Fri, 31 Jan 2003 13:03:47 +0000
[gaim-migrate @ 4766]
aim_user is dead. long live gaim_account.
| 2086 | 1 | /* |
| 2 | The contents of this file are subject to the Mozilla Public License | |
| 3 | Version 1.1 (the "License"); you may not use this file except in | |
| 4 | compliance with the License. You may obtain a copy of the License at | |
| 5 | http://www.mozilla.org/MPL/ | |
| 6 | ||
| 7 | Software distributed under the License is distributed on an "AS IS" | |
| 8 | basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the | |
| 9 | License for the specific language governing rights and limitations | |
| 10 | under the License. | |
| 11 | ||
| 12 | The Original Code is expat. | |
| 13 | ||
| 14 | The Initial Developer of the Original Code is James Clark. | |
| 15 | Portions created by James Clark are Copyright (C) 1998, 1999 | |
| 16 | James Clark. All Rights Reserved. | |
| 17 | ||
| 18 | Contributor(s): | |
| 19 | ||
| 20 | */ | |
| 21 | ||
| 22 | #ifndef IS_INVALID_CHAR | |
| 23 | #define IS_INVALID_CHAR(enc, ptr, n) (0) | |
| 24 | #endif | |
| 25 | ||
| 26 | #define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \ | |
| 27 | case BT_LEAD ## n: \ | |
| 28 | if (end - ptr < n) \ | |
| 29 | return XML_TOK_PARTIAL_CHAR; \ | |
| 30 | if (IS_INVALID_CHAR(enc, ptr, n)) { \ | |
| 31 | *(nextTokPtr) = (ptr); \ | |
| 32 | return XML_TOK_INVALID; \ | |
| 33 | } \ | |
| 34 | ptr += n; \ | |
| 35 | break; | |
| 36 | ||
| 37 | #define INVALID_CASES(ptr, nextTokPtr) \ | |
| 38 | INVALID_LEAD_CASE(2, ptr, nextTokPtr) \ | |
| 39 | INVALID_LEAD_CASE(3, ptr, nextTokPtr) \ | |
| 40 | INVALID_LEAD_CASE(4, ptr, nextTokPtr) \ | |
| 41 | case BT_NONXML: \ | |
| 42 | case BT_MALFORM: \ | |
| 43 | case BT_TRAIL: \ | |
| 44 | *(nextTokPtr) = (ptr); \ | |
| 45 | return XML_TOK_INVALID; | |
| 46 | ||
| 47 | #define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \ | |
| 48 | case BT_LEAD ## n: \ | |
| 49 | if (end - ptr < n) \ | |
| 50 | return XML_TOK_PARTIAL_CHAR; \ | |
| 51 | if (!IS_NAME_CHAR(enc, ptr, n)) { \ | |
| 52 | *nextTokPtr = ptr; \ | |
| 53 | return XML_TOK_INVALID; \ | |
| 54 | } \ | |
| 55 | ptr += n; \ | |
| 56 | break; | |
| 57 | ||
| 58 | #define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \ | |
| 59 | case BT_NONASCII: \ | |
| 60 | if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \ | |
| 61 | *nextTokPtr = ptr; \ | |
| 62 | return XML_TOK_INVALID; \ | |
| 63 | } \ | |
| 64 | case BT_NMSTRT: \ | |
| 65 | case BT_HEX: \ | |
| 66 | case BT_DIGIT: \ | |
| 67 | case BT_NAME: \ | |
| 68 | case BT_MINUS: \ | |
| 69 | ptr += MINBPC(enc); \ | |
| 70 | break; \ | |
| 71 | CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \ | |
| 72 | CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \ | |
| 73 | CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr) | |
| 74 | ||
| 75 | #define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \ | |
| 76 | case BT_LEAD ## n: \ | |
| 77 | if (end - ptr < n) \ | |
| 78 | return XML_TOK_PARTIAL_CHAR; \ | |
| 79 | if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \ | |
| 80 | *nextTokPtr = ptr; \ | |
| 81 | return XML_TOK_INVALID; \ | |
| 82 | } \ | |
| 83 | ptr += n; \ | |
| 84 | break; | |
| 85 | ||
| 86 | #define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \ | |
| 87 | case BT_NONASCII: \ | |
| 88 | if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \ | |
| 89 | *nextTokPtr = ptr; \ | |
| 90 | return XML_TOK_INVALID; \ | |
| 91 | } \ | |
| 92 | case BT_NMSTRT: \ | |
| 93 | case BT_HEX: \ | |
