePages, (const unsigned char *)p);
 135 }
 136
 137 #define utf8_isName4 isNever
 138
 139 static int PTRFASTCALL
 140 utf8_isNmstrt2(const ENCODING *enc, const char *p)
 141 {
 142   return UTF8_GET_NAMING2(nmstrtPages, (const unsigned char *)p);
 143 }
 144
 145 static int PTRFASTCALL
 146 utf8_isNmstrt3(const ENCODING *enc, const char *p)
 147 {
 148   return UTF8_GET_NAMING3(nmstrtPages, (const unsigned char *)p);
 149 }
 150
 151 #define utf8_isNmstrt4 isNever
 152
 153 static int PTRFASTCALL
 154 utf8_isInvalid2(const ENCODING *enc, const char *p)
 155 {
 156   return UTF8_INVALID2((const unsigned char *)p);
 157 }
 158
 159 static int PTRFASTCALL
 160 utf8_isInvalid3(const ENCODING *enc, const char *p)
 161 {
 162   return UTF8_INVALID3((const unsigned char *)p);
 163 }
 164
 165 static int PTRFASTCALL
 166 utf8_isInvalid4(const ENCODING *enc, const char *p)
 167 {
 168   return UTF8_INVALID4((const unsigned char *)p);
 169 }
 170
 171 struct normal_encoding {
 172   ENCODING enc;
 173   unsigned char type[256];
 174 #ifdef XML_MIN_SIZE
 175   int (PTRFASTCALL *byteType)(const ENCODING *, const char *);
 176   int (PTRFASTCALL *isNameMin)(const ENCODING *, const char *);
 177   int (PTRFASTCALL *isNmstrtMin)(const ENCODING *, const char *);
 178   int (PTRFASTCALL *byteToAscii)(const ENCODING *, const char *);
 179   int (PTRCALL *charMatches)(const ENCODING *, const char *, int);
 180 #endif /* XML_MIN_SIZE */
 181   int (PTRFASTCALL *isName2)(const ENCODING *, const char *);
 182   int (PTRFASTCALL *isName3)(const ENCODING *, const char *);
 183   int (PTRFASTCALL *isName4)(const ENCODING *, const char *);
 184   int (PTRFASTCALL *isNmstrt2)(const ENCODING *, const char *);
 185   int (PTRFASTCALL *isNmstrt3)(const ENCODING *, const char *);
 186   int (PTRFASTCALL *isNmstrt4)(const ENCODING *, const char *);
 187   int (PTRFASTCALL *isInvalid2)(const ENCODING *, const char *);
 188   int (PTRFASTCALL *isInvalid3)(const ENCODING *, const char *);
 189   int (PTRFASTCALL *isInvalid4)(const ENCODING *, const char *);
 190 };
 191
 192 #define AS_NORMAL_ENCODING(enc)   ((const struct normal_encoding *) (enc))
 193
 194 #ifdef XML_MIN_SIZE
 195
 196 #define STANDARD_VTABLE(E) \
 197  E ## byteType, \
 198  E ## isNameMin, \
 199  E ## isNmstrtMin, \
 200  E ## byteToAscii, \
 201  E ## charMatches,
 202
 203 #else
 204
 205 #define STANDARD_VTABLE(E) /* as nothing */
 206
 207 #endif
 208
 209 #define NORMAL_VTABLE(E) \
 210  E ## isName2, \
 211  E ## isName3, \
 212  E ## isName4, \
 213  E ## isNmstrt2, \
 214  E ## isNmstrt3, \
 215  E ## isNmstrt4, \
 216  E ## isInvalid2, \
 217  E ## isInvalid3, \
 218  E ## isInvalid4
 219
 220 static int FASTCALL checkCharRefNumber(int);
 221
 222 #include "xmltok_impl.h"
 223 #include "ascii.h"
 224
 225 #ifdef XML_MIN_SIZE
 226 #define sb_isNameMin isNever
 227 #define sb_isNmstrtMin isNever
 228 #endif
 229
 230 #ifdef XML_MIN_SIZE
 231 #define MINBPC(enc) ((enc)->minBytesPerChar)
 232 #else
 233 /* minimum bytes per character */
 234 #define MINBPC(enc) 1
 235 #endif
 236
 237 #define SB_BYTE_TYPE(enc, p) \
 238   (((struct normal_encoding *)(enc))->type[(unsigned char)*(p)])
 239
 240 #ifdef XML_MIN_SIZE
 241 static int PTRFASTCALL
 242 sb_byteType(const ENCODING *enc, const char *p)
 243 {
 244   return SB_BYTE_TYPE(enc, p);
 245 }
 246 #define BYTE_TYPE(enc, p) \
 247  (AS_NORMAL_ENCODING(enc)->byteType(enc, p))
 248 #else
 249 #define BYTE_TYPE(enc, p) SB_BYTE_TYPE(enc, p)
 250 #endif
 251
 252 #ifdef XML_MIN_SIZE
 253 #define BYTE_TO_ASCII(enc, p) \
 254  (AS_NORMAL_ENCODING(enc)->byteToAscii(enc, p))
 255 static int PTRFASTCALL
 256 sb_byteToAscii(const ENCODING *enc, const char *p)
 257 {
 258   return *p;
 259 }
 260 #else
 261 #define BYTE_TO_ASCII(enc, p) (*(p))
 262 #endif
 263
 264 #define IS_NAME_CHAR(enc, p, n) \
 265  (AS_NORMAL_ENCODING(enc)->isName ## n(enc, p))
 266 #define IS_NMSTRT_CHAR(enc, p, n) \
 267  (AS_NORMAL_ENCODING(enc)->isNmstrt ## n(enc, p))
 268 #define IS_INVALID_CHAR(enc, p, n) \
 269  (AS_NORMAL_ENCODING(enc)->isInvalid ## n(enc, p))
 270
 271 #ifdef XML_MIN_SIZE
 272 #define IS_NAME_CHAR_MINBPC(enc, p) \
 273  (AS_NORMAL_ENCODING(enc)->isNameMin(enc, p))
 274 #define IS_NMSTRT_CHAR_MINBPC(enc, p) \
 275  (AS_NORMAL_ENCODING(enc)->isNmstrtMin(enc, p))
 276 #else
 277 #define IS_NAME_CHAR_MINBPC(enc, p) (0)
 278 #define IS_NMSTRT_CHAR_MINBPC(enc, p) (0)
 279 #endif
 280
 281 #ifdef XML_MIN_SIZE
 282 #define CHAR_MATCHES(enc, p, c) \
 283  (AS_NORMAL_ENCODING(enc)->charMatches(enc, p, c))
 284 static int PTRCALL
 285 sb_charMatches(const ENCODING *enc, const char *p, int c)
 286 {
 287   return *p == c;
 288 }
 289 #else
 290 /* c is an ASCII character */
 291 #define CHAR_MATCHES(enc, p, c) (*(p) == c)
 292 #endif
 293
 294 #define PREFIX(ident) normal_ ## ident
 295 #include "xmltok_impl.c"
 296
 297 #undef MINBPC
 298 #undef BYTE_TYPE
 299 #undef BYTE_TO_ASCII
 300 #undef CHAR_MATCHES
 301 #undef IS_NAME_CHAR
 302 #undef IS_NAME_CHAR_MINBPC
 303 #undef IS_NMSTRT_CHAR
 304 #undef IS_NMSTRT_CHAR_MINBPC
 305 #undef IS_INVALID_CHAR
 306
 307 enum {  /* UTF8_cvalN is value of masked first byte of N byte sequence */
 308   UTF8_cval1 = 0x00,
 309   UTF8_cval2 = 0xc0,
 310   UTF8_cval3 = 0xe0,
 311   UTF8_cval4 = 0xf0
 312 };
 313
 314 static void PTRCALL
 315 utf8_toUtf8(const ENCODING *enc,
 316             const char **fromP, const char *fromLim,
 317             char **toP, const char *toLim)
 318 {
 319   char *to;
 320   const char *from;
 321   if (fromLim - *fromP > toLim - *toP) {
