1 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd 2 See the file COPYING for copying permission. 3 */ 4 5 #include <stddef.h> 6 7 #ifdef WIN32 8 #include "winconfig.h" 9 #elif defined(MACOS_CLASSIC) 10 #include "macconfig.h" 11 #elif defined(__amigaos__) 12 #include "amigaconfig.h" 13 #elif defined(__WATCOMC__) 14 #include "watcomconfig.h" 15 #else 16 #ifdef HAVE_EXPAT_CONFIG_H 17 #include <expat_config.h> 18 #endif 19 #endif /* ndef WIN32 */ 20 21 #include "expat_external.h" 22 #include "internal.h" 23 #include "xmltok.h" 24 #include "nametab.h" 25 26 #ifdef XML_DTD 27 #define IGNORE_SECTION_TOK_VTABLE , PREFIX(ignoreSectionTok) 28 #else 29 #define IGNORE_SECTION_TOK_VTABLE /* as nothing */ 30 #endif 31 32 #define VTABLE1 \ 33 { PREFIX(prologTok), PREFIX(contentTok), \ 34 PREFIX(cdataSectionTok) IGNORE_SECTION_TOK_VTABLE }, \ 35 { PREFIX(attributeValueTok), PREFIX(entityValueTok) }, \ 36 PREFIX(sameName), \ 37 PREFIX(nameMatchesAscii), \ 38 PREFIX(nameLength), \ 39 PREFIX(skipS), \ 40 PREFIX(getAtts), \ 41 PREFIX(charRefNumber), \ 42 PREFIX(predefinedEntityName), \ 43 PREFIX(updatePosition), \ 44 PREFIX(isPublicId) 45 46 #define VTABLE VTABLE1, PREFIX(toUtf8), PREFIX(toUtf16) 47 48 #define UCS2_GET_NAMING(pages, hi, lo) \ 49 (namingBitmap[(pages[hi] << 3) + ((lo) >> 5)] & (1u << ((lo) & 0x1F))) 50 51 /* A 2 byte UTF-8 representation splits the characters 11 bits between 52 the bottom 5 and 6 bits of the bytes. We need 8 bits to index into 53 pages, 3 bits to add to that index and 5 bits to generate the mask. 54 */ 55 #define UTF8_GET_NAMING2(pages, byte) \ 56 (namingBitmap[((pages)[(((byte)[0]) >> 2) & 7] << 3) \ 57 + ((((byte)[0]) & 3) << 1) \ 58 + ((((byte)[1]) >> 5) & 1)] \ 59 & (1u << (((byte)[1]) & 0x1F))) 60 61 /* A 3 byte UTF-8 representation splits the characters 16 bits between 62 the bottom 4, 6 and 6 bits of the bytes. We need 8 bits to index 63 into pages, 3 bits to add to that index and 5 bits to generate the 64 mask. 65 */ 66 #define UTF8_GET_NAMING3(pages, byte) \ 67 (namingBitmap[((pages)[((((byte)[0]) & 0xF) << 4) \ 68 + ((((byte)[1]) >> 2) & 0xF)] \ 69 << 3) \ 70 + ((((byte)[1]) & 3) << 1) \ 71 + ((((byte)[2]) >> 5) & 1)] \ 72 & (1u << (((byte)[2]) & 0x1F))) 73 74 #define UTF8_GET_NAMING(pages, p, n) \ 75 ((n) == 2 \ 76 ? UTF8_GET_NAMING2(pages, (const unsigned char *)(p)) \ 77 : ((n) == 3 \ 78 ? UTF8_GET_NAMING3(pages, (const unsigned char *)(p)) \ 79 : 0)) 80 81 /* Detection of invalid UTF-8 sequences is based on Table 3.1B 82 of Unicode 3.2: http://www.unicode.org/unicode/reports/tr28/ 83 with the additional restriction of not allowing the Unicode 84 code points 0xFFFF and 0xFFFE (sequences EF,BF,BF and EF,BF,BE). 85 Implementation details: 86 (A & 0x80) == 0 means A < 0x80 87 and 88 (A & 0xC0) == 0xC0 means A > 0xBF 89 */ 90 91 #define UTF8_INVALID2(p) \ 92 ((*p) < 0xC2 || ((p)[1] & 0x80) == 0 || ((p)[1] & 0xC0) == 0xC0) 93 94 #define UTF8_INVALID3(p) \ 95 (((p)[2] & 0x80) == 0 \ 96 || \ 97 ((*p) == 0xEF && (p)[1] == 0xBF \ 98 ? \ 99 (p)[2] > 0xBD \ 100 : \ 101 ((p)[2] & 0xC0) == 0xC0) \ 102 || \ 103 ((*p) == 0xE0 \ 104 ? \ 105 (p)[1] < 0xA0 || ((p)[1] & 0xC0) == 0xC0 \ 106 : \ 107 ((p)[1] & 0x80) == 0 \ 108 || \ 109 ((*p) == 0xED ? (p)[1] > 0x9F : ((p)[1] & 0xC0) == 0xC0))) 110 111 #define UTF8_INVALID4(p) \ 112 (((p)[3] & 0x80) == 0 || ((p)[3] & 0xC0) == 0xC0 \ 113 || \ 114 ((p)[2] & 0x80) == 0 || ((p)[2] & 0xC0) == 0xC0 \ 115 || \ 116 ((*p) == 0xF0 \ 117 ? \ 118 (p)[1] < 0x90 || ((p)[1] & 0xC0) == 0xC0 \ 119 : \ 120 ((p)[1] & 0x80) == 0 \ 121 || \ 122 ((*p) == 0xF4 ? (p)[1] > 0x8F : ((p)[1] & 0xC0) == 0xC0))) 123 124 static int PTRFASTCALL 125 isNever(const ENCODING *UNUSED_P(enc), const char *UNUSED_P(p)) 126 { 127 return 0; 128 } 129 130 static int PTRFASTCALL 131 utf8_isName2(const ENCODING *UNUSED_P(enc), const char *p) 132 { 133 return UTF8_GET_NAMING2(namePages, (const unsigned char *)p); 134 } 135 136 static int PTRFASTCALL 137 utf8_isName3(const ENCODING *UNUSED_P(enc), const char *p) 138 { 139 return UTF8_GET_NAMING3(namePages, (const unsigned char *)p); 140 } 141 142 #define utf8_isName4 isNever 143 144 static int PTRFASTCALL 145 utf8_isNmstrt2(const ENCODING *UNUSED_P(enc), const char *p) 146 { 147 return UTF8_GET_NAMING2(nmstrtPages, (const unsigned char *)p); 148 } 149 150 static int PTRFASTCALL 151 utf8_isNmstrt3(const ENCODING *UNUSED_P(enc), const char *p) 152 { 153 return UTF8_GET_NAMING3(nmstrtPages, (const unsigned char *)p); 154 } 155 156 #define utf8_isNmstrt4 isNever 157 158 static int PTRFASTCALL 159 utf8_isInvalid2(const ENCODING *UNUSED_P(enc), const char *p) 160 { 161 return UTF8_INVALID2((const unsigned char *)p); 162 } 163 164 static int PTRFASTCALL 165 utf8_isInvalid3(const ENCODING *UNUSED_P(enc), const char *p) 166 { 167 return UTF8_INVALID3((const unsigned char *)p); 168 } 169 170 static int PTRFASTCALL 171 utf8_isInvalid4(const ENCODING *UNUSED_P(enc), const char *p) 172 { 173 return UTF8_INVALID4((const unsigned char *)p); 174 } 175 176 struct normal_encoding { 177 ENCODING enc; 178 unsigned char type[256]; 179 #ifdef XML_MIN_SIZE 180 int (PTRFASTCALL *byteType)(const ENCODING *, const char *); 181 int (PTRFASTCALL *isNameMin)(const ENCODING *, const char *); 182 int (PTRFASTCALL *isNmstrtMin)(const ENCODING *, const char *); 183 int (PTRFASTCALL *byteToAscii)(const ENCODING *, const char *); 184 int (PTRCALL *charMatches)(const ENCODING *, const char *, int); 185 #endif /* XML_MIN_SIZE */ 186 int (PTRFASTCALL *isName2)(const ENCODING *, const char *); 187 int (PTRFASTCALL *isName3)(const ENCODING *, const char *); 188 int (PTRFASTCALL *isName4)(const ENCODING *, const char *); 189 int (PTRFASTCALL *isNmstrt2)(const ENCODING *, const char *); 190 int (PTRFASTCALL *isNmstrt3)(const ENCODING *, const char *); 191 int (PTRFASTCALL *isNmstrt4)(const ENCODING *, const char *); 192 int (PTRFASTCALL *isInvalid2)(const ENCODING *, const char *); 193 int (PTRFASTCALL *isInvalid3)(const