| 94 | ptr += MINBPC(enc); \ | |
| 95 | break; \ | |
| 96 | CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \ | |
| 97 | CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \ | |
| 98 | CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr) | |
| 99 | ||
| 100 | #ifndef PREFIX | |
| 101 | #define PREFIX(ident) ident | |
| 102 | #endif | |
| 103 | ||
| 104 | /* ptr points to character following "<!-" */ | |
| 105 | ||
| 106 | static | |
| 107 | int PREFIX(scanComment)(const ENCODING *enc, const char *ptr, const char *end, | |
| 108 | const char **nextTokPtr) | |
| 109 | { | |
| 110 | if (ptr != end) { | |
| 111 | if (!CHAR_MATCHES(enc, ptr, '-')) { | |
| 112 | *nextTokPtr = ptr; | |
| 113 | return XML_TOK_INVALID; | |
| 114 | } | |
| 115 | ptr += MINBPC(enc); | |
| 116 | while (ptr != end) { | |
| 117 | switch (BYTE_TYPE(enc, ptr)) { | |
| 118 | INVALID_CASES(ptr, nextTokPtr) | |
| 119 | case BT_MINUS: | |
| 120 | if ((ptr += MINBPC(enc)) == end) | |
| 121 | return XML_TOK_PARTIAL; | |
| 122 | if (CHAR_MATCHES(enc, ptr, '-')) { | |
| 123 | if ((ptr += MINBPC(enc)) == end) | |
| 124 | return XML_TOK_PARTIAL; | |
| 125 | if (!CHAR_MATCHES(enc, ptr, '>')) { | |
| 126 | *nextTokPtr = ptr; | |
| 127 | return XML_TOK_INVALID; | |
| 128 | } | |
| 129 | *nextTokPtr = ptr + MINBPC(enc); | |
| 130 | return XML_TOK_COMMENT; | |
| 131 | } | |
| 132 | break; | |
| 133 | default: | |
| 134 | ptr += MINBPC(enc); | |
| 135 | break; | |
| 136 | } | |
| 137 | } | |
| 138 | } | |
| 139 | return XML_TOK_PARTIAL; | |
| 140 | } | |
| 141 | ||
| 142 | /* ptr points to character following "<!" */ | |
| 143 | ||
| 144 | static | |
| 145 | int PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, const char *end, | |
| 146 | const char **nextTokPtr) | |
| 147 | { | |
| 148 | if (ptr == end) | |
| 149 | return XML_TOK_PARTIAL; | |
| 150 | switch (BYTE_TYPE(enc, ptr)) { | |
| 151 | case BT_MINUS: | |
| 152 | return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 153 | case BT_LSQB: | |
| 154 | *nextTokPtr = ptr + MINBPC(enc); | |
| 155 | return XML_TOK_COND_SECT_OPEN; | |
| 156 | case BT_NMSTRT: | |
| 157 | case BT_HEX: | |
| 158 | ptr += MINBPC(enc); | |
| 159 | break; | |
| 160 | default: | |
| 161 | *nextTokPtr = ptr; | |
| 162 | return XML_TOK_INVALID; | |
| 163 | } | |
| 164 | while (ptr != end) { | |
| 165 | switch (BYTE_TYPE(enc, ptr)) { | |
| 166 | case BT_PERCNT: | |
| 167 | if (ptr + MINBPC(enc) == end) | |
| 168 | return XML_TOK_PARTIAL; | |
| 169 | /* don't allow <!ENTITY% foo "whatever"> */ | |
| 170 | switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) { | |
| 171 | case BT_S: case BT_CR: case BT_LF: case BT_PERCNT: | |
| 172 | *nextTokPtr = ptr; | |
| 173 | return XML_TOK_INVALID; | |
| 174 | } | |
| 175 | /* fall through */ | |
| 176 | case BT_S: case BT_CR: case BT_LF: | |
| 177 | *nextTokPtr = ptr; | |
| 178 | return XML_TOK_DECL_OPEN; | |
| 179 | case BT_NMSTRT: | |
| 180 | case BT_HEX: | |
| 181 | ptr += MINBPC(enc); | |
| 182 | break; | |
| 183 | default: | |
| 184 | *nextTokPtr = ptr; | |
| 185 | return XML_TOK_INVALID; | |
| 186 | } | |
| 187 | } | |
| 188 | return XML_TOK_PARTIAL; | |
| 189 | } | |
| 190 | ||
| 191 | static | |
| 192 | int PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, const char *end, int *tokPtr) | |
| 193 | { | |
| 194 | int upper = 0; | |
| 195 | *tokPtr = XML_TOK_PI; | |
| 196 | if (end - ptr != MINBPC(enc)*3) | |
| 197 | return 1; | |
| 198 | switch (BYTE_TO_ASCII(enc, ptr)) { | |
| 199 | case 'x': | |