 322     /* Avoid copying partial characters. */
 323     for (fromLim = *fromP + (toLim - *toP); fromLim > *fromP; fromLim--)
 324       if (((unsigned char)fromLim[-1] & 0xc0) != 0x80)
 325         break;
 326   }
 327   for (to = *toP, from = *fromP; from != fromLim; from++, to++)
 328     *to = *from;
 329   *fromP = from;
 330   *toP = to;
 331 }
 332
 333 static void PTRCALL
 334 utf8_toUtf16(const ENCODING *enc,
 335              const char **fromP, const char *fromLim,
 336              unsigned short **toP, const unsigned short *toLim)
 337 {
 338   unsigned short *to = *toP;
 339   const char *from = *fromP;
 340   while (from != fromLim && to != toLim) {
 341     switch (((struct normal_encoding *)enc)->type[(unsigned char)*from]) {
 342     case BT_LEAD2:
 343       *to++ = (unsigned short)(((from[0] & 0x1f) << 6) | (from[1] & 0x3f));
 344       from += 2;
 345       break;
 346     case BT_LEAD3:
 347       *to++ = (unsigned short)(((from[0] & 0xf) << 12)
 348                                | ((from[1] & 0x3f) << 6) | (from[2] & 0x3f));
 349       from += 3;
 350       break;
 351     case BT_LEAD4:
 352       {
 353         unsigned long n;
 354         if (to + 1 == toLim)
 355           goto after;
 356         n = ((from[0] & 0x7) << 18) | ((from[1] & 0x3f) << 12)
 357             | ((from[2] & 0x3f) << 6) | (from[3] & 0x3f);
 358         n -= 0x10000;
 359         to[0] = (unsigned short)((n >> 10) | 0xD800);
 360         to[1] = (unsigned short)((n & 0x3FF) | 0xDC00);
 361         to += 2;
 362         from += 4;
 363       }
 364       break;
 365     default:
 366       *to++ = *from++;
 367       break;
 368     }
 369   }
 370 after:
 371   *fromP = from;
 372   *toP = to;
 373 }
 374
 375 #ifdef XML_NS
 376 static const struct normal_encoding utf8_encoding_ns = {
 377   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 378   {
 379 #include "asciitab.h"
 380 #include "utf8tab.h"
 381   },
 382   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 383 };
 384 #endif
 385
 386 static const struct normal_encoding utf8_encoding = {
 387   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 388   {
 389 #define BT_COLON BT_NMSTRT
 390 #include "asciitab.h"
 391 #undef BT_COLON
 392 #include "utf8tab.h"
 393   },
 394   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 395 };
 396
 397 #ifdef XML_NS
 398
 399 static const struct normal_encoding internal_utf8_encoding_ns = {
 400   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 401   {
 402 #include "iasciitab.h"
 403 #include "utf8tab.h"
 404   },
 405   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 406 };
 407
 408 #endif
 409
 410 static const struct normal_encoding internal_utf8_encoding = {
 411   { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 },
 412   {
 413 #define BT_COLON BT_NMSTRT
 414 #include "iasciitab.h"
 415 #undef BT_COLON
 416 #include "utf8tab.h"
 417   },
 418   STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_)
 419 };
 420
 421 static void PTRCALL
 422 latin1_toUtf8(const ENCODING *enc,
 423               const char **fromP, const char *fromLim,
 424               char **toP, const char *toLim)
 425 {
 426   for (;;) {
 427     unsigned char c;
 428     if (*fromP == fromLim)
 429       break;
 430     c = (unsigned char)**fromP;
 431     if (c & 0x80) {
 432       if (toLim - *toP < 2)
 433         break;
 434       *(*toP)++ = (char)((c >> 6) | UTF8_cval2);
 435       *(*toP)++ = (char)((c & 0x3f) | 0x80);
 436       (*fromP)++;
 437     }
 438     else {
 439       if (*toP == toLim)
 440         break;
 441       *(*toP)++ = *(*fromP)++;
 442     }
 443   }
 444 }
 445
 446 static void PTRCALL
 447 latin1_toUtf16(const ENCODING *enc,
 448                const char **fromP, const char *fromLim,
 449                unsigned short **toP, const unsigned short *toLim)
 450 {
 451   while (*fromP != fromLim && *toP != toLim)
 452     *(*toP)++ = (unsigned char)*(*fromP)++;
 453 }
 454
 455 #ifdef XML_NS
 456
 457 static const struct normal_encoding latin1_encoding_ns = {
 458   { VTABLE1, latin1_toUtf8, latin1_toUtf16, 1, 0, 0 },
 459   {
 460 #include "asciitab.h"
 461 #include "latin1tab.h"
 462   },
 463   STANDARD_VTABLE(sb_)
 464 };
 465
 466 #endif
 467
 468 static const struct normal_encoding latin1_encoding = {
 469   { VTABLE1, latin1_toUtf8, latin1_toUtf16, 1, 0, 0 },
 470   {
 471 #define BT_COLON BT_NMSTRT
 472 #include "asciitab.h"
 473 #undef BT_COLON
 474 #include "latin1tab.h"
 475   },
 476   STANDARD_VTABLE(sb_)
 477 };
 478
 479 static void PTRCALL
 480 ascii_toUtf8(const ENCODING *enc,
 481              const char **fromP, const char *fromLim,
 482              char **toP, const char *toLim)
 483 {
 484   while (*fromP != fromLim && *toP != toLim)
 485     *(*toP)++ = *(*fromP)++;
 486 }
 487
 488 #ifdef XML_NS
 489
 490 static const struct normal_encoding ascii_encoding_ns = {
 491   { VTABLE1, ascii_toUtf8, latin1_toUtf16, 1, 1, 0 },
 492   {
 493 #include "asciitab.h"
 494 /* BT_NONXML == 0 */
 495   },
 496   STANDARD_VTABLE(sb_)
 497 };
 498
 499 #endif
 500
 501 static const struct normal_encoding ascii_encoding = {
 502   { VTABLE1, ascii_toUtf8, latin1_toUtf16, 1, 1, 0 },
 503   {
 504 #define BT_COLON BT_NMSTRT
 505 #include "asciitab.h"
 506 #undef BT_COLON
 507 /* BT_NONXML == 0 */
 508   },
 509   STANDARD_VTABLE(sb_)
 510 };
 511
 512 static int PTRFASTCALL
 513 unicode_byte_type(char hi, char lo)
 514 {
 515   switch ((unsigned char)hi) {
 516   case 0xD8: case 0xD9: case 0xDA: case 0xDB:
 517     return BT_LEAD4;
 518   case 0xDC: case 0xDD: case 0xDE: case 0xDF:
 519     return BT_TRAIL;
 520   case 0xFF:
 521     switch ((unsigned char)lo) {
 522     case 0xFF:
 523     case 0xFE:
 524       return BT_NONXML;
 525     }
 526     break;
 527   }
 528   return BT_NONASCII;
 529 }
 530
 531 #define DEFINE_UTF16_TO_UTF8(E) \
 532 static void  PTRCALL \
 533 E ## toUtf8(const ENCODING *enc, \
 534             const char **fromP, const char *fromLim, \
 535             char **toP, const char *toLim) \
 536 { \
 537   const char *from; \
 538   for (from = *fromP; from != fromLim; from += 2) { \
 539     int plane; \
 540     unsigned char lo2; \
 541     unsigned char lo = GET_LO(from); \
 542     unsigned char hi = GET_HI(from); \
 543     switch (hi) { \
 544     case 0: \
 545       if (lo < 0x80) { \
 546         if (*toP == toLim) { \
 547           *fromP = from; \
 548           return; \
 549         } \
 550         *(*toP)++ = lo; \
 551         break; \
 552       } \
 553       /* fall through */ \
 554     case 0x1: case 0x2: case 0x3: \
 555     case 0x4: case 0x5: case 0x6: case 0x7: \
 556       if (toLim -  *toP < 2) { \
 557         *fromP = from; \
 558         return; \
 559       } \
 560       *(*toP)++ = ((lo >> 6) | (hi << 2) |  UTF8_cval2); \
 561       *(*toP)++ = ((lo & 0x3f) | 0x80); \
 562       break; \
 563     default: \
 564       if (toLim -  *toP < 3)  { \
 565         *fromP = from; \
 566         return; \
 567       } \
 568       /* 16 bits divided 4, 6, 6 amongst 3 bytes */ \
 569       *(*toP)++ = ((hi >> 4) | UTF8_cval3); \
 570       *(*toP)++ = (((hi & 0xf) << 2) | (lo >> 6) | 0x80); \
 571       *(*toP)++ = ((lo & 0x3f) | 0x80); \
 572       break; \
 573     case 0xD8: case 0xD9: case 0xDA: case 0xDB: \
 574       if (toLim -  *toP < 4) { \
 575         *fromP = from; \
 576         return; \
 577       } \
 578       plane = (((hi & 0x3) << 2) | ((lo >> 6) & 0x3)) + 1; \
 579       *(*toP)++ = ((plane >> 2) | UTF8_cval4); \
 580       *(*toP)++ = (((lo >> 2) & 0xF) | ((plane & 0x3) << 4) | 0x80); \
 581       from += 2; \
 582       lo2 = GET_LO(from); \
 583       *(*toP)++ = (((lo & 0x3) << 4) \
 584                    | ((GET_HI(from) & 0x3) << 2) \
 585                    | (lo2 >> 6) \
 586                    | 0x80); \
 587       *(*toP)++ = ((lo2 & 0x3f) | 0x80); \
 588       break; \
 589     } \
 590   } \
 591   *fromP = from; \
 592 }
 593
 594 #define DEFINE_UTF16_TO_UTF16(E) \
 595 static void  PTRCALL \
 596 E ## toUtf16(const ENCODING *enc, \
 597              const char **fromP, const char *fromLim, \
 598              unsigned short **toP, const unsigned short *toLim) \
 599 { \
 600   /* Avoid copying first half only of surrogate */ \
 601   if (fromLim - *fromP > ((toLim - *toP) << 1) \
 602       && (GET_HI(fromLim - 2) & 0xF8) == 0xD8) \
 603     fromLim -= 2; \
 604   for (; *fromP != fromLim && *toP != toLim; *fromP += 2) \
 605     *(*toP)++ = (GET_HI(*fromP) << 8) | GET_LO(*fromP); \
 606 }
 607
 608 #define SET2(ptr, ch) \
 609   (((ptr)[0] = ((ch) & 0xff)), ((ptr)[1] = ((ch) >> 8)))
 610 #define GET_LO(ptr) ((unsigned char)(ptr)[0])
 611 #define GET_HI(ptr) ((unsigned char)(ptr)[1])
 612
 613 DEFINE_UTF16_TO_UTF8(little2_)
 614 DEFINE_UTF16_TO_UTF16(little2_)
 615
 616 #undef SET2
 617 #undef GET_LO
 618 #undef GET_HI
 619
 620 #define SET2(ptr, ch) \
 621   (((ptr)[0] = ((ch) >> 8)), ((ptr)[1] = ((ch) & 0xFF)))
 622 #define GET_LO(ptr) ((unsigned char)(ptr)[1])
 623 #define GET_HI(ptr) ((unsigned char)(ptr)[0])
 624
 625 DEFINE_UTF16_TO_UTF8(big2_)
 626 DEFINE_UTF16_TO_UTF16(big2_)
 627
 628 #undef SET2
 629 #undef GET_LO
 630 #undef GET_HI
 631
 632 #define LITTLE2_BYTE_TYPE(enc, p) \
 633  ((p)[1] == 0 \
 634   ? ((struct normal_encoding *)(enc))->type[(unsigned char)*(p)] \
 635   : unicode_byte_type((p)[1], (p)[0]))
 636 #define LITTLE2_BYTE_TO_ASCII(enc, p) ((p)[1] == 0 ? (p)[0] : -1)
 637 #define LITTLE2_CHAR_MATCHES(enc, p, c) ((p)[1] == 0 && (p)[0] == c)
 638 #define LITTLE2_IS_NAME_CHAR_MINBPC(enc, p) \
 639   UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0])
 640 #define LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p) \
 641   UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0])
 642
 643 #ifdef XML_MIN_SIZE
 644
 645 static int PTRFASTCALL
 646 little2_byteType(const ENCODING *enc, const char *p)
 647 {
 648   return LITTLE2_BYTE_TYPE(enc, p);
 649 }
 650
 651 static int PTRFASTCALL
 652 little2_byteToAscii(const ENCODING *enc, const char *p)
 653 {
 654   return LITTLE2_BYTE_TO_ASCII(enc, p);
 655 }
 656
 657 static int PTRCALL
 658 little2_charMatches(const ENCODING *enc, const char *p, int c)
 659 {
 660   return LITTLE2_CHAR_MATCHES(enc, p, c);
 661 }
 662
 663 static int PTRFASTCALL
 664 little2_isNameMin(const ENCODING *enc, const char *p)
 665 {
 666   return LITTLE2_IS_NAME_CHAR_MINBPC(enc, p);
 667 }
 668
 669 static int PTRFASTCALL
 670 little2_isNmstrtMin(const ENCODING *enc, const char *p)
 671 {
 672   return LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p);
 673 }
 674
 675 #undef VTABLE
 676 #define VTABLE VTABLE1, little2_toUtf8, little2_toUtf16
 677
 678 #else /* not XML_MIN_SIZE */
 679
 680 #undef PREFIX
 681 #define PREFIX(ident) little2_ ## ident
 682 #define MINBPC(enc) 2
 683 /* CHAR_MATCHES is guaranteed to have MINBPC bytes available. */
 684 #define BYTE_TYPE(enc, p) LITTLE2_BYTE_TYPE(enc, p)
 685 #define BYTE_TO_ASCII(enc, p) LITTLE2_BYTE_TO_ASCII(enc, p)
 686 #define CHAR_MATCHES(enc, p, c) LITTLE2_CHAR_MATCHES(enc, p, c)