ENCODING *, const char *); 194 int (PTRFASTCALL *isInvalid4)(const ENCODING *, const char *); 195 }; 196 197 #define AS_NORMAL_ENCODING(enc) ((const struct normal_encoding *) (enc)) 198 199 #ifdef XML_MIN_SIZE 200 201 #define STANDARD_VTABLE(E) \ 202 E ## byteType, \ 203 E ## isNameMin, \ 204 E ## isNmstrtMin, \ 205 E ## byteToAscii, \ 206 E ## charMatches, 207 208 #else 209 210 #define STANDARD_VTABLE(E) /* as nothing */ 211 212 #endif 213 214 #define NORMAL_VTABLE(E) \ 215 E ## isName2, \ 216 E ## isName3, \ 217 E ## isName4, \ 218 E ## isNmstrt2, \ 219 E ## isNmstrt3, \ 220 E ## isNmstrt4, \ 221 E ## isInvalid2, \ 222 E ## isInvalid3, \ 223 E ## isInvalid4 224 225 #define NULL_VTABLE \ 226 /* isName2 */ NULL, \ 227 /* isName3 */ NULL, \ 228 /* isName4 */ NULL, \ 229 /* isNmstrt2 */ NULL, \ 230 /* isNmstrt3 */ NULL, \ 231 /* isNmstrt4 */ NULL, \ 232 /* isInvalid2 */ NULL, \ 233 /* isInvalid3 */ NULL, \ 234 /* isInvalid4 */ NULL 235 236 static int FASTCALL checkCharRefNumber(int); 237 238 #include "xmltok_impl.h" 239 #include "ascii.h" 240 241 #ifdef XML_MIN_SIZE 242 #define sb_isNameMin isNever 243 #define sb_isNmstrtMin isNever 244 #endif 245 246 #ifdef XML_MIN_SIZE 247 #define MINBPC(enc) ((enc)->minBytesPerChar) 248 #else 249 /* minimum bytes per character */ 250 #define MINBPC(enc) 1 251 #endif 252 253 #define SB_BYTE_TYPE(enc, p) \ 254 (((struct normal_encoding *)(enc))->type[(unsigned char)*(p)]) 255 256 #ifdef XML_MIN_SIZE 257 static int PTRFASTCALL 258 sb_byteType(const ENCODING *enc, const char *p) 259 { 260 return SB_BYTE_TYPE(enc, p); 261 } 262 #define BYTE_TYPE(enc, p) \ 263 (AS_NORMAL_ENCODING(enc)->byteType(enc, p)) 264 #else 265 #define BYTE_TYPE(enc, p) SB_BYTE_TYPE(enc, p) 266 #endif 267 268 #ifdef XML_MIN_SIZE 269 #define BYTE_TO_ASCII(enc, p) \ 270 (AS_NORMAL_ENCODING(enc)->byteToAscii(enc, p)) 271 static int PTRFASTCALL 272 sb_byteToAscii(const ENCODING *enc, const char *p) 273 { 274 return *p; 275 } 276 #else 277 #define BYTE_TO_ASCII(enc, p) (*(p)) 278 #endif 279 280 #define IS_NAME_CHAR(enc, p, n) \ 281 (AS_NORMAL_ENCODING(enc)->isName ## n(enc, p)) 282 #define IS_NMSTRT_CHAR(enc, p, n) \ 283 (AS_NORMAL_ENCODING(enc)->isNmstrt ## n(enc, p)) 284 #define IS_INVALID_CHAR(enc, p, n) \ 285 (AS_NORMAL_ENCODING(enc)->isInvalid ## n(enc, p)) 286 287 #ifdef XML_MIN_SIZE 288 #define IS_NAME_CHAR_MINBPC(enc, p) \ 289 (AS_NORMAL_ENCODING(enc)->isNameMin(enc, p)) 290 #define IS_NMSTRT_CHAR_MINBPC(enc, p) \ 291 (AS_NORMAL_ENCODING(enc)->isNmstrtMin(enc, p)) 292 #else 293 #define IS_NAME_CHAR_MINBPC(enc, p) (0) 294 #define IS_NMSTRT_CHAR_MINBPC(enc, p) (0) 295 #endif 296 297 #ifdef XML_MIN_SIZE 298 #define CHAR_MATCHES(enc, p, c) \ 299 (AS_NORMAL_ENCODING(enc)->charMatches(enc, p, c)) 300 static int PTRCALL 301 sb_charMatches(const ENCODING *enc, const char *p, int c) 302 { 303 return *p == c; 304 } 305 #else 306 /* c is an ASCII character */ 307 #define CHAR_MATCHES(enc, p, c) (*(p) == c) 308 #endif 309 310 #define PREFIX(ident) normal_ ## ident 311 #define XML_TOK_IMPL_C 312 #include "xmltok_impl.c" 313 #undef XML_TOK_IMPL_C 314 315 #undef MINBPC 316 #undef BYTE_TYPE 317 #undef BYTE_TO_ASCII 318 #undef CHAR_MATCHES 319 #undef IS_NAME_CHAR 320 #undef IS_NAME_CHAR_MINBPC 321 #undef IS_NMSTRT_CHAR 322 #undef IS_NMSTRT_CHAR_MINBPC 323 #undef IS_INVALID_CHAR 324 325 enum { /* UTF8_cvalN is value of masked first byte of N byte sequence */ 326 UTF8_cval1 = 0x00, 327 UTF8_cval2 = 0xc0, 328 UTF8_cval3 = 0xe0, 329 UTF8_cval4 = 0xf0 330 }; 331 332 void 333 align_limit_to_full_utf8_characters(const char * from, const char ** fromLimRef) 334 { 335 const char * fromLim = *fromLimRef; 336 size_t walked = 0; 337 for (; fromLim > from; fromLim--, walked++) { 338 const unsigned char prev = (unsigned char)fromLim[-1]; 339 if ((prev & 0xf8u) == 0xf0u) { /* 4-byte character, lead by 0b11110xxx byte */ 340 if (walked + 1 >= 4) { 341 fromLim += 4 - 1; 342 break; 343 } else { 344 walked = 0; 345 } 346 } else if ((prev & 0xf0u) == 0xe0u) { /* 3-byte character, lead by 0b1110xxxx byte */ 347 if (walked + 1 >= 3) { 348 fromLim += 3 - 1; 349 break; 350 } else { 351 walked = 0; 352 } 353 } else if ((prev & 0xe0u) == 0xc0u) { /* 2-byte character, lead by 0b110xxxxx byte */ 354 if (walked + 1 >= 2) { 355 fromLim += 2 - 1; 356 break; 357 } else { 358 walked = 0; 359 } 360 } else if ((prev & 0x80u) == 0x00u) { /* 1-byte character, matching 0b0xxxxxxx */ 361 break; 362 } 363 } 364 *fromLimRef = fromLim; 365 } 366 367 static enum XML_Convert_Result PTRCALL 368 utf8_toUtf8(const ENCODING *UNUSED_P(enc), 369 const char **fromP, const char *fromLim, 370 char **toP, const char *toLim) 371 { 372 enum XML_Convert_Result res = XML_CONVERT_COMPLETED; 373 char *to; 374 const char *from; 375 if (fromLim - *fromP > toLim - *toP) { 376 /* Avoid copying partial characters. */ 377 res = XML_CONVERT_OUTPUT_EXHAUSTED; 378 fromLim = *fromP + (toLim - *toP); 379 align_limit_to_full_utf8_characters(*fromP, &fromLim); 380 } 381 for (to = *toP, from = *fromP; (from < fromLim) && (to < toLim); from++, to++) 382 *to = *from; 383 *fromP = from; 384 *toP = to; 385 386 if ((to == toLim) && (from < fromLim)) 387 return XML_CONVERT_OUTPUT_EXHAUSTED; 388 else 389 return res; 390 } 391 392 static enum XML_Convert_Result PTRCALL 393 utf8_toUtf16(const ENCODING *enc, 394 const char **fromP, const char *fromLim, 395 unsigned short **toP, const unsigned short *toLim) 396 { 397 enum XML_Convert_Result res = XML_CONVERT_COMPLETED; 398 unsigned short *to = *toP; 399 const char *from = *fromP; 400 while (from < fromLim && to < toLim) { 401 switch (((struct normal_encoding *)enc)->type[(unsigned char)*from]) { 402 case BT_LEAD2: 403 if (fromLim - from < 2) { 404 res = XML_CONVERT_INPUT_INCOMPLETE; 405 break; 406 } 407 *to++ = (unsigned short)(((from[0] & 0x1f) << 6) | (from[1] & 0x3f)); 408 from += 2; 409 break; 410 case BT_LEAD3: 411 if (fromLim - from < 3) { 412 res = XML_CONVERT_INPUT_INCOMPLETE; 