| 200 | break; | |
| 201 | case 'X': | |
| 202 | upper = 1; | |
| 203 | break; | |
| 204 | default: | |
| 205 | return 1; | |
| 206 | } | |
| 207 | ptr += MINBPC(enc); | |
| 208 | switch (BYTE_TO_ASCII(enc, ptr)) { | |
| 209 | case 'm': | |
| 210 | break; | |
| 211 | case 'M': | |
| 212 | upper = 1; | |
| 213 | break; | |
| 214 | default: | |
| 215 | return 1; | |
| 216 | } | |
| 217 | ptr += MINBPC(enc); | |
| 218 | switch (BYTE_TO_ASCII(enc, ptr)) { | |
| 219 | case 'l': | |
| 220 | break; | |
| 221 | case 'L': | |
| 222 | upper = 1; | |
| 223 | break; | |
| 224 | default: | |
| 225 | return 1; | |
| 226 | } | |
| 227 | if (upper) | |
| 228 | return 0; | |
| 229 | *tokPtr = XML_TOK_XML_DECL; | |
| 230 | return 1; | |
| 231 | } | |
| 232 | ||
| 233 | /* ptr points to character following "<?" */ | |
| 234 | ||
| 235 | static | |
| 236 | int PREFIX(scanPi)(const ENCODING *enc, const char *ptr, const char *end, | |
| 237 | const char **nextTokPtr) | |
| 238 | { | |
| 239 | int tok; | |
| 240 | const char *target = ptr; | |
| 241 | if (ptr == end) | |
| 242 | return XML_TOK_PARTIAL; | |
| 243 | switch (BYTE_TYPE(enc, ptr)) { | |
| 244 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 245 | default: | |
| 246 | *nextTokPtr = ptr; | |
| 247 | return XML_TOK_INVALID; | |
| 248 | } | |
| 249 | while (ptr != end) { | |
| 250 | switch (BYTE_TYPE(enc, ptr)) { | |
| 251 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 252 | case BT_S: case BT_CR: case BT_LF: | |
| 253 | if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) { | |
| 254 | *nextTokPtr = ptr; | |
| 255 | return XML_TOK_INVALID; | |
| 256 | } | |
| 257 | ptr += MINBPC(enc); | |
| 258 | while (ptr != end) { | |
| 259 | switch (BYTE_TYPE(enc, ptr)) { | |
| 260 | INVALID_CASES(ptr, nextTokPtr) | |
| 261 | case BT_QUEST: | |
| 262 | ptr += MINBPC(enc); | |
| 263 | if (ptr == end) | |
| 264 | return XML_TOK_PARTIAL; | |
| 265 | if (CHAR_MATCHES(enc, ptr, '>')) { | |
| 266 | *nextTokPtr = ptr + MINBPC(enc); | |
| 267 | return tok; | |
| 268 | } | |
| 269 | break; | |
| 270 | default: | |
| 271 | ptr += MINBPC(enc); | |
| 272 | break; | |
| 273 | } | |
| 274 | } | |
| 275 | return XML_TOK_PARTIAL; | |
| 276 | case BT_QUEST: | |
| 277 | if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) { | |
| 278 | *nextTokPtr = ptr; | |
| 279 | return XML_TOK_INVALID; | |
| 280 | } | |
| 281 | ptr += MINBPC(enc); | |
| 282 | if (ptr == end) | |
| 283 | return XML_TOK_PARTIAL; | |
| 284 | if (CHAR_MATCHES(enc, ptr, '>')) { | |
| 285 | *nextTokPtr = ptr + MINBPC(enc); | |
| 286 | return tok; | |
| 287 | } | |
| 288 | /* fall through */ | |
| 289 | default: | |
| 290 | *nextTokPtr = ptr; | |
| 291 | return XML_TOK_INVALID; | |
| 292 | } | |
| 293 | } | |
| 294 | return XML_TOK_PARTIAL; | |
| 295 | } | |
| 296 | ||
| 297 | ||
| 298 | static | |
| 299 | int PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, const char *end, | |
| 300 | const char **nextTokPtr) | |
| 301 | { | |
| 302 | int i; | |
| 303 | /* CDATA[ */ | |
| 304 | if (end - ptr < 6 * MINBPC(enc)) | |
| 305 | return XML_TOK_PARTIAL; | |
| 306 | for (i = 0; i < 6; i++, ptr += MINBPC(enc)) { | |
| 307 | if (!CHAR_MATCHES(enc, ptr, "CDATA["[i])) { | |
| 308 | *nextTokPtr = ptr; | |
| 309 | return XML_TOK_INVALID; | |
| 310 | } | |
| 311 | } | |
| 312 | *nextTokPtr = ptr; | |
| 313 | return XML_TOK_CDATA_SECT_OPEN; | |
| 314 | } | |
| 315 | ||
| 316 | static | |
| 317 | int PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, const char *end, | |