 687 #define IS_NAME_CHAR(enc, p, n) 0
 688 #define IS_NAME_CHAR_MINBPC(enc, p) LITTLE2_IS_NAME_CHAR_MINBPC(enc, p)
 689 #define IS_NMSTRT_CHAR(enc, p, n) (0)
 690 #define IS_NMSTRT_CHAR_MINBPC(enc, p) LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p)
 691
 692 #include "xmltok_impl.c"
 693
 694 #undef MINBPC
 695 #undef BYTE_TYPE
 696 #undef BYTE_TO_ASCII
 697 #undef CHAR_MATCHES
 698 #undef IS_NAME_CHAR
 699 #undef IS_NAME_CHAR_MINBPC
 700 #undef IS_NMSTRT_CHAR
 701 #undef IS_NMSTRT_CHAR_MINBPC
 702 #undef IS_INVALID_CHAR
 703
 704 #endif /* not XML_MIN_SIZE */
 705
 706 #ifdef XML_NS
 707
 708 static const struct normal_encoding little2_encoding_ns = {
 709   { VTABLE, 2, 0,
 710 #if BYTEORDER == 1234
 711     1
 712 #else
 713     0
 714 #endif
 715   },
 716   {
 717 #include "asciitab.h"
 718 #include "latin1tab.h"
 719   },
 720   STANDARD_VTABLE(little2_)
 721 };
 722
 723 #endif
 724
 725 static const struct normal_encoding little2_encoding = {
 726   { VTABLE, 2, 0,
 727 #if BYTEORDER == 1234
 728     1
 729 #else
 730     0
 731 #endif
 732   },
 733   {
 734 #define BT_COLON BT_NMSTRT
 735 #include "asciitab.h"
 736 #undef BT_COLON
 737 #include "latin1tab.h"
 738   },
 739   STANDARD_VTABLE(little2_)
 740 };
 741
 742 #if BYTEORDER != 4321
 743
 744 #ifdef XML_NS
 745
 746 static const struct normal_encoding internal_little2_encoding_ns = {
 747   { VTABLE, 2, 0, 1 },
 748   {
 749 #include "iasciitab.h"
 750 #include "latin1tab.h"
 751   },
 752   STANDARD_VTABLE(little2_)
 753 };
 754
 755 #endif
 756
 757 static const struct normal_encoding internal_little2_encoding = {
 758   { VTABLE, 2, 0, 1 },
 759   {
 760 #define BT_COLON BT_NMSTRT
 761 #include "iasciitab.h"
 762 #undef BT_COLON
 763 #include "latin1tab.h"
 764   },
 765   STANDARD_VTABLE(little2_)
 766 };
 767
 768 #endif
 769
 770
 771 #define BIG2_BYTE_TYPE(enc, p) \
 772  ((p)[0] == 0 \
 773   ? ((struct normal_encoding *)(enc))->type[(unsigned char)(p)[1]] \
 774   : unicode_byte_type((p)[0], (p)[1]))
 775 #define BIG2_BYTE_TO_ASCII(enc, p) ((p)[0] == 0 ? (p)[1] : -1)
 776 #define BIG2_CHAR_MATCHES(enc, p, c) ((p)[0] == 0 && (p)[1] == c)
 777 #define BIG2_IS_NAME_CHAR_MINBPC(enc, p) \
 778   UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1])
 779 #define BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p) \
 780   UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1])
 781
 782 #ifdef XML_MIN_SIZE
 783
 784 static int PTRFASTCALL
 785 big2_byteType(const ENCODING *enc, const char *p)
 786 {
 787   return BIG2_BYTE_TYPE(enc, p);
 788 }
 789
 790 static int PTRFASTCALL
 791 big2_byteToAscii(const ENCODING *enc, const char *p)
 792 {
 793   return BIG2_BYTE_TO_ASCII(enc, p);
 794 }
 795
 796 static int PTRCALL
 797 big2_charMatches(const ENCODING *enc, const char *p, int c)
 798 {
 799   return BIG2_CHAR_MATCHES(enc, p, c);
 800 }
 801
 802 static int PTRFASTCALL
 803 big2_isNameMin(const ENCODING *enc, const char *p)
 804 {
 805   return BIG2_IS_NAME_CHAR_MINBPC(enc, p);
 806 }
 807
 808 static int PTRFASTCALL
 809 big2_isNmstrtMin(const ENCODING *enc, const char *p)
 810 {
 811   return BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p);
 812 }
 813
 814 #undef VTABLE
 815 #define VTABLE VTABLE1, big2_toUtf8, big2_toUtf16
 816
 817 #else /* not XML_MIN_SIZE */
 818
 819 #undef PREFIX
 820 #define PREFIX(ident) big2_ ## ident
 821 #define MINBPC(enc) 2
 822 /* CHAR_MATCHES is guaranteed to have MINBPC bytes available. */
 823 #define BYTE_TYPE(enc, p) BIG2_BYTE_TYPE(enc, p)
 824 #define BYTE_TO_ASCII(enc, p) BIG2_BYTE_TO_ASCII(enc, p)
 825 #define CHAR_MATCHES(enc, p, c) BIG2_CHAR_MATCHES(enc, p, c)
 826 #define IS_NAME_CHAR(enc, p, n) 0
 827 #define IS_NAME_CHAR_MINBPC(enc, p) BIG2_IS_NAME_CHAR_MINBPC(enc, p)
 828 #define IS_NMSTRT_CHAR(enc, p, n) (0)
 829 #define IS_NMSTRT_CHAR_MINBPC(enc, p) BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p)
 830
 831 #include "xmltok_impl.c"
 832
 833 #undef MINBPC
 834 #undef BYTE_TYPE
 835 #undef BYTE_TO_ASCII
 836 #undef CHAR_MATCHES
 837 #undef IS_NAME_CHAR
 838 #undef IS_NAME_CHAR_MINBPC
 839 #undef IS_NMSTRT_CHAR
 840 #undef IS_NMSTRT_CHAR_MINBPC
 841 #undef IS_INVALID_CHAR
 842
 843 #endif /* not XML_MIN_SIZE */
 844
 845 #ifdef XML_NS
 846
 847 static const struct normal_encoding big2_encoding_ns = {
 848   { VTABLE, 2, 0,
 849 #if BYTEORDER == 4321
 850   1
 851 #else
 852   0
 853 #endif
 854   },
 855   {
 856 #include "asciitab.h"
 857 #include "latin1tab.h"
 858   },
 859   STANDARD_VTABLE(big2_)
 860 };
 861
 862 #endif
 863
 864 static const struct normal_encoding big2_encoding = {
 865   { VTABLE, 2, 0,
 866 #if BYTEORDER == 4321
 867   1
 868 #else
 869   0
 870 #endif
 871   },
 872   {
 873 #define BT_COLON BT_NMSTRT
 874 #include "asciitab.h"
 875 #undef BT_COLON
 876 #include "latin1tab.h"
 877   },
 878   STANDARD_VTABLE(big2_)
 879 };
 880
 881 #if BYTEORDER != 1234
 882
 883 #ifdef XML_NS
 884
 885 static const struct normal_encoding internal_big2_encoding_ns = {
 886   { VTABLE, 2, 0, 1 },
 887   {
 888 #include "iasciitab.h"
 889 #include "latin1tab.h"
 890   },
 891   STANDARD_VTABLE(big2_)
 892 };
 893
 894 #endif
 895
 896 static const struct normal_encoding internal_big2_encoding = {
 897   { VTABLE, 2, 0, 1 },
 898   {
 899 #define BT_COLON BT_NMSTRT
 900 #include "iasciitab.h"
 901 #undef BT_COLON
 902 #include "latin1tab.h"
 903   },
 904   STANDARD_VTABLE(big2_)
 905 };
 906
 907 #endif
 908
 909 #undef PREFIX
 910
 911 static int FASTCALL
 912 streqci(const char *s1, const char *s2)