413 break; 414 } 415 *to++ = (unsigned short)(((from[0] & 0xf) << 12) 416 | ((from[1] & 0x3f) << 6) | (from[2] & 0x3f)); 417 from += 3; 418 break; 419 case BT_LEAD4: 420 { 421 unsigned long n; 422 if (toLim - to < 2) { 423 res = XML_CONVERT_OUTPUT_EXHAUSTED; 424 goto after; 425 } 426 if (fromLim - from < 4) { 427 res = XML_CONVERT_INPUT_INCOMPLETE; 428 goto after; 429 } 430 n = ((from[0] & 0x7) << 18) | ((from[1] & 0x3f) << 12) 431 | ((from[2] & 0x3f) << 6) | (from[3] & 0x3f); 432 n -= 0x10000; 433 to[0] = (unsigned short)((n >> 10) | 0xD800); 434 to[1] = (unsigned short)((n & 0x3FF) | 0xDC00); 435 to += 2; 436 from += 4; 437 } 438 break; 439 default: 440 *to++ = *from++; 441 break; 442 } 443 } 444 after: 445 *fromP = from; 446 *toP = to; 447 return res; 448 } 449 450 #ifdef XML_NS 451 static const struct normal_encoding utf8_encoding_ns = { 452 { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 }, 453 { 454 #include "asciitab.h" 455 #include "utf8tab.h" 456 }, 457 STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_) 458 }; 459 #endif 460 461 static const struct normal_encoding utf8_encoding = { 462 { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 }, 463 { 464 #define BT_COLON BT_NMSTRT 465 #include "asciitab.h" 466 #undef BT_COLON 467 #include "utf8tab.h" 468 }, 469 STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_) 470 }; 471 472 #ifdef XML_NS 473 474 static const struct normal_encoding internal_utf8_encoding_ns = { 475 { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 }, 476 { 477 #include "iasciitab.h" 478 #include "utf8tab.h" 479 }, 480 STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_) 481 }; 482 483 #endif 484 485 static const struct normal_encoding internal_utf8_encoding = { 486 { VTABLE1, utf8_toUtf8, utf8_toUtf16, 1, 1, 0 }, 487 { 488 #define BT_COLON BT_NMSTRT 489 #include "iasciitab.h" 490 #undef BT_COLON 491 #include "utf8tab.h" 492 }, 493 STANDARD_VTABLE(sb_) NORMAL_VTABLE(utf8_) 494 }; 495 496 static enum XML_Convert_Result PTRCALL 497 latin1_toUtf8(const ENCODING *UNUSED_P(enc), 498 const char **fromP, const char *fromLim, 499 char **toP, const char *toLim) 500 { 501 for (;;) { 502 unsigned char c; 503 if (*fromP == fromLim) 504 return XML_CONVERT_COMPLETED; 505 c = (unsigned char)**fromP; 506 if (c & 0x80) { 507 if (toLim - *toP < 2) 508 return XML_CONVERT_OUTPUT_EXHAUSTED; 509 *(*toP)++ = (char)((c >> 6) | UTF8_cval2); 510 *(*toP)++ = (char)((c & 0x3f) | 0x80); 511 (*fromP)++; 512 } 513 else { 514 if (*toP == toLim) 515 return XML_CONVERT_OUTPUT_EXHAUSTED; 516 *(*toP)++ = *(*fromP)++; 517 } 518 } 519 } 520 521 static enum XML_Convert_Result PTRCALL 522 latin1_toUtf16(const ENCODING *UNUSED_P(enc), 523 const char **fromP, const char *fromLim, 524 unsigned short **toP, const unsigned short *toLim) 525 { 526 while (*fromP < fromLim && *toP < toLim) 527 *(*toP)++ = (unsigned char)*(*fromP)++; 528 529 if ((*toP == toLim) && (*fromP < fromLim)) 530 return XML_CONVERT_OUTPUT_EXHAUSTED; 531 else 532 return XML_CONVERT_COMPLETED; 533 } 534 535 #ifdef XML_NS 536 537 static const struct normal_encoding latin1_encoding_ns = { 538 { VTABLE1, latin1_toUtf8, latin1_toUtf16, 1, 0, 0 }, 539 { 540 #include "asciitab.h" 541 #include "latin1tab.h" 542 }, 543 STANDARD_VTABLE(sb_) NULL_VTABLE 544 }; 545 546 #endif 547 548 static const struct normal_encoding latin1_encoding = { 549 { VTABLE1, latin1_toUtf8, latin1_toUtf16, 1, 0, 0 }, 550 { 551 #define BT_COLON BT_NMSTRT 552 #include "asciitab.h" 553 #undef BT_COLON 554 #include "latin1tab.h" 555 }, 556 STANDARD_VTABLE(sb_) NULL_VTABLE 557 }; 558 559 static enum XML_Convert_Result PTRCALL 560 ascii_toUtf8(const ENCODING *UNUSED_P(enc), 561 const char **fromP, const char *fromLim, 562 char **toP, const char *toLim) 563 { 564 while (*fromP < fromLim && *toP < toLim) 565 *(*toP)++ = *(*fromP)++; 566 567 if ((*toP == toLim) && (*fromP < fromLim)) 568 return XML_CONVERT_OUTPUT_EXHAUSTED; 569 else 570 return XML_CONVERT_COMPLETED; 571 } 572 573 #ifdef XML_NS 574 575 static const struct normal_encoding ascii_encoding_ns = { 576 { VTABLE1, ascii_toUtf8, latin1_toUtf16, 1, 1, 0 }, 577 { 578 #include "asciitab.h" 579 /* BT_NONXML == 0 */ 580 }, 581 STANDARD_VTABLE(sb_) NULL_VTABLE 582 }; 583 584 #endif 585 586 static const struct normal_encoding ascii_encoding = { 587 { VTABLE1, ascii_toUtf8, latin1_toUtf16, 1, 1, 0 }, 588 { 589 #define BT_COLON BT_NMSTRT 590 #include "asciitab.h" 591 #undef BT_COLON 592 /* BT_NONXML == 0 */ 593 }, 594 STANDARD_VTABLE(sb_) NULL_VTABLE 595 }; 596 597 static int PTRFASTCALL 598 unicode_byte_type(char hi, char lo) 599 { 600 switch ((unsigned char)hi) { 601 case 0xD8: case 0xD9: case 0xDA: case 0xDB: 602 return BT_LEAD4; 603 case 0xDC: case 0xDD: case 0xDE: case 0xDF: 604 return BT_TRAIL; 605 case 0xFF: 606 switch ((unsigned char)lo) { 607 case 0xFF: 608 case 0xFE: 609 return BT_NONXML; 610 } 611 break; 612 } 613 return BT_NONASCII; 614 } 615 616 #define DEFINE_UTF16_TO_UTF8(E) \ 617 static enum XML_Convert_Result PTRCALL \ 618 E ## toUtf8(const ENCODING *UNUSED_P(enc), \ 619 const char **fromP, const char *fromLim, \ 620 char **toP, const char *toLim) \ 621 { \ 622 const char *from = *fromP; \ 623 fromLim = from + (((fromLim - from) >> 1) << 1); /* shrink to even */ \ 624 for (; from < fromLim; from += 2) { \ 625 int plane; \ 626 unsigned char lo2; \ 627 unsigned char lo = GET_LO(from); \ 628 unsigned char hi = GET_HI(from); \ 629 switch (hi) { \ 630 case 0: \ 631 if (lo < 0x80) { \ 632 if (*toP == toLim) { \ 633 *fromP = from; \ 634 return XML_CONVERT_OUTPUT_EXHAUSTED; \ 635 } \ 636 *(*toP)++ = lo; \ 637 break; \ 638 } \ 639 /* fall through */ \ 640 case 0x1: case 0x2: case 0x3: \ 641 case 0x4: case 0x5: case 0x6: case 0x7: \ 642 if (toLim - *toP < 2) { \ 643 *fromP = from; \ 644 return XML_CONVERT_OUTPUT_EXHAUSTED; \ 645 } \ 646 *(*toP)++ = ((lo >> 6) | (hi << 2) | UTF8_cval2); \ 647 *(*toP)++ = ((lo & 0x3f) | 0x80); \ 648 break; \ 649 default: \ 650 if (toLim - *toP < 3) { \ 651 *fromP = from; \ 652 return XML_CONVERT_OUTPUT_EXHAUSTED; \ 653 } \ 654 /* 16 bits divided 4, 6, 6 amongst 3 bytes */ \ 655 *(*toP)++ = ((hi >> 4) | UTF8_cval3); \ 656 *(*toP)++ = (((hi & 0xf) << 2) | (lo >> 6) | 0x80); \ 657 *(*toP)++ = ((lo & 0x3f) | 0x80); \ 658 break; \ 659 case 0xD8: case 0xD9: case 0xDA: case 0xDB: \ 660 if (toLim - *toP < 4) { \ 661 *fromP = from; \ 662 return XML_CONVERT_OUTPUT_EXHAUSTED; \ 663 } \ 664 if (fromLim - from < 4) { \ 665 *fromP = from; \ 666 return XML_CONVERT_INPUT_INCOMPLETE; \ 667 } \ 668 plane = (((hi & 0x3) << 2) | ((lo >> 6) & 0x3)) + 1; \ 669 *(*toP)++ = ((plane >> 2) | UTF8_cval4); \ 670 *(*toP)++ = (((lo >> 2) & 0xF) | ((plane & 0x3) << 4) | 0x80); \ 671 from += 2; \ 672 lo2 = GET_LO(from); \ 673 *(*toP)++ = (((lo & 0x3) << 4) \ 674 | ((GET_HI(from) & 0x3) << 2) \ 675 | (lo2 >> 6) \ 676 | 0x80); \ 677 *(*toP)++ = ((lo2 & 0x3f) | 0x80); \ 678 break; \ 679 } \ 680 } \ 681 *fromP = from; \ 682 if (from < fromLim) \ 683 return XML_CONVERT_INPUT_INCOMPLETE; \ 684 else \ 685 return XML_CONVERT_COMPLETED; \ 686 } 687 688 #define DEFINE_UTF16_TO_UTF16(E) \ 689 static enum XML_Convert_Result PTRCALL \ 690 E ## toUtf16(const ENCODING *UNUSED_P(enc), \ 691 const char **fromP, const char *fromLim, \ 692 unsigned short **toP, const unsigned short *toLim) \ 693 { \ 694 enum XML_Convert_Result res = XML_CONVERT_COMPLETED; \ 695 fromLim = *fromP + (((fromLim - *fromP) >> 1) << 1); /* shrink to even */ \ 696 /* Avoid copying first half only of surrogate */ \ 697 if (fromLim - *fromP > ((toLim - *toP) << 1) \ 698 && (GET_HI(fromLim - 2) & 0xF8) == 0xD8) { \ 699 fromLim -= 2; \ 700 res = XML_CONVERT_INPUT_INCOMPLETE; \ 701 } \ 702 for (; *fromP < fromLim && *toP < toLim; *fromP += 2) \ 703 *(*toP)++ = (GET_HI(*fromP) << 8) | GET_LO(*fromP); \ 704 if ((*toP == toLim) && (*fromP < fromLim)) \ 705 return XML_CONVERT_OUTPUT_EXHAUSTED; \ 706 else \ 707 return res; \ 708 } 709 710 #define SET2(ptr, ch) \ 711 (((ptr)[0] = ((ch) & 0xff)), ((ptr)[1] = ((ch) >> 8))) 712 #define GET_LO(ptr) ((unsigned char)(ptr)[0]) 713 #define GET_HI(ptr) ((unsigned char)(ptr)[1]) 714 715 DEFINE_UTF16_TO_UTF8(little2_) 716 DEFINE_UTF16_TO_UTF16(little2_) 717 718 #undef SET2 719 #undef GET_LO 720 #undef GET_HI 721 722 #define SET2(ptr, ch) \ 723 (((ptr)[0] = ((ch) >> 8)), ((ptr)[1] = ((ch) & 0xFF))) 724 #define GET_LO(ptr) ((unsigned char)(ptr)[1]) 725 #define GET_HI(ptr) ((unsigned char)(ptr)[0]) 726 727 DEFINE_UTF16_TO_UTF8(big2_) 728 DEFINE_UTF16_TO_UTF16(big2_) 729 730 #undef SET2 731 #undef GET_LO 732 #undef GET_HI 733 734 #define LITTLE2_BYTE_TYPE(enc, p) \ 735 ((p)[1] == 0 \ 736 ? ((struct normal_encoding *)(enc))->type[(unsigned char)*(p)] \ 737 : unicode_byte_type((p)[1], (p)[0])) 738 #define LITTLE2_BYTE_TO_ASCII(enc, p) ((p)[1] == 0 ? (p)[0] : -1) 739 #define LITTLE2_CHAR_MATCHES(enc, p, c) ((p)[1] == 0 && (p)[0] == c) 740 #define LITTLE2_IS_NAME_CHAR_MINBPC(enc, p) \ 741 UCS2_GET_NAMING(namePages, (unsigned char)p[1], (unsigned char)p[0]) 742 #define LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p) \ 743 UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[1], (unsigned char)p[0]) 744 745 #ifdef XML_MIN_SIZE 746 747 static int PTRFASTCALL 748 little2_byteType(const ENCODING *enc, const char *p) 749 { 750 return LITTLE2_BYTE_TYPE(enc, p); 751 } 752 753 static int PTRFASTCALL 754 little2_byteToAscii(const ENCODING *enc, const char *p) 755 { 756 return LITTLE2_BYTE_TO_ASCII(enc, p); 757 } 758 759 static int PTRCALL 760 little2_charMatches(const ENCODING *enc, const char *p, int c) 761 { 762 return LITTLE2_CHAR_MATCHES(enc, p, c); 763 } 764 765 static int PTRFASTCALL 766 little2_isNameMin(const ENCODING *enc, const char *p) 767 { 768 return LITTLE2_IS_NAME_CHAR_MINBPC(enc, p); 769 } 770 771 static int PTRFASTCALL 772 little2_isNmstrtMin(const ENCODING *enc, const char *p) 773 { 774 return LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p); 775 } 776 777 #undef VTABLE 778 #define VTABLE VTABLE1, little2_toUtf8, little2_toUtf16 779 780 #else /* not XML_MIN_SIZE */ 781 782 #undef PREFIX 783 #define PREFIX(ident) little2_ ## ident 784 #define MINBPC(enc) 2 785 /* CHAR_MATCHES is guaranteed to have MINBPC bytes available. */ 786 #define BYTE_TYPE(enc, p) LITTLE2_BYTE_TYPE(enc, p) 787 #define BYTE_TO_ASCII(enc, p) LITTLE2_BYTE_TO_ASCII(enc, p) 788 #define CHAR_MATCHES(enc, p, c) LITTLE2_CHAR_MATCHES(enc, p, c) 789 #define IS_NAME_CHAR(enc, p, n) 0 790 #define IS_NAME_CHAR_MINBPC(enc, p) LITTLE2_IS_NAME_CHAR_MINBPC(enc, p) 791 #define IS_NMSTRT_CHAR(enc, p, n) (0) 792 #define IS_NMSTRT_CHAR_MINBPC(enc, p) LITTLE2_IS_NMSTRT_CHAR_MINBPC(enc, p) 793 794 #define XML_TOK_IMPL_C 795 #include "xmltok_impl.c" 796 #undef XML_TOK_IMPL_C 797 798 #undef MINBPC 799 #undef BYTE_TYPE 800 #undef BYTE_TO_ASCII 801 #undef CHAR_MATCHES 802 #undef IS_NAME_CHAR 803 #undef IS_NAME_CHAR_MINBPC 804 #undef IS_NMSTRT_CHAR 805 #undef IS_NMSTRT_CHAR_MINBPC 806 #undef IS_INVALID_CHAR 807 808 #endif /* not XML_MIN_SIZE */ 809 810 #ifdef XML_NS 811 812 static const struct normal_encoding little2_encoding_ns = { 813 { VTABLE, 2, 0, 814 #if BYTEORDER == 1234 815 1 816 #else 817 0 818 #endif 819 }, 820 { 821 #include "asciitab.h" 822 #include "latin1tab.h" 823 }, 824 STANDARD_VTABLE(little2_) NULL_VTABLE 825 }; 826 827 #endif 828 829 static const struct normal_encoding little2_encoding = { 830 { VTABLE, 2, 0, 831 #if BYTEORDER == 1234 832 1 833 #else 834 0 835 #endif 836 }, 837 { 838 #define BT_COLON BT_NMSTRT 839 #include "asciitab.h" 840 #undef BT_COLON 841 #include "latin1tab.h" 842 }, 843 STANDARD_VTABLE(little2_) NULL_VTABLE 844 }; 845 846 #if BYTEORDER != 4321 847 848 #ifdef XML_NS 849 850 static const struct normal_encoding internal_little2_encoding_ns = { 851 { VTABLE, 2, 0, 1 }, 852 { 853 #include "iasciitab.h" 854 #include "latin1tab.h" 855 }, 856 STANDARD_VTABLE(little2_) NULL_VTABLE 857 }; 858 859 #endif 860 