| 318 | const char **nextTokPtr) | |
| 319 | { | |
| 320 | if (ptr == end) | |
| 321 | return XML_TOK_NONE; | |
| 322 | if (MINBPC(enc) > 1) { | |
| 323 | size_t n = end - ptr; | |
| 324 | if (n & (MINBPC(enc) - 1)) { | |
| 325 | n &= ~(MINBPC(enc) - 1); | |
| 326 | if (n == 0) | |
| 327 | return XML_TOK_PARTIAL; | |
| 328 | end = ptr + n; | |
| 329 | } | |
| 330 | } | |
| 331 | switch (BYTE_TYPE(enc, ptr)) { | |
| 332 | case BT_RSQB: | |
| 333 | ptr += MINBPC(enc); | |
| 334 | if (ptr == end) | |
| 335 | return XML_TOK_PARTIAL; | |
| 336 | if (!CHAR_MATCHES(enc, ptr, ']')) | |
| 337 | break; | |
| 338 | ptr += MINBPC(enc); | |
| 339 | if (ptr == end) | |
| 340 | return XML_TOK_PARTIAL; | |
| 341 | if (!CHAR_MATCHES(enc, ptr, '>')) { | |
| 342 | ptr -= MINBPC(enc); | |
| 343 | break; | |
| 344 | } | |
| 345 | *nextTokPtr = ptr + MINBPC(enc); | |
| 346 | return XML_TOK_CDATA_SECT_CLOSE; | |
| 347 | case BT_CR: | |
| 348 | ptr += MINBPC(enc); | |
| 349 | if (ptr == end) | |
| 350 | return XML_TOK_PARTIAL; | |
| 351 | if (BYTE_TYPE(enc, ptr) == BT_LF) | |
| 352 | ptr += MINBPC(enc); | |
| 353 | *nextTokPtr = ptr; | |
| 354 | return XML_TOK_DATA_NEWLINE; | |
| 355 | case BT_LF: | |
| 356 | *nextTokPtr = ptr + MINBPC(enc); | |
| 357 | return XML_TOK_DATA_NEWLINE; | |
| 358 | INVALID_CASES(ptr, nextTokPtr) | |
| 359 | default: | |
| 360 | ptr += MINBPC(enc); | |
| 361 | break; | |
| 362 | } | |
| 363 | while (ptr != end) { | |
| 364 | switch (BYTE_TYPE(enc, ptr)) { | |
| 365 | #define LEAD_CASE(n) \ | |
| 366 | case BT_LEAD ## n: \ | |
| 367 | if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \ | |
| 368 | *nextTokPtr = ptr; \ | |
| 369 | return XML_TOK_DATA_CHARS; \ | |
| 370 | } \ | |
| 371 | ptr += n; \ | |
| 372 | break; | |
| 373 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 374 | #undef LEAD_CASE | |
| 375 | case BT_NONXML: | |
| 376 | case BT_MALFORM: | |
| 377 | case BT_TRAIL: | |
| 378 | case BT_CR: | |
| 379 | case BT_LF: | |
| 380 | case BT_RSQB: | |
| 381 | *nextTokPtr = ptr; | |
| 382 | return XML_TOK_DATA_CHARS; | |
| 383 | default: | |
| 384 | ptr += MINBPC(enc); | |
| 385 | break; | |
| 386 | } | |
| 387 | } | |
| 388 | *nextTokPtr = ptr; | |
| 389 | return XML_TOK_DATA_CHARS; | |
| 390 | } | |
| 391 | ||
| 392 | /* ptr points to character following "</" */ | |
| 393 | ||
| 394 | static | |
| 395 | int PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, const char *end, | |
| 396 | const char **nextTokPtr) | |
| 397 | { | |
| 398 | if (ptr == end) | |
| 399 | return XML_TOK_PARTIAL; | |
| 400 | switch (BYTE_TYPE(enc, ptr)) { | |
| 401 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 402 | default: | |
| 403 | *nextTokPtr = ptr; | |
| 404 | return XML_TOK_INVALID; | |
| 405 | } | |
| 406 | while (ptr != end) { | |
| 407 | switch (BYTE_TYPE(enc, ptr)) { | |
| 408 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 409 | case BT_S: case BT_CR: case BT_LF: | |
| 410 | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) { | |
| 411 | switch (BYTE_TYPE(enc, ptr)) { | |
| 412 | case BT_S: case BT_CR: case BT_LF: | |
| 413 | break; | |
| 414 | case BT_GT: | |
| 415 | *nextTokPtr = ptr + MINBPC(enc); | |
| 416 | return XML_TOK_END_TAG; | |
| 417 | default: | |
| 418 | *nextTokPtr = ptr; | |
| 419 | return XML_TOK_INVALID; | |
| 420 | } | |
| 421 | } | |
| 422 | return XML_TOK_PARTIAL; | |
| 423 | #ifdef XML_NS | |
| 424 | case BT_COLON: | |
| 425 | /* no need to check qname syntax here, since end-tag must match exactly */ | |