 913 {
 914   for (;;) {
 915     char c1 = *s1++;
 916     char c2 = *s2++;
 917     if (ASCII_a <= c1 && c1 <= ASCII_z)
 918       c1 += ASCII_A - ASCII_a;
 919     if (ASCII_a <= c2 && c2 <= ASCII_z)
 920       c2 += ASCII_A - ASCII_a;
 921     if (c1 != c2)
 922       return 0;
 923     if (!c1)
 924       break;
 925   }
 926   return 1;
 927 }
 928
 929 static void PTRCALL
 930 initUpdatePosition(const ENCODING *enc, const char *ptr,
 931                    const char *end, POSITION *pos)
 932 {
 933   normal_updatePosition(&utf8_encoding.enc, ptr, end, pos);
 934 }
 935
 936 static int
 937 toAscii(const ENCODING *enc, const char *ptr, const char *end)
 938 {
 939   char buf[1];
 940   char *p = buf;
 941   XmlUtf8Convert(enc, &ptr, end, &p, p + 1);
 942   if (p == buf)
 943     return -1;
 944   else
 945     return buf[0];
 946 }
 947
 948 static int FASTCALL
 949 isSpace(int c)
 950 {
 951   switch (c) {
 952   case 0x20:
 953   case 0xD:
 954   case 0xA:
 955   case 0x9:
 956     return 1;
 957   }
 958   return 0;
 959 }
 960
 961 /* Return 1 if there's just optional white space or there's an S
 962    followed by name=val.
 963 */
 964 static int
 965 parsePseudoAttribute(const ENCODING *enc,
 966                      const char *ptr,
 967                      const char *end,
 968                      const char **namePtr,
 969                      const char **nameEndPtr,
 970                      const char **valPtr,
 971                      const char **nextTokPtr)
 972 {
 973   int c;
 974   char open;
 975   if (ptr == end) {
 976     *namePtr = NULL;
 977     return 1;
 978   }
 979   if (!isSpace(toAscii(enc, ptr, end))) {
 980     *nextTokPtr = ptr;
 981     return 0;
 982   }
 983   do {
 984     ptr += enc->minBytesPerChar;
 985   } while (isSpace(toAscii(enc, ptr, end)));
 986   if (ptr == end) {
 987     *namePtr = NULL;
 988     return 1;
 989   }
 990   *namePtr = ptr;
 991   for (;;) {
 992     c = toAscii(enc, ptr, end);
 993     if (c == -1) {
 994       *nextTokPtr = ptr;
 995       return 0;
 996     }
 997     if (c == ASCII_EQUALS) {
 998       *nameEndPtr = ptr;
 999       break;
1000     }
1001     if (isSpace(c)) {
1002       *nameEndPtr = ptr;
1003       do {
1004         ptr += enc->minBytesPerChar;
1005       } while (isSpace(c = toAscii(enc, ptr, end)));
1006       if (c != ASCII_EQUALS) {
1007         *nextTokPtr = ptr;
1008         return 0;
1009       }
1010       break;
1011     }
1012     ptr += enc->minBytesPerChar;
1013   }
1014   if (ptr == *namePtr) {
1015     *nextTokPtr = ptr;
1016     return 0;
1017   }
1018   ptr += enc->minBytesPerChar;
1019   c = toAscii(enc, ptr, end);
1020   while (isSpace(c)) {
1021     ptr += enc->minBytesPerChar;
1022     c = toAscii(enc, ptr, end);
1023   }
1024   if (c != ASCII_QUOT && c != ASCII_APOS) {
1025     *nextTokPtr = ptr;
1026     return 0;
1027   }
1028   open = (char)c;
1029   ptr += enc->minBytesPerChar;
1030   *valPtr = ptr;
1031   for (;; ptr += enc->minBytesPerChar) {
1032     c = toAscii(enc, ptr, end);
1033     if (c == open)
1034       break;
1035     if (!(ASCII_a <= c && c <= ASCII_z)
1036         && !(ASCII_A <= c && c <= ASCII_Z)
1037         && !(ASCII_0 <= c && c <= ASCII_9)
1038         && c != ASCII_PERIOD
1039         && c != ASCII_MINUS
1040         && c != ASCII_UNDERSCORE) {
1041       *nextTokPtr = ptr;
1042       return 0;
1043     }
1044   }
1045   *nextTokPtr = ptr + enc->minBytesPerChar;
1046   return 1;
1047 }
1048
1049 static const char KW_version[] = {
1050   ASCII_v, ASCII_e, ASCII_r, ASCII_s, ASCII_i, ASCII_o, ASCII_n, '\0'
1051 };
1052
1053 static const char KW_encoding[] = {
1054   ASCII_e, ASCII_n, ASCII_c, ASCII_o, ASCII_d, ASCII_i, ASCII_n, ASCII_g, '\0'
1055 };
1056
1057 static const char KW_standalone[] = {
1058   ASCII_s, ASCII_t, ASCII_a, ASCII_n, ASCII_d, ASCII_a, ASCII_l, ASCII_o,
1059   ASCII_n, ASCII_e, '\0'
1060 };
1061
1062 static const char KW_yes[] = {
1063   ASCII_y, ASCII_e, ASCII_s,  '\0'
1064 };
1065
1066 static const char KW_no[] = {
1067   ASCII_n, ASCII_o,  '\0'
1068 };
1069
1070 static int
1071 doParseXmlDecl(const ENCODING *(*encodingFinder)(const ENCODING *,
1072                                                  const char *,
1073                                                  const char *),
1074                int isGeneralTextEntity,
1075                const ENCODING *enc,
1076                const char *ptr,
1077                const char *end,
1078                const char **badPtr,
1079                const char **versionPtr,
1080                const char **versionEndPtr,
1081                const char **encodingName,
1082                const ENCODING **encoding,
1083                int *standalone)
1084 {
1085   const char *val = NULL;
1086   const char *name = NULL;
1087   const char *nameEnd = NULL;
1088   ptr += 5 * enc->minBytesPerChar;
1089   end -= 2 * enc->minBytesPerChar;
1090   if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)
1091       || !name) {
1092     *badPtr = ptr;
1093     return 0;
1094   }
1095   if (!XmlNameMatchesAscii(enc, name, nameEnd, KW_version)) {
1096     if (!isGeneralTextEntity) {
1097       *badPtr = name;
1098       return 0;
1099     }
1100   }
1101   else {
1102     if (versionPtr)
1103       *versionPtr = val;
1104     if (versionEndPtr)
1105       *versionEndPtr = ptr;
1106     if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)) {
1107       *badPtr = ptr;
1108       return 0;
1109     }
1110     if (!name) {
1111       if (isGeneralTextEntity) {