861 static const struct normal_encoding internal_little2_encoding = { 862 { VTABLE, 2, 0, 1 }, 863 { 864 #define BT_COLON BT_NMSTRT 865 #include "iasciitab.h" 866 #undef BT_COLON 867 #include "latin1tab.h" 868 }, 869 STANDARD_VTABLE(little2_) NULL_VTABLE 870 }; 871 872 #endif 873 874 875 #define BIG2_BYTE_TYPE(enc, p) \ 876 ((p)[0] == 0 \ 877 ? ((struct normal_encoding *)(enc))->type[(unsigned char)(p)[1]] \ 878 : unicode_byte_type((p)[0], (p)[1])) 879 #define BIG2_BYTE_TO_ASCII(enc, p) ((p)[0] == 0 ? (p)[1] : -1) 880 #define BIG2_CHAR_MATCHES(enc, p, c) ((p)[0] == 0 && (p)[1] == c) 881 #define BIG2_IS_NAME_CHAR_MINBPC(enc, p) \ 882 UCS2_GET_NAMING(namePages, (unsigned char)p[0], (unsigned char)p[1]) 883 #define BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p) \ 884 UCS2_GET_NAMING(nmstrtPages, (unsigned char)p[0], (unsigned char)p[1]) 885 886 #ifdef XML_MIN_SIZE 887 888 static int PTRFASTCALL 889 big2_byteType(const ENCODING *enc, const char *p) 890 { 891 return BIG2_BYTE_TYPE(enc, p); 892 } 893 894 static int PTRFASTCALL 895 big2_byteToAscii(const ENCODING *enc, const char *p) 896 { 897 return BIG2_BYTE_TO_ASCII(enc, p); 898 } 899 900 static int PTRCALL 901 big2_charMatches(const ENCODING *enc, const char *p, int c) 902 { 903 return BIG2_CHAR_MATCHES(enc, p, c); 904 } 905 906 static int PTRFASTCALL 907 big2_isNameMin(const ENCODING *enc, const char *p) 908 { 909 return BIG2_IS_NAME_CHAR_MINBPC(enc, p); 910 } 911 912 static int PTRFASTCALL 913 big2_isNmstrtMin(const ENCODING *enc, const char *p) 914 { 915 return BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p); 916 } 917 918 #undef VTABLE 919 #define VTABLE VTABLE1, big2_toUtf8, big2_toUtf16 920 921 #else /* not XML_MIN_SIZE */ 922 923 #undef PREFIX 924 #define PREFIX(ident) big2_ ## ident 925 #define MINBPC(enc) 2 926 /* CHAR_MATCHES is guaranteed to have MINBPC bytes available. */ 927 #define BYTE_TYPE(enc, p) BIG2_BYTE_TYPE(enc, p) 928 #define BYTE_TO_ASCII(enc, p) BIG2_BYTE_TO_ASCII(enc, p) 929 #define CHAR_MATCHES(enc, p, c) BIG2_CHAR_MATCHES(enc, p, c) 930 #define IS_NAME_CHAR(enc, p, n) 0 931 #define IS_NAME_CHAR_MINBPC(enc, p) BIG2_IS_NAME_CHAR_MINBPC(enc, p) 932 #define IS_NMSTRT_CHAR(enc, p, n) (0) 933 #define IS_NMSTRT_CHAR_MINBPC(enc, p) BIG2_IS_NMSTRT_CHAR_MINBPC(enc, p) 934 935 #define XML_TOK_IMPL_C 936 #include "xmltok_impl.c" 937 #undef XML_TOK_IMPL_C 938 939 #undef MINBPC 940 #undef BYTE_TYPE 941 #undef BYTE_TO_ASCII 942 #undef CHAR_MATCHES 943 #undef IS_NAME_CHAR 944 #undef IS_NAME_CHAR_MINBPC 945 #undef IS_NMSTRT_CHAR 946 #undef IS_NMSTRT_CHAR_MINBPC 947 #undef IS_INVALID_CHAR 948 949 #endif /* not XML_MIN_SIZE */ 950 951 #ifdef XML_NS 952 953 static const struct normal_encoding big2_encoding_ns = { 954 { VTABLE, 2, 0, 955 #if BYTEORDER == 4321 956 1 957 #else 958 0 959 #endif 960 }, 961 { 962 #include "asciitab.h" 963 #include "latin1tab.h" 964 }, 965 STANDARD_VTABLE(big2_) NULL_VTABLE 966 }; 967 968 #endif 969 970 static const struct normal_encoding big2_encoding = { 971 { VTABLE, 2, 0, 972 #if BYTEORDER == 4321 973 1 974 #else 975 0 976 #endif 977 }, 978 { 979 #define BT_COLON BT_NMSTRT 980 #include "asciitab.h" 981 #undef BT_COLON 982 #include "latin1tab.h" 983 }, 984 STANDARD_VTABLE(big2_) NULL_VTABLE 985 }; 986 987 #if BYTEORDER != 1234 988 989 #ifdef XML_NS 990 991 static const struct normal_encoding internal_big2_encoding_ns = { 992 { VTABLE, 2, 0, 1 }, 993 { 994 #include "iasciitab.h" 995 #include "latin1tab.h" 996 }, 997 STANDARD_VTABLE(big2_) NULL_VTABLE 998 }; 999 1000 #endif 1001 1002 static const struct normal_encoding internal_big2_encoding = { 1003 { VTABLE, 2, 0, 1 }, 1004 { 1005 #define BT_COLON BT_NMSTRT 1006 #include "iasciitab.h" 1007 #undef BT_COLON 1008 #include "latin1tab.h" 1009 }, 1010 STANDARD_VTABLE(big2_) NULL_VTABLE 1011 }; 1012 1013 #endif 1014 1015 #undef PREFIX 1016 1017 static int FASTCALL 1018 streqci(const char *s1, const char *s2) 1019 { 1020 for (;;) { 1021 char c1 = *s1++; 1022 char c2 = *s2++; 1023 if (ASCII_a <= c1 && c1 <= ASCII_z) 1024 c1 += ASCII_A - ASCII_a; 1025 if (ASCII_a <= c2 && c2 <= ASCII_z) 1026 c2 += ASCII_A - ASCII_a; 1027 if (c1 != c2) 1028 return 0; 1029 if (!c1) 1030 break; 1031 } 1032 return 1; 1033 } 1034 1035 static void PTRCALL 1036 initUpdatePosition(const ENCODING *UNUSED_P(enc), const char *ptr, 1037 const char *end, POSITION *pos) 1038 { 1039 normal_updatePosition(&utf8_encoding.enc, ptr, end, pos); 1040 } 1041 1042 static int 1043 toAscii(const ENCODING *enc, const char *ptr, const char *end) 1044 { 1045 char buf[1]; 1046 char *p = buf; 1047 XmlUtf8Convert(enc, &ptr, end, &p, p + 1); 1048 if (p == buf) 1049 return -1; 1050 else 1051 return buf[0]; 1052 } 1053 1054 static int FASTCALL 1055 isSpace(int c) 1056 { 1057 switch (c) { 1058 case 0x20: 1059 case 0xD: 1060 case 0xA: 1061 case 0x9: 1062 return 1; 1063 } 1064 return 0; 1065 } 1066 1067 /* Return 1 if there's just optional white space or there's an S 1068 followed by name=val. 1069 */ 1070 static int 1071 parsePseudoAttribute(const ENCODING *enc, 1072 const char *ptr, 1073 const char *end, 1074 const char **namePtr, 1075 const char **nameEndPtr, 1076 const char **valPtr, 1077 const char **nextTokPtr) 1078 { 1079 int c; 1080 char open; 1081 if (ptr == end) { 1082 *namePtr = NULL; 1083 return 1; 1084 } 1085 if (!isSpace(toAscii(enc, ptr, end))) { 1086 *nextTokPtr = ptr; 1087 return 0; 1088 } 1089 do { 1090 ptr += enc->minBytesPerChar; 1091 } while (isSpace(toAscii(enc, ptr, end))); 1092 if (ptr == end) { 1093 *namePtr = NULL; 1094 return 1; 1095 } 1096 *namePtr = ptr; 1097 for (;;) { 1098 c = toAscii(enc, ptr, end); 1099 if (c == -1) { 1100 *nextTokPtr = ptr; 1101 return 0; 1102 } 1103 if (c == ASCII_EQUALS) { 1104 *nameEndPtr = ptr; 1105 break; 1106 } 1107 if (isSpace(c)) { 1108 *nameEndPtr = ptr; 1109 do { 1110 ptr += enc->minBytesPerChar; 1111 } while (isSpace(c = toAscii(enc, ptr, end))); 1112 if (c != ASCII_EQUALS) { 1113 *nextTokPtr = ptr; 1114 return 0; 1115 } 1116 break; 1117 } 1118 ptr += enc->minBytesPerChar; 1119 } 1120 if (ptr == *namePtr) { 1121 *nextTokPtr = ptr; 1122 return 0; 1123 } 1124 ptr += enc->minBytesPerChar; 1125 c = toAscii(enc, ptr, end); 1126 while (isSpace(c)) { 1127 ptr += enc->minBytesPerChar; 1128 c = toAscii(enc, ptr, end); 1129 } 1130 if (c != ASCII_QUOT && c != ASCII_APOS) { 1131 *nextTokPtr = ptr; 1132 return 0; 1133 } 1134 open = (char)c; 1135 ptr += enc->minBytesPerChar; 1136 *valPtr = ptr; 1137 for (;; ptr += enc->minBytesPerChar) { 1138 c = toAscii(enc, ptr, end); 1139 if (c == open) 1140 break; 1141 if (!