| 426 | ptr += MINBPC(enc); | |
| 427 | break; | |
| 428 | #endif | |
| 429 | case BT_GT: | |
| 430 | *nextTokPtr = ptr + MINBPC(enc); | |
| 431 | return XML_TOK_END_TAG; | |
| 432 | default: | |
| 433 | *nextTokPtr = ptr; | |
| 434 | return XML_TOK_INVALID; | |
| 435 | } | |
| 436 | } | |
| 437 | return XML_TOK_PARTIAL; | |
| 438 | } | |
| 439 | ||
| 440 | /* ptr points to character following "&#X" */ | |
| 441 | ||
| 442 | static | |
| 443 | int PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, const char *end, | |
| 444 | const char **nextTokPtr) | |
| 445 | { | |
| 446 | if (ptr != end) { | |
| 447 | switch (BYTE_TYPE(enc, ptr)) { | |
| 448 | case BT_DIGIT: | |
| 449 | case BT_HEX: | |
| 450 | break; | |
| 451 | default: | |
| 452 | *nextTokPtr = ptr; | |
| 453 | return XML_TOK_INVALID; | |
| 454 | } | |
| 455 | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) { | |
| 456 | switch (BYTE_TYPE(enc, ptr)) { | |
| 457 | case BT_DIGIT: | |
| 458 | case BT_HEX: | |
| 459 | break; | |
| 460 | case BT_SEMI: | |
| 461 | *nextTokPtr = ptr + MINBPC(enc); | |
| 462 | return XML_TOK_CHAR_REF; | |
| 463 | default: | |
| 464 | *nextTokPtr = ptr; | |
| 465 | return XML_TOK_INVALID; | |
| 466 | } | |
| 467 | } | |
| 468 | } | |
| 469 | return XML_TOK_PARTIAL; | |
| 470 | } | |
| 471 | ||
| 472 | /* ptr points to character following "&#" */ | |
| 473 | ||
| 474 | static | |
| 475 | int PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, const char *end, | |
| 476 | const char **nextTokPtr) | |
| 477 | { | |
| 478 | if (ptr != end) { | |
| 479 | if (CHAR_MATCHES(enc, ptr, 'x')) | |
| 480 | return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 481 | switch (BYTE_TYPE(enc, ptr)) { | |
| 482 | case BT_DIGIT: | |
| 483 | break; | |
| 484 | default: | |
| 485 | *nextTokPtr = ptr; | |
| 486 | return XML_TOK_INVALID; | |
| 487 | } | |
| 488 | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) { | |
| 489 | switch (BYTE_TYPE(enc, ptr)) { | |
| 490 | case BT_DIGIT: | |
| 491 | break; | |
| 492 | case BT_SEMI: | |
| 493 | *nextTokPtr = ptr + MINBPC(enc); | |
| 494 | return XML_TOK_CHAR_REF; | |
| 495 | default: | |
| 496 | *nextTokPtr = ptr; | |
| 497 | return XML_TOK_INVALID; | |
| 498 | } | |
| 499 | } | |
| 500 | } | |
| 501 | return XML_TOK_PARTIAL; | |
| 502 | } | |
| 503 | ||
| 504 | /* ptr points to character following "&" */ | |
| 505 | ||
| 506 | static | |
| 507 | int PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end, | |
| 508 | const char **nextTokPtr) | |
| 509 | { | |
| 510 | if (ptr == end) | |
| 511 | return XML_TOK_PARTIAL; | |
| 512 | switch (BYTE_TYPE(enc, ptr)) { | |
| 513 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 514 | case BT_NUM: | |
| 515 | return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 516 | default: | |
| 517 | *nextTokPtr = ptr; | |
| 518 | return XML_TOK_INVALID; | |
| 519 | } | |
| 520 | while (ptr != end) { | |
| 521 | switch (BYTE_TYPE(enc, ptr)) { | |
| 522 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 523 | case BT_SEMI: | |
| 524 | *nextTokPtr = ptr + MINBPC(enc); | |
| 525 | return XML_TOK_ENTITY_REF; | |
| 526 | default: | |
| 527 | *nextTokPtr = ptr; | |
| 528 | return XML_TOK_INVALID; | |
| 529 | } | |
| 530 | } | |
| 531 | return XML_TOK_PARTIAL; | |
| 532 | } | |
| 533 | ||
| 534 | /* ptr points to character following first character of attribute name */ | |
| 535 | ||
| 536 | static | |
| 537 | int PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end, | |