1112         /* a TextDecl must have an EncodingDecl */
1113         *badPtr = ptr;
1114         return 0;
1115       }
1116       return 1;
1117     }
1118   }
1119   if (XmlNameMatchesAscii(enc, name, nameEnd, KW_encoding)) {
1120     int c = toAscii(enc, val, end);
1121     if (!(ASCII_a <= c && c <= ASCII_z) && !(ASCII_A <= c && c <= ASCII_Z)) {
1122       *badPtr = val;
1123       return 0;
1124     }
1125     if (encodingName)
1126       *encodingName = val;
1127     if (encoding)
1128       *encoding = encodingFinder(enc, val, ptr - enc->minBytesPerChar);
1129     if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)) {
1130       *badPtr = ptr;
1131       return 0;
1132     }
1133     if (!name)
1134       return 1;
1135   }
1136   if (!XmlNameMatchesAscii(enc, name, nameEnd, KW_standalone)
1137       || isGeneralTextEntity) {
1138     *badPtr = name;
1139     return 0;
1140   }
1141   if (XmlNameMatchesAscii(enc, val, ptr - enc->minBytesPerChar, KW_yes)) {
1142     if (standalone)
1143       *standalone = 1;
1144   }
1145   else if (XmlNameMatchesAscii(enc, val, ptr - enc->minBytesPerChar, KW_no)) {
1146     if (standalone)
1147       *standalone = 0;
1148   }
1149   else {
1150     *badPtr = val;
1151     return 0;
1152   }
1153   while (isSpace(toAscii(enc, ptr, end)))
1154     ptr += enc->minBytesPerChar;
1155   if (ptr != end) {
1156     *badPtr = ptr;
1157     return 0;
1158   }
1159   return 1;
1160 }
1161
1162 static int FASTCALL
1163 checkCharRefNumber(int result)
1164 {
1165   switch (result >> 8) {
1166   case 0xD8: case 0xD9: case 0xDA: case 0xDB:
1167   case 0xDC: case 0xDD: case 0xDE: case 0xDF:
1168     return -1;
1169   case 0:
1170     if (latin1_encoding.type[result] == BT_NONXML)
1171       return -1;
1172     break;
1173   case 0xFF:
1174     if (result == 0xFFFE || result == 0xFFFF)
1175       return -1;
1176     break;
1177   }
1178   return result;
1179 }
1180
1181 int FASTCALL
1182 XmlUtf8Encode(int c, char *buf)
1183 {
1184   enum {
1185     /* minN is minimum legal resulting value for N byte sequence */
1186     min2 = 0x80,
1187     min3 = 0x800,
1188     min4 = 0x10000
1189   };
1190
1191   if (c < 0)
1192     return 0;
1193   if (c < min2) {
1194     buf[0] = (char)(c | UTF8_cval1);
1195     return 1;
1196   }
1197   if (c < min3) {
1198     buf[0] = (char)((c >> 6) | UTF8_cval2);
1199     buf[1] = (char)((c & 0x3f) | 0x80);
1200     return 2;
1201   }
1202   if (c < min4) {
1203     buf[0] = (char)((c >> 12) | UTF8_cval3);
1204     buf[1] = (char)(((c >> 6) & 0x3f) | 0x80);
1205     buf[2] = (char)((c & 0x3f) | 0x80);
1206     return 3;
1207   }
1208   if (c < 0x110000) {
1209     buf[0] = (char)((c >> 18) | UTF8_cval4);
1210     buf[1] = (char)(((c >> 12) & 0x3f) | 0x80);
1211     buf[2] = (char)(((c >> 6) & 0x3f) | 0x80);
1212     buf[3] = (char)((c & 0x3f) | 0x80);
1213     return 4;
1214   }
1215   return 0;
1216 }
1217
1218 int FASTCALL
1219 XmlUtf16Encode(int charNum, unsigned short *buf)
1220 {
1221   if (charNum < 0)
1222     return 0;
1223   if (charNum < 0x10000) {
1224     buf[0] = (unsigned short)charNum;
1225     return 1;
1226   }
1227   if (charNum < 0x110000) {
1228     charNum -= 0x10000;
1229     buf[0] = (unsigned short)((charNum >> 10) + 0xD800);
1230     buf[1] = (unsigned short)((charNum & 0x3FF) + 0xDC00);
1231     return 2;
1232   }
1233   return 0;
1234 }
1235
1236 struct unknown_encoding {
1237   struct normal_encoding normal;
1238   int (*convert)(void *userData, const char *p);
1239   void *userData;
1240   unsigned short utf16[256];
1241   char utf8[256][4];
1242 };
1243
1244 #define AS_UNKNOWN_ENCODING(enc)  ((const struct unknown_encoding *) (enc))
1245
1246 int
1247 XmlSizeOfUnknownEncoding(void)
1248 {
1249   return sizeof(struct unknown_encoding);
1250 }
1251
1252 static int PTRFASTCALL
1253 unknown_isName(const ENCODING *enc, const char *p)
1254 {
1255   const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc);
1256   int c = uenc->convert(uenc->userData, p);
1257   if (c & ~0xFFFF)
1258     return 0;
1259   return UCS2_GET_NAMING(namePages, c >> 8, c & 0xFF);
1260 }
1261
1262 static int PTRFASTCALL
1263 unknown_isNmstrt(const ENCODING *enc, const char *p)
1264 {
1265   const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc);
1266   int c = uenc->convert(uenc->userData, p);
1267   if (c & ~0xFFFF)
1268     return 0;
1269   return UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xFF);
1270 }
1271
1272 static int PTRFASTCALL
1273 unknown_isInvalid(const ENCODING *enc, const char *p)
1274 {
1275   const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc);
1276   int c = uenc->convert(uenc->userData, p);
1277   return (c & ~0xFFFF) || checkCharRefNumber(c) < 0;
1278 }
1279
1280 static void PTRCALL
1281 unknown_toUtf8(const ENCODING *enc,
1282                const char **fromP, const char *fromLim,
1283                char **toP, const char *toLim)
1284 {
1285   const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc);
1286   char buf[XML_UTF8_ENCODE_MAX];
1287   for (;;) {
1288     const char *utf8;
1289     int n;
1290     if (*fromP == fromLim)
1291       break;
1292     utf8 = uenc->utf8[(unsigned char)**fromP];
1293     n = *utf8++;
1294     if (n == 0) {
1295       int c = uenc->convert(uenc->userData, *fromP);
1296       n = XmlUtf8Encode(c, buf);
1297       if (n > toLim - *toP)
1298         break;
1299       utf8 = buf;
1300       *fromP += (AS_NORMAL_ENCODING(enc)->type[(unsigned char)**fromP]
1301                  - (BT_LEAD2 - 2));
1302     }
1303     else {