(ASCII_a <= c && c <= ASCII_z) 1142 && !(ASCII_A <= c && c <= ASCII_Z) 1143 && !(ASCII_0 <= c && c <= ASCII_9) 1144 && c != ASCII_PERIOD 1145 && c != ASCII_MINUS 1146 && c != ASCII_UNDERSCORE) { 1147 *nextTokPtr = ptr; 1148 return 0; 1149 } 1150 } 1151 *nextTokPtr = ptr + enc->minBytesPerChar; 1152 return 1; 1153 } 1154 1155 static const char KW_version[] = { 1156 ASCII_v, ASCII_e, ASCII_r, ASCII_s, ASCII_i, ASCII_o, ASCII_n, '\0' 1157 }; 1158 1159 static const char KW_encoding[] = { 1160 ASCII_e, ASCII_n, ASCII_c, ASCII_o, ASCII_d, ASCII_i, ASCII_n, ASCII_g, '\0' 1161 }; 1162 1163 static const char KW_standalone[] = { 1164 ASCII_s, ASCII_t, ASCII_a, ASCII_n, ASCII_d, ASCII_a, ASCII_l, ASCII_o, 1165 ASCII_n, ASCII_e, '\0' 1166 }; 1167 1168 static const char KW_yes[] = { 1169 ASCII_y, ASCII_e, ASCII_s, '\0' 1170 }; 1171 1172 static const char KW_no[] = { 1173 ASCII_n, ASCII_o, '\0' 1174 }; 1175 1176 static int 1177 doParseXmlDecl(const ENCODING *(*encodingFinder)(const ENCODING *, 1178 const char *, 1179 const char *), 1180 int isGeneralTextEntity, 1181 const ENCODING *enc, 1182 const char *ptr, 1183 const char *end, 1184 const char **badPtr, 1185 const char **versionPtr, 1186 const char **versionEndPtr, 1187 const char **encodingName, 1188 const ENCODING **encoding, 1189 int *standalone) 1190 { 1191 const char *val = NULL; 1192 const char *name = NULL; 1193 const char *nameEnd = NULL; 1194 ptr += 5 * enc->minBytesPerChar; 1195 end -= 2 * enc->minBytesPerChar; 1196 if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr) 1197 || !name) { 1198 *badPtr = ptr; 1199 return 0; 1200 } 1201 if (!XmlNameMatchesAscii(enc, name, nameEnd, KW_version)) { 1202 if (!isGeneralTextEntity) { 1203 *badPtr = name; 1204 return 0; 1205 } 1206 } 1207 else { 1208 if (versionPtr) 1209 *versionPtr = val; 1210 if (versionEndPtr) 1211 *versionEndPtr = ptr; 1212 if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)) { 1213 *badPtr = ptr; 1214 return 0; 1215 } 1216 if (!name) { 1217 if (isGeneralTextEntity) { 1218 /* a TextDecl must have an EncodingDecl */ 1219 *badPtr = ptr; 1220 return 0; 1221 } 1222 return 1; 1223 } 1224 } 1225 if (XmlNameMatchesAscii(enc, name, nameEnd, KW_encoding)) { 1226 int c = toAscii(enc, val, end); 1227 if (!(ASCII_a <= c && c <= ASCII_z) && !(ASCII_A <= c && c <= ASCII_Z)) { 1228 *badPtr = val; 1229 return 0; 1230 } 1231 if (encodingName) 1232 *encodingName = val; 1233 if (encoding) 1234 *encoding = encodingFinder(enc, val, ptr - enc->minBytesPerChar); 1235 if (!parsePseudoAttribute(enc, ptr, end, &name, &nameEnd, &val, &ptr)) { 1236 *badPtr = ptr; 1237 return 0; 1238 } 1239 if (!name) 1240 return 1; 1241 } 1242 if (!XmlNameMatchesAscii(enc, name, nameEnd, KW_standalone) 1243 || isGeneralTextEntity) { 1244 *badPtr = name; 1245 return 0; 1246 } 1247 if (XmlNameMatchesAscii(enc, val, ptr - enc->minBytesPerChar, KW_yes)) { 1248 if (standalone) 1249 *standalone = 1; 1250 } 1251 else if (XmlNameMatchesAscii(enc, val, ptr - enc->minBytesPerChar, KW_no)) { 1252 if (standalone) 1253 *standalone = 0; 1254 } 1255 else { 1256 *badPtr = val; 1257 return 0; 1258 } 1259 while (isSpace(toAscii(enc, ptr, end))) 1260 ptr += enc->minBytesPerChar; 1261 if (ptr != end) { 1262 *badPtr = ptr; 1263 return 0; 1264 } 1265 return 1; 1266 } 1267 1268 static int FASTCALL 1269 checkCharRefNumber(int result) 1270 { 1271 switch (result >> 8) { 1272 case 0xD8: case 0xD9: case 0xDA: case 0xDB: 1273 case 0xDC: case 0xDD: case 0xDE: case 0xDF: 1274 return -1; 1275 case 0: 1276 if (latin1_encoding.type[result] == BT_NONXML) 1277 return -1; 1278 break; 1279 case 0xFF: 1280 if (result == 0xFFFE || result == 0xFFFF) 1281 return -1; 1282 break; 1283 } 1284 return result; 1285 } 1286 1287 int FASTCALL 1288 XmlUtf8Encode(int c, char *buf) 1289 { 1290 enum { 1291 /* minN is minimum legal resulting value for N byte sequence */ 1292 min2 = 0x80, 1293 min3 = 0x800, 1294 min4 = 0x10000 1295 }; 1296 1297 if (c < 0) 1298 return 0; 1299 if (c < min2) { 1300 buf[0] = (char)(c | UTF8_cval1); 1301 return 1; 1302 } 1303 if (c < min3) { 1304 buf[0] = (char)((c >> 6) | UTF8_cval2); 1305 buf[1] = (char)((c & 0x3f) | 0x80); 1306 return 2; 1307 } 1308 if (c < min4) { 1309 buf[0] = (char)((c >> 12) | UTF8_cval3); 1310 buf[1] = (char)(((c >> 6) & 0x3f) | 0x80); 1311 buf[2] = (char)((c & 0x3f) | 0x80); 1312 return 3; 1313 } 1314 if (c < 0x110000) { 1315 buf[0] = (char)((c >> 18) | UTF8_cval4); 1316 buf[1] = (char)(((c >> 12) & 0x3f) | 0x80); 1317 buf[2] = (char)(((c >> 6) & 0x3f) | 0x80); 1318 buf[3] = (char)((c & 0x3f) | 0x80); 1319 return 4; 1320 } 1321 return 0; 1322 } 1323 1324 int FASTCALL 1325 XmlUtf16Encode(int charNum, unsigned short *buf) 1326 { 1327 if (charNum < 0) 1328 return 0; 1329 if (charNum < 0x10000) { 1330 buf[0] = (unsigned short)charNum; 1331 return 1; 1332 } 1333 if (charNum < 0x110000) { 1334 charNum -= 0x10000; 1335 buf[0] = (unsigned short)((charNum >> 10) + 0xD800); 1336 buf[1] = (unsigned short)((charNum & 0x3FF) + 0xDC00); 1337 return 2; 1338 } 1339 return 0; 1340 } 1341 1342 struct unknown_encoding { 1343 struct normal_encoding normal; 1344 CONVERTER convert; 1345 void *userData; 1346 unsigned short utf16[256]; 1347 char utf8[256][4]; 1348 }; 1349 1350 #define AS_UNKNOWN_ENCODING(enc) ((const struct unknown_encoding *) (enc)) 1351 1352 int 1353 XmlSizeOfUnknownEncoding(void) 