| 538 | const char **nextTokPtr) | |
| 539 | { | |
| 540 | #ifdef XML_NS | |
| 541 | int hadColon = 0; | |
| 542 | #endif | |
| 543 | while (ptr != end) { | |
| 544 | switch (BYTE_TYPE(enc, ptr)) { | |
| 545 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 546 | #ifdef XML_NS | |
| 547 | case BT_COLON: | |
| 548 | if (hadColon) { | |
| 549 | *nextTokPtr = ptr; | |
| 550 | return XML_TOK_INVALID; | |
| 551 | } | |
| 552 | hadColon = 1; | |
| 553 | ptr += MINBPC(enc); | |
| 554 | if (ptr == end) | |
| 555 | return XML_TOK_PARTIAL; | |
| 556 | switch (BYTE_TYPE(enc, ptr)) { | |
| 557 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 558 | default: | |
| 559 | *nextTokPtr = ptr; | |
| 560 | return XML_TOK_INVALID; | |
| 561 | } | |
| 562 | break; | |
| 563 | #endif | |
| 564 | case BT_S: case BT_CR: case BT_LF: | |
| 565 | for (;;) { | |
| 566 | int t; | |
| 567 | ||
| 568 | ptr += MINBPC(enc); | |
| 569 | if (ptr == end) | |
| 570 | return XML_TOK_PARTIAL; | |
| 571 | t = BYTE_TYPE(enc, ptr); | |
| 572 | if (t == BT_EQUALS) | |
| 573 | break; | |
| 574 | switch (t) { | |
| 575 | case BT_S: | |
| 576 | case BT_LF: | |
| 577 | case BT_CR: | |
| 578 | break; | |
| 579 | default: | |
| 580 | *nextTokPtr = ptr; | |
| 581 | return XML_TOK_INVALID; | |
| 582 | } | |
| 583 | } | |
| 584 | /* fall through */ | |
| 585 | case BT_EQUALS: | |
| 586 | { | |
| 587 | int open; | |
| 588 | #ifdef XML_NS | |
| 589 | hadColon = 0; | |
| 590 | #endif | |
| 591 | for (;;) { | |
| 592 | ||
| 593 | ptr += MINBPC(enc); | |
| 594 | if (ptr == end) | |
| 595 | return XML_TOK_PARTIAL; | |
| 596 | open = BYTE_TYPE(enc, ptr); | |
| 597 | if (open == BT_QUOT || open == BT_APOS) | |
| 598 | break; | |
| 599 | switch (open) { | |
| 600 | case BT_S: | |
| 601 | case BT_LF: | |
| 602 | case BT_CR: | |
| 603 | break; | |
| 604 | default: | |
| 605 | *nextTokPtr = ptr; | |
| 606 | return XML_TOK_INVALID; | |
| 607 | } | |
| 608 | } | |
| 609 | ptr += MINBPC(enc); | |
| 610 | /* in attribute value */ | |
| 611 | for (;;) { | |
| 612 | int t; | |
| 613 | if (ptr == end) | |
| 614 | return XML_TOK_PARTIAL; | |
| 615 | t = BYTE_TYPE(enc, ptr); | |
| 616 | if (t == open) | |
| 617 | break; | |
| 618 | switch (t) { | |
| 619 | INVALID_CASES(ptr, nextTokPtr) | |
| 620 | case BT_AMP: | |
| 621 | { | |
| 622 | int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr); | |
| 623 | if (tok <= 0) { | |
| 624 | if (tok == XML_TOK_INVALID) | |
| 625 | *nextTokPtr = ptr; | |
| 626 | return tok; | |
| 627 | } | |
| 628 | break; | |
| 629 | } | |
| 630 | case BT_LT: | |
| 631 | *nextTokPtr = ptr; | |
| 632 | return XML_TOK_INVALID; | |
| 633 | default: | |
| 634 | ptr += MINBPC(enc); | |
| 635 | break; | |
| 636 | } | |
| 637 | } | |
| 638 | ptr += MINBPC(enc); | |
| 639 | if (ptr == end) | |
| 640 | return XML_TOK_PARTIAL; | |
| 641 | switch (BYTE_TYPE(enc, ptr)) { | |
| 642 | case BT_S: | |
| 643 | case BT_CR: | |
| 644 | case BT_LF: | |
| 645 | break; | |
| 646 | case BT_SOL: | |
| 647 | goto sol; | |
| 648 | case BT_GT: | |
| 649 | goto gt; | |
| 650 | default: | |
| 651 | *nextTokPtr = ptr; | |
| 652 | return XML_TOK_INVALID; | |
| 653 | } | |
| 654 | /* ptr points to closing quote */ | |
| 655 | for (;;) { | |
| 656 | ptr += MINBPC(enc); | |
| 657 | if (ptr == end) | |
| 658 | return XML_TOK_PARTIAL; | |
| 659 | switch (BYTE_TYPE(enc, ptr)) { | |
| 660 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 661 | case BT_S: case BT_CR: case BT_LF: | |
| 662 | continue; | |
| 663 | case BT_GT: | |
| 664 | gt: | |