1304       if (n > toLim - *toP)
1305         break;
1306       (*fromP)++;
1307     }
1308     do {
1309       *(*toP)++ = *utf8++;
1310     } while (--n != 0);
1311   }
1312 }
1313
1314 static void PTRCALL
1315 unknown_toUtf16(const ENCODING *enc,
1316                 const char **fromP, const char *fromLim,
1317                 unsigned short **toP, const unsigned short *toLim)
1318 {
1319   const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc);
1320   while (*fromP != fromLim && *toP != toLim) {
1321     unsigned short c = uenc->utf16[(unsigned char)**fromP];
1322     if (c == 0) {
1323       c = (unsigned short)
1324           uenc->convert(uenc->userData, *fromP);
1325       *fromP += (AS_NORMAL_ENCODING(enc)->type[(unsigned char)**fromP]
1326                  - (BT_LEAD2 - 2));
1327     }
1328     else
1329       (*fromP)++;
1330     *(*toP)++ = c;
1331   }
1332 }
1333
1334 ENCODING *
1335 XmlInitUnknownEncoding(void *mem,
1336                        int *table,
1337                        CONVERTER convert,
1338                        void *userData)
1339 {
1340   int i;
1341   struct unknown_encoding *e = (struct unknown_encoding *)mem;
1342   for (i = 0; i < (int)sizeof(struct normal_encoding); i++)
1343     ((char *)mem)[i] = ((char *)&latin1_encoding)[i];
1344   for (i = 0; i < 128; i++)
1345     if (latin1_encoding.type[i] != BT_OTHER
1346         && latin1_encoding.type[i] != BT_NONXML
1347         && table[i] != i)
1348       return 0;
1349   for (i = 0; i < 256; i++) {
1350     int c = table[i];
1351     if (c == -1) {
1352       e->normal.type[i] = BT_MALFORM;
1353       /* This shouldn't really get used. */
1354       e->utf16[i] = 0xFFFF;
1355       e->utf8[i][0] = 1;
1356       e->utf8[i][1] = 0;
1357     }
1358     else if (c < 0) {
1359       if (c < -4)
1360         return 0;
1361       e->normal.type[i] = (unsigned char)(BT_LEAD2 - (c + 2));
1362       e->utf8[i][0] = 0;
1363       e->utf16[i] = 0;
1364     }
1365     else if (c < 0x80) {
1366       if (latin1_encoding.type[c] != BT_OTHER
1367           && latin1_encoding.type[c] != BT_NONXML
1368           && c != i)
1369         return 0;
1370       e->normal.type[i] = latin1_encoding.type[c];
1371       e->utf8[i][0] = 1;
1372       e->utf8[i][1] = (char)c;
1373       e->utf16[i] = (unsigned short)(c == 0 ? 0xFFFF : c);
1374     }
1375     else if (checkCharRefNumber(c) < 0) {
1376       e->normal.type[i] = BT_NONXML;
1377       /* This shouldn't really get used. */
1378       e->utf16[i] = 0xFFFF;
1379       e->utf8[i][0] = 1;
1380       e->utf8[i][1] = 0;
1381     }
1382     else {
1383       if (c > 0xFFFF)
1384         return 0;
1385       if (UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xff))
1386         e->normal.type[i] = BT_NMSTRT;
1387       else if (UCS2_GET_NAMING(namePages, c >> 8, c & 0xff))
1388         e->normal.type[i] = BT_NAME;
1389       else
1390         e->normal.type[i] = BT_OTHER;
1391       e->utf8[i][0] = (char)XmlUtf8Encode(c, e->utf8[i] + 1);
1392       e->utf16[i] = (unsigned short)c;
1393     }
1394   }
1395   e->userData = userData;
1396   e->convert = convert;
1397   if (convert) {
1398     e->normal.isName2 = unknown_isName;
1399     e->normal.isName3 = unknown_isName;
1400     e->normal.isName4 = unknown_isName;
1401     e->normal.isNmstrt2 = unknown_isNmstrt;
1402     e->normal.isNmstrt3 = unknown_isNmstrt;
1403     e->normal.isNmstrt4 = unknown_isNmstrt;
1404     e->normal.isInvalid2 = unknown_isInvalid;
1405     e->normal.isInvalid3 = unknown_isInvalid;
1406     e->normal.isInvalid4 = unknown_isInvalid;
1407   }
1408   e->normal.enc.utf8Convert = unknown_toUtf8;
1409   e->normal.enc.utf16Convert = unknown_toUtf16;
1410   return &(e->normal.enc);
1411 }
1412
1413 /* If this enumeration is changed, getEncodingIndex and encodings
1414 must also be changed. */
1415 enum {
1416   UNKNOWN_ENC = -1,
1417   ISO_8859_1_ENC = 0,
1418   US_ASCII_ENC,
1419   UTF_8_ENC,
1420   UTF_16_ENC,
1421   UTF_16BE_ENC,
1422   UTF_16LE_ENC,
1423   /* must match encodingNames up to here */
1424   NO_ENC
1425 };
1426
1427 static const char KW_ISO_8859_1[] = {
1428   ASCII_I, ASCII_S, ASCII_O, ASCII_MINUS, ASCII_8, ASCII_8, ASCII_5, ASCII_9,
1429   ASCII_MINUS, ASCII_1, '\0'
1430 };
1431 static const char KW_US_ASCII[] = {
1432   ASCII_U, ASCII_S, ASCII_MINUS, ASCII_A, ASCII_S, ASCII_C, ASCII_I, ASCII_I,
1433   '\0'
1434 };
1435 static const char KW_UTF_8[] =  {
1436   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_8, '\0'
1437 };
1438 static const char KW_UTF_16[] = {
1439   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, '\0'
1440 };
1441 static const char KW_UTF_16BE[] = {
1442   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, ASCII_B, ASCII_E,
1443   '\0'
1444 };
1445 static const char KW_UTF_16LE[] = {
1446   ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, ASCII_L, ASCII_E,
1447   '\0'
1448 };
1449
1450 static int FASTCALL
1451 getEncodingIndex(const char *name)
1452 {
1453   static const char *encodingNames[] = {
1454     KW_ISO_8859_1,
1455     KW_US_ASCII,
1456     KW_UTF_8,
1457     KW_UTF_16,
1458     KW_UTF_16BE,
1459     KW_UTF_16LE,
1460   };
1461   int i;
1462   if (name == NULL)
1463     return NO_ENC;
1464   for (i = 0; i < (int)(sizeof(encodingNames)/sizeof(encodingNames[0])); i++)
1465     if (streqci(name, encodingNames[i]))
1466       return i;
1467   return UNKNOWN_ENC;
1468 }
1469
1470 /* For binary compatibility, we store the index of the encoding
1471    specified at initialization in the isUtf16 member.