1354 { 1355 return sizeof(struct unknown_encoding); 1356 } 1357 1358 static int PTRFASTCALL 1359 unknown_isName(const ENCODING *enc, const char *p) 1360 { 1361 const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc); 1362 int c = uenc->convert(uenc->userData, p); 1363 if (c & ~0xFFFF) 1364 return 0; 1365 return UCS2_GET_NAMING(namePages, c >> 8, c & 0xFF); 1366 } 1367 1368 static int PTRFASTCALL 1369 unknown_isNmstrt(const ENCODING *enc, const char *p) 1370 { 1371 const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc); 1372 int c = uenc->convert(uenc->userData, p); 1373 if (c & ~0xFFFF) 1374 return 0; 1375 return UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xFF); 1376 } 1377 1378 static int PTRFASTCALL 1379 unknown_isInvalid(const ENCODING *enc, const char *p) 1380 { 1381 const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc); 1382 int c = uenc->convert(uenc->userData, p); 1383 return (c & ~0xFFFF) || checkCharRefNumber(c) < 0; 1384 } 1385 1386 static enum XML_Convert_Result PTRCALL 1387 unknown_toUtf8(const ENCODING *enc, 1388 const char **fromP, const char *fromLim, 1389 char **toP, const char *toLim) 1390 { 1391 const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc); 1392 char buf[XML_UTF8_ENCODE_MAX]; 1393 for (;;) { 1394 const char *utf8; 1395 int n; 1396 if (*fromP == fromLim) 1397 return XML_CONVERT_COMPLETED; 1398 utf8 = uenc->utf8[(unsigned char)**fromP]; 1399 n = *utf8++; 1400 if (n == 0) { 1401 int c = uenc->convert(uenc->userData, *fromP); 1402 n = XmlUtf8Encode(c, buf); 1403 if (n > toLim - *toP) 1404 return XML_CONVERT_OUTPUT_EXHAUSTED; 1405 utf8 = buf; 1406 *fromP += (AS_NORMAL_ENCODING(enc)->type[(unsigned char)**fromP] 1407 - (BT_LEAD2 - 2)); 1408 } 1409 else { 1410 if (n > toLim - *toP) 1411 return XML_CONVERT_OUTPUT_EXHAUSTED; 1412 (*fromP)++; 1413 } 1414 do { 1415 *(*toP)++ = *utf8++; 1416 } while (--n != 0); 1417 } 1418 } 1419 1420 static enum XML_Convert_Result PTRCALL 1421 unknown_toUtf16(const ENCODING *enc, 1422 const char **fromP, const char *fromLim, 1423 unsigned short **toP, const unsigned short *toLim) 1424 { 1425 const struct unknown_encoding *uenc = AS_UNKNOWN_ENCODING(enc); 1426 while (*fromP < fromLim && *toP < toLim) { 1427 unsigned short c = uenc->utf16[(unsigned char)**fromP]; 1428 if (c == 0) { 1429 c = (unsigned short) 1430 uenc->convert(uenc->userData, *fromP); 1431 *fromP += (AS_NORMAL_ENCODING(enc)->type[(unsigned char)**fromP] 1432 - (BT_LEAD2 - 2)); 1433 } 1434 else 1435 (*fromP)++; 1436 *(*toP)++ = c; 1437 } 1438 1439 if ((*toP == toLim) && (*fromP < fromLim)) 1440 return XML_CONVERT_OUTPUT_EXHAUSTED; 1441 else 1442 return XML_CONVERT_COMPLETED; 1443 } 1444 1445 ENCODING * 1446 XmlInitUnknownEncoding(void *mem, 1447 int *table, 1448 CONVERTER convert, 1449 void *userData) 1450 { 1451 int i; 1452 struct unknown_encoding *e = (struct unknown_encoding *)mem; 1453 for (i = 0; i < (int)sizeof(struct normal_encoding); i++) 1454 ((char *)mem)[i] = ((char *)&latin1_encoding)[i]; 1455 for (i = 0; i < 128; i++) 1456 if (latin1_encoding.type[i] != BT_OTHER 1457 && latin1_encoding.type[i] != BT_NONXML 1458 && table[i] != i) 1459 return 0; 1460 for (i = 0; i < 256; i++) { 1461 int c = table[i]; 1462 if (c == -1) { 1463 e->normal.type[i] = BT_MALFORM; 1464 /* This shouldn't really get used. */ 1465 e->utf16[i] = 0xFFFF; 1466 e->utf8[i][0] = 1; 1467 e->utf8[i][1] = 0; 1468 } 1469 else if (c < 0) { 1470 if (c < -4) 1471 return 0; 1472 e->normal.type[i] = (unsigned char)(BT_LEAD2 - (c + 2)); 1473 e->utf8[i][0] = 0; 1474 e->utf16[i] = 0; 1475 } 1476 else if (c < 0x80) { 1477 if (latin1_encoding.type[c] != BT_OTHER 1478 && latin1_encoding.type[c] != BT_NONXML 1479 && c != i) 1480 return 0; 1481 e->normal.type[i] = latin1_encoding.type[c]; 1482 e->utf8[i][0] = 1; 1483 e->utf8[i][1] = (char)c; 1484 e->utf16[i] = (unsigned short)(c == 0 ? 0xFFFF : c); 1485 } 1486 else if (checkCharRefNumber(c) < 0) { 1487 e->normal.type[i] = BT_NONXML; 1488 /* This shouldn't really get used. */ 1489 e->utf16[i] = 0xFFFF; 1490 e->utf8[i][0] = 1; 1491 e->utf8[i][1] = 0; 1492 } 1493 else { 1494 if (c > 0xFFFF) 1495 return 0; 1496 if (UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xff)) 1497 e->normal.type[i] = BT_NMSTRT; 1498 else if (UCS2_GET_NAMING(namePages, c >> 8, c & 0xff)) 1499 e->normal.type[i] = BT_NAME; 1500 else 1501 e->normal.type[i] = BT_OTHER; 1502 e->utf8[i][0] = (char)XmlUtf8Encode(c, e->utf8[i] + 1); 1503 e->utf16[i] = (unsigned short)c; 1504 } 1505 } 1506 e->userData = userData; 1507 e->convert = convert; 1508 if (convert) { 1509 e->normal.isName2 = unknown_isName; 1510 e->normal.isName3 = unknown_isName; 1511 e->normal.isName4 = unknown_isName; 1512 e->normal.isNmstrt2 = unknown_isNmstrt; 1513 e->normal.isNmstrt3 = unknown_isNmstrt; 1514 e->normal.isNmstrt4 = unknown_isNmstrt; 1515 e->normal.isInvalid2 = unknown_isInvalid; 1516 e->normal.isInvalid3 = unknown_isInvalid; 1517 e->normal.isInvalid4 = unknown_isInvalid; 1518 } 1519 e->normal.enc.utf8Convert = unknown_toUtf8; 1520 e->normal.enc.utf16Convert = unknown_toUtf16; 1521 return &(e->normal.enc); 1522 } 1523 1524 /* If this enumeration is changed, getEncodingIndex and encodings 1525 must also be changed. */ 1526 enum { 1527 UNKNOWN_ENC = -1, 1528 ISO_8859_1_ENC = 0, 1529 US_ASCII_ENC, 1530 UTF_8_ENC, 1531 UTF_16_ENC, 1532 UTF_16BE_ENC, 1533 UTF_16LE_ENC, 1534 /* must match encodingNames up to here */ 1535 NO_ENC 1536 }; 1537 1538 static const char KW_ISO_8859_1[] = { 1539 ASCII_I, ASCII_S, ASCII_O, ASCII_MINUS, ASCII_8, ASCII_8, ASCII_5, ASCII_9, 1540 ASCII_MINUS, ASCII_1, '\0' 1541 }; 1542 static const char KW_US_ASCII[] = { 1543 ASCII_U, ASCII_S, ASCII_MINUS, ASCII_A, ASCII_S, ASCII_C, ASCII_I, ASCII_I, 1544 '\0' 1545 }; 1546 static const char KW_UTF_8[] = { 1547 ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_8, '\0' 1548 }; 1549 static const char KW_UTF_16[] = { 1550 ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, '\0' 