| 665 | *nextTokPtr = ptr + MINBPC(enc); | |
| 666 | return XML_TOK_START_TAG_WITH_ATTS; | |
| 667 | case BT_SOL: | |
| 668 | sol: | |
| 669 | ptr += MINBPC(enc); | |
| 670 | if (ptr == end) | |
| 671 | return XML_TOK_PARTIAL; | |
| 672 | if (!CHAR_MATCHES(enc, ptr, '>')) { | |
| 673 | *nextTokPtr = ptr; | |
| 674 | return XML_TOK_INVALID; | |
| 675 | } | |
| 676 | *nextTokPtr = ptr + MINBPC(enc); | |
| 677 | return XML_TOK_EMPTY_ELEMENT_WITH_ATTS; | |
| 678 | default: | |
| 679 | *nextTokPtr = ptr; | |
| 680 | return XML_TOK_INVALID; | |
| 681 | } | |
| 682 | break; | |
| 683 | } | |
| 684 | break; | |
| 685 | } | |
| 686 | default: | |
| 687 | *nextTokPtr = ptr; | |
| 688 | return XML_TOK_INVALID; | |
| 689 | } | |
| 690 | } | |
| 691 | return XML_TOK_PARTIAL; | |
| 692 | } | |
| 693 | ||
| 694 | /* ptr points to character following "<" */ | |
| 695 | ||
| 696 | static | |
| 697 | int PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end, | |
| 698 | const char **nextTokPtr) | |
| 699 | { | |
| 700 | #ifdef XML_NS | |
| 701 | int hadColon; | |
| 702 | #endif | |
| 703 | if (ptr == end) | |
| 704 | return XML_TOK_PARTIAL; | |
| 705 | switch (BYTE_TYPE(enc, ptr)) { | |
| 706 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 707 | case BT_EXCL: | |
| 708 | if ((ptr += MINBPC(enc)) == end) | |
| 709 | return XML_TOK_PARTIAL; | |
| 710 | switch (BYTE_TYPE(enc, ptr)) { | |
| 711 | case BT_MINUS: | |
| 712 | return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 713 | case BT_LSQB: | |
| 714 | return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 715 | } | |
| 716 | *nextTokPtr = ptr; | |
| 717 | return XML_TOK_INVALID; | |
| 718 | case BT_QUEST: | |
| 719 | return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 720 | case BT_SOL: | |
| 721 | return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 722 | default: | |
| 723 | *nextTokPtr = ptr; | |
| 724 | return XML_TOK_INVALID; | |
| 725 | } | |
| 726 | #ifdef XML_NS | |
| 727 | hadColon = 0; | |
| 728 | #endif | |
| 729 | /* we have a start-tag */ | |
| 730 | while (ptr != end) { | |
| 731 | switch (BYTE_TYPE(enc, ptr)) { | |
| 732 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 733 | #ifdef XML_NS | |
| 734 | case BT_COLON: | |
| 735 | if (hadColon) { | |
| 736 | *nextTokPtr = ptr; | |
| 737 | return XML_TOK_INVALID; | |
| 738 | } | |
| 739 | hadColon = 1; | |
| 740 | ptr += MINBPC(enc); | |
| 741 | if (ptr == end) | |
| 742 | return XML_TOK_PARTIAL; | |
| 743 | switch (BYTE_TYPE(enc, ptr)) { | |
| 744 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 745 | default: | |
| 746 | *nextTokPtr = ptr; | |
| 747 | return XML_TOK_INVALID; | |
| 748 | } | |
| 749 | break; | |
| 750 | #endif | |
| 751 | case BT_S: case BT_CR: case BT_LF: | |
| 752 | { | |
| 753 | ptr += MINBPC(enc); | |
| 754 | while (ptr != end) { | |
| 755 | switch (BYTE_TYPE(enc, ptr)) { | |
| 756 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 757 | case BT_GT: | |
| 758 | goto gt; | |
| 759 | case BT_SOL: | |
| 760 | goto sol; | |
| 761 | case BT_S: case BT_CR: case BT_LF: | |
| 762 | ptr += MINBPC(enc); | |
| 763 | continue; | |
| 764 | default: | |
| 765 | *nextTokPtr = ptr; | |
| 766 | return XML_TOK_INVALID; | |
| 767 | } | |
| 768 | return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr); | |
| 769 | } | |
| 770 | return XML_TOK_PARTIAL; | |
| 771 | } | |
| 772 | case BT_GT: | |
| 773 | gt: | |
| 774 | *nextTokPtr = ptr + MINBPC(enc); | |
| 775 | return XML_TOK_START_TAG_NO_ATTS; | |