1472 */
1473
1474 #define INIT_ENC_INDEX(enc) ((int)(enc)->initEnc.isUtf16)
1475 #define SET_INIT_ENC_INDEX(enc, i) ((enc)->initEnc.isUtf16 = (char)i)
1476
1477 /* This is what detects the encoding.  encodingTable maps from
1478    encoding indices to encodings; INIT_ENC_INDEX(enc) is the index of
1479    the external (protocol) specified encoding; state is
1480    XML_CONTENT_STATE if we're parsing an external text entity, and
1481    XML_PROLOG_STATE otherwise.
1482 */
1483
1484
1485 static int
1486 initScan(const ENCODING **encodingTable,
1487          const INIT_ENCODING *enc,
1488          int state,
1489          const char *ptr,
1490          const char *end,
1491          const char **nextTokPtr)
1492 {
1493   const ENCODING **encPtr;
1494
1495   if (ptr == end)
1496     return XML_TOK_NONE;
1497   encPtr = enc->encPtr;
1498   if (ptr + 1 == end) {
1499     /* only a single byte available for auto-detection */
1500 #ifndef XML_DTD /* FIXME */
1501     /* a well-formed document entity must have more than one byte */
1502     if (state != XML_CONTENT_STATE)
1503       return XML_TOK_PARTIAL;
1504 #endif
1505     /* so we're parsing an external text entity... */
1506     /* if UTF-16 was externally specified, then we need at least 2 bytes */
1507     switch (INIT_ENC_INDEX(enc)) {
1508     case UTF_16_ENC:
1509     case UTF_16LE_ENC:
1510     case UTF_16BE_ENC:
1511       return XML_TOK_PARTIAL;
1512     }
1513     switch ((unsigned char)*ptr) {
1514     case 0xFE:
1515     case 0xFF:
1516     case 0xEF: /* possibly first byte of UTF-8 BOM */
1517       if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC
1518           && state == XML_CONTENT_STATE)
1519         break;
1520       /* fall through */
1521     case 0x00:
1522     case 0x3C:
1523       return XML_TOK_PARTIAL;
1524     }
1525   }
1526   else {
1527     switch (((unsigned char)ptr[0] << 8) | (unsigned char)ptr[1]) {
1528     case 0xFEFF:
1529       if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC
1530           && state == XML_CONTENT_STATE)
1531         break;
1532       *nextTokPtr = ptr + 2;
1533       *encPtr = encodingTable[UTF_16BE_ENC];
1534       return XML_TOK_BOM;
1535     /* 00 3C is handled in the default case */
1536     case 0x3C00:
1537       if ((INIT_ENC_INDEX(enc) == UTF_16BE_ENC
1538            || INIT_ENC_INDEX(enc) == UTF_16_ENC)
1539           && state == XML_CONTENT_STATE)
1540         break;
1541       *encPtr = encodingTable[UTF_16LE_ENC];
1542       return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1543     case 0xFFFE:
1544       if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC
1545           && state == XML_CONTENT_STATE)
1546         break;
1547       *nextTokPtr = ptr + 2;
1548       *encPtr = encodingTable[UTF_16LE_ENC];
1549       return XML_TOK_BOM;
1550     case 0xEFBB:
1551       /* Maybe a UTF-8 BOM (EF BB BF) */
1552       /* If there's an explicitly specified (external) encoding
1553          of ISO-8859-1 or some flavour of UTF-16
1554          and this is an external text entity,
1555          don't look for the BOM,
1556          because it might be a legal data.
1557       */
1558       if (state == XML_CONTENT_STATE) {
1559         int e = INIT_ENC_INDEX(enc);
1560         if (e == ISO_8859_1_ENC || e == UTF_16BE_ENC
1561             || e == UTF_16LE_ENC || e == UTF_16_ENC)
1562           break;
1563       }
1564       if (ptr + 2 == end)
1565         return XML_TOK_PARTIAL;
1566       if ((unsigned char)ptr[2] == 0xBF) {
1567         *nextTokPtr = ptr + 3;
1568         *encPtr = encodingTable[UTF_8_ENC];
1569         return XML_TOK_BOM;
1570       }
1571       break;
1572     default:
1573       if (ptr[0] == '\0') {
1574         /* 0 isn't a legal data character. Furthermore a document
1575            entity can only start with ASCII characters.  So the only
1576            way this can fail to be big-endian UTF-16 if it it's an
1577            external parsed general entity that's labelled as
1578            UTF-16LE.
1579         */
1580         if (state == XML_CONTENT_STATE && INIT_ENC_INDEX(enc) == UTF_16LE_ENC)
1581           break;
1582         *encPtr = encodingTable[UTF_16BE_ENC];
1583         return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1584       }
1585       else if (ptr[1] == '\0') {
1586         /* We could recover here in the case:
1587             - parsing an external entity
1588             - second byte is 0
1589             - no externally specified encoding
1590             - no encoding declaration
1591            by assuming UTF-16LE.  But we don't, because this would mean when
1592            presented just with a single byte, we couldn't reliably determine
1593            whether we needed further bytes.
1594         */
1595         if (state == XML_CONTENT_STATE)
1596           break;
1597         *encPtr = encodingTable[UTF_16LE_ENC];
1598         return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1599       }
1600       break;
1601     }
1602   }
1603   *encPtr = encodingTable[INIT_ENC_INDEX(enc)];
1604   return XmlTok(*encPtr, state, ptr, end, nextTokPtr);
1605 }
1606
1607
1608 #define NS(x) x
1609 #define ns(x) x
1610 #include "xmltok_ns.c"
1611 #undef NS
1612 #undef ns
1613
1614 #ifdef XML_NS
1615
1616 #define NS(x) x ## NS
1617 #define ns(x) x ## _ns
1618
1619 #include "xmltok_ns.c"
1620
1621 #undef NS
1622 #undef ns
1623
1624 ENCODING *
1625 XmlInitUnknownEncodingNS(void *mem,
1626                          int *table,
1627                          CONVERTER convert,
1628                          void *userData)
1629 {
1630   ENCODING *enc = XmlInitUnknownEncoding(mem, table, convert, userData);
1631   if (enc)
1632     ((struct normal_encoding *)enc)->type[ASCII_COLON] = BT_COLON;
1633   return enc;
1634 }
1635
1636 #endif /* XML_NS */