1551 }; 1552 static const char KW_UTF_16BE[] = { 1553 ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, ASCII_B, ASCII_E, 1554 '\0' 1555 }; 1556 static const char KW_UTF_16LE[] = { 1557 ASCII_U, ASCII_T, ASCII_F, ASCII_MINUS, ASCII_1, ASCII_6, ASCII_L, ASCII_E, 1558 '\0' 1559 }; 1560 1561 static int FASTCALL 1562 getEncodingIndex(const char *name) 1563 { 1564 static const char * const encodingNames[] = { 1565 KW_ISO_8859_1, 1566 KW_US_ASCII, 1567 KW_UTF_8, 1568 KW_UTF_16, 1569 KW_UTF_16BE, 1570 KW_UTF_16LE, 1571 }; 1572 int i; 1573 if (name == NULL) 1574 return NO_ENC; 1575 for (i = 0; i < (int)(sizeof(encodingNames)/sizeof(encodingNames[0])); i++) 1576 if (streqci(name, encodingNames[i])) 1577 return i; 1578 return UNKNOWN_ENC; 1579 } 1580 1581 /* For binary compatibility, we store the index of the encoding 1582 specified at initialization in the isUtf16 member. 1583 */ 1584 1585 #define INIT_ENC_INDEX(enc) ((int)(enc)->initEnc.isUtf16) 1586 #define SET_INIT_ENC_INDEX(enc, i) ((enc)->initEnc.isUtf16 = (char)i) 1587 1588 /* This is what detects the encoding. encodingTable maps from 1589 encoding indices to encodings; INIT_ENC_INDEX(enc) is the index of 1590 the external (protocol) specified encoding; state is 1591 XML_CONTENT_STATE if we're parsing an external text entity, and 1592 XML_PROLOG_STATE otherwise. 1593 */ 1594 1595 1596 static int 1597 initScan(const ENCODING * const *encodingTable, 1598 const INIT_ENCODING *enc, 1599 int state, 1600 const char *ptr, 1601 const char *end, 1602 const char **nextTokPtr) 1603 { 1604 const ENCODING **encPtr; 1605 1606 if (ptr >= end) 1607 return XML_TOK_NONE; 1608 encPtr = enc->encPtr; 1609 if (ptr + 1 == end) { 1610 /* only a single byte available for auto-detection */ 1611 #ifndef XML_DTD /* FIXME */ 1612 /* a well-formed document entity must have more than one byte */ 1613 if (state != XML_CONTENT_STATE) 1614 return XML_TOK_PARTIAL; 1615 #endif 1616 /* so we're parsing an external text entity... */ 1617 /* if UTF-16 was externally specified, then we need at least 2 bytes */ 1618 switch (INIT_ENC_INDEX(enc)) { 1619 case UTF_16_ENC: 1620 case UTF_16LE_ENC: 1621 case UTF_16BE_ENC: 1622 return XML_TOK_PARTIAL; 1623 } 1624 switch ((unsigned char)*ptr) { 1625 case 0xFE: 1626 case 0xFF: 1627 case 0xEF: /* possibly first byte of UTF-8 BOM */ 1628 if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC 1629 && state == XML_CONTENT_STATE) 1630 break; 1631 /* fall through */ 1632 case 0x00: 1633 case 0x3C: 1634 return XML_TOK_PARTIAL; 1635 } 1636 } 1637 else { 1638 switch (((unsigned char)ptr[0] << 8) | (unsigned char)ptr[1]) { 1639 case 0xFEFF: 1640 if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC 1641 && state == XML_CONTENT_STATE) 1642 break; 1643 *nextTokPtr = ptr + 2; 1644 *encPtr = encodingTable[UTF_16BE_ENC]; 1645 return XML_TOK_BOM; 1646 /* 00 3C is handled in the default case */ 1647 case 0x3C00: 1648 if ((INIT_ENC_INDEX(enc) == UTF_16BE_ENC 1649 || INIT_ENC_INDEX(enc) == UTF_16_ENC) 1650 && state == XML_CONTENT_STATE) 1651 break; 1652 *encPtr = encodingTable[UTF_16LE_ENC]; 1653 return XmlTok(*encPtr, state, ptr, end, nextTokPtr); 1654 case 0xFFFE: 1655 if (INIT_ENC_INDEX(enc) == ISO_8859_1_ENC 1656 && state == XML_CONTENT_STATE) 1657 break; 1658 *nextTokPtr = ptr + 2; 1659 *encPtr = encodingTable[UTF_16LE_ENC]; 1660 return XML_TOK_BOM; 1661 case 0xEFBB: 1662 /* Maybe a UTF-8 BOM (EF BB BF) */ 1663 /* If there's an explicitly specified (external) encoding 1664 of ISO-8859-1 or some flavour of UTF-16 1665 and this is an external text entity, 1666 don't look for the BOM, 1667 because it might be a legal data. 1668 */ 1669 if (state == XML_CONTENT_STATE) { 1670 int e = INIT_ENC_INDEX(enc); 1671 if (e == ISO_8859_1_ENC || e == UTF_16BE_ENC 1672 || e == UTF_16LE_ENC || e == UTF_16_ENC) 1673 break; 1674 } 1675 if (ptr + 2 == end) 1676 return XML_TOK_PARTIAL; 1677 if ((unsigned char)ptr[2] == 0xBF) { 1678 *nextTokPtr = ptr + 3; 1679 *encPtr = encodingTable[UTF_8_ENC]; 1680 return XML_TOK_BOM; 1681 } 1682 break; 1683 default: 1684 if (ptr[0] == '\0') { 1685 /* 0 isn't a legal data character. Furthermore a document 1686 entity can only start with ASCII characters. So the only 1687 way this can fail to be big-endian UTF-16 if it it's an 1688 external parsed general entity that's labelled as 1689 UTF-16LE. 1690 */ 1691 if (state == XML_CONTENT_STATE && INIT_ENC_INDEX(enc) == UTF_16LE_ENC) 1692 break; 1693 *encPtr = encodingTable[UTF_16BE_ENC]; 1694 return XmlTok(*encPtr, state, ptr, end, nextTokPtr); 1695 } 1696 else if (ptr[1] == '\0') { 1697 /* We could recover here in the case: 1698 - parsing an external entity 1699 - second byte is 0 1700 - no externally specified encoding 1701 - no encoding declaration 1702 by assuming UTF-16LE. But we don't, because this would mean when 1703 presented just with a single byte, we couldn't reliably determine 1704 whether we needed further bytes. 1705 */ 1706 if (state == XML_CONTENT_STATE) 1707 break; 1708 *encPtr = encodingTable[UTF_16LE_ENC]; 1709 return XmlTok(*encPtr, state, ptr, end, nextTokPtr); 1710 } 1711 break; 1712 } 1713 } 1714 *encPtr = encodingTable[INIT_ENC_INDEX(enc)]; 1715 return XmlTok(*encPtr, state, ptr, end, nextTokPtr); 1716 } 1717 1718 1719 #define NS(x) x 1720 #define ns(x) x 1721 #define XML_TOK_NS_C 1722 #include "xmltok_ns.c" 1723 #undef XML_TOK_NS_C 1724 #undef NS 1725 #undef ns 1726 1727 #ifdef XML_NS 1728 1729 #define NS(x) x ## NS 1730 #define ns(x) x ## _ns 1731 1732 #define XML_TOK_NS_C 1733 #include "xmltok_ns.c" 1734 #undef XML_TOK_NS_C 1735 1736 #undef NS 1737 #undef ns 1738 1739 ENCODING * 1740 XmlInitUnknownEncodingNS(void *mem, 1741 int *table, 1742 CONVERTER convert, 1743 void *userData) 1744 { 1745 ENCODING *enc = XmlInitUnknownEncoding(mem, table, convert, userData); 1746 if (enc) 1747 ((struct normal_encoding *)enc)->type[ASCII_COLON] = BT_COLON; 1748 return enc; 1749 } 1750 1751 #endif /* XML_NS */ 1752