| 776 | case BT_SOL: | |
| 777 | sol: | |
| 778 | ptr += MINBPC(enc); | |
| 779 | if (ptr == end) | |
| 780 | return XML_TOK_PARTIAL; | |
| 781 | if (!CHAR_MATCHES(enc, ptr, '>')) { | |
| 782 | *nextTokPtr = ptr; | |
| 783 | return XML_TOK_INVALID; | |
| 784 | } | |
| 785 | *nextTokPtr = ptr + MINBPC(enc); | |
| 786 | return XML_TOK_EMPTY_ELEMENT_NO_ATTS; | |
| 787 | default: | |
| 788 | *nextTokPtr = ptr; | |
| 789 | return XML_TOK_INVALID; | |
| 790 | } | |
| 791 | } | |
| 792 | return XML_TOK_PARTIAL; | |
| 793 | } | |
| 794 | ||
| 795 | static | |
| 796 | int PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end, | |
| 797 | const char **nextTokPtr) | |
| 798 | { | |
| 799 | if (ptr == end) | |
| 800 | return XML_TOK_NONE; | |
| 801 | if (MINBPC(enc) > 1) { | |
| 802 | size_t n = end - ptr; | |
| 803 | if (n & (MINBPC(enc) - 1)) { | |
| 804 | n &= ~(MINBPC(enc) - 1); | |
| 805 | if (n == 0) | |
| 806 | return XML_TOK_PARTIAL; | |
| 807 | end = ptr + n; | |
| 808 | } | |
| 809 | } | |
| 810 | switch (BYTE_TYPE(enc, ptr)) { | |
| 811 | case BT_LT: | |
| 812 | return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 813 | case BT_AMP: | |
| 814 | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 815 | case BT_CR: | |
| 816 | ptr += MINBPC(enc); | |
| 817 | if (ptr == end) | |
| 818 | return XML_TOK_TRAILING_CR; | |
| 819 | if (BYTE_TYPE(enc, ptr) == BT_LF) | |
| 820 | ptr += MINBPC(enc); | |
| 821 | *nextTokPtr = ptr; | |
| 822 | return XML_TOK_DATA_NEWLINE; | |
| 823 | case BT_LF: | |
| 824 | *nextTokPtr = ptr + MINBPC(enc); | |
| 825 | return XML_TOK_DATA_NEWLINE; | |
| 826 | case BT_RSQB: | |
| 827 | ptr += MINBPC(enc); | |
| 828 | if (ptr == end) | |
| 829 | return XML_TOK_TRAILING_RSQB; | |
| 830 | if (!CHAR_MATCHES(enc, ptr, ']')) | |
| 831 | break; | |
| 832 | ptr += MINBPC(enc); | |
| 833 | if (ptr == end) | |
| 834 | return XML_TOK_TRAILING_RSQB; | |
| 835 | if (!CHAR_MATCHES(enc, ptr, '>')) { | |
| 836 | ptr -= MINBPC(enc); | |
| 837 | break; | |
| 838 | } | |
| 839 | *nextTokPtr = ptr; | |
| 840 | return XML_TOK_INVALID; | |
| 841 | INVALID_CASES(ptr, nextTokPtr) | |
| 842 | default: | |
| 843 | ptr += MINBPC(enc); | |
| 844 | break; | |
| 845 | } | |
| 846 | while (ptr != end) { | |
| 847 | switch (BYTE_TYPE(enc, ptr)) { | |
| 848 | #define LEAD_CASE(n) \ | |
| 849 | case BT_LEAD ## n: \ | |
| 850 | if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \ | |
| 851 | *nextTokPtr = ptr; \ | |
| 852 | return XML_TOK_DATA_CHARS; \ | |
| 853 | } \ | |
| 854 | ptr += n; \ | |
| 855 | break; | |
| 856 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 857 | #undef LEAD_CASE | |
| 858 | case BT_RSQB: | |
| 859 | if (ptr + MINBPC(enc) != end) { | |
| 860 | if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ']')) { | |
| 861 | ptr += MINBPC(enc); | |
| 862 | break; | |
| 863 | } | |
| 864 | if (ptr + 2*MINBPC(enc) != end) { | |
| 865 | if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), '>')) { | |
| 866 | ptr += MINBPC(enc); | |
| 867 | break; | |
| 868 | } | |
| 869 | *nextTokPtr = ptr + 2*MINBPC(enc); | |
| 870 | return XML_TOK_INVALID; | |
| 871 | } | |
| 872 | } | |
| 873 | /* fall through */ | |
| 874 | case BT_AMP: | |
| 875 | case BT_LT: | |
| 876 | case BT_NONXML: | |
| 877 | case BT_MALFORM: | |
| 878 | case BT_TRAIL: | |
| 879 | case BT_CR: | |
| 880 | case BT_LF: | |
| 881 | *nextTokPtr = ptr; | |
| 882 | return XML_TOK_DATA_CHARS; | |
| 883 | default: | |
| 884 | ptr += MINBPC(enc); | |
| 885 | break; | |
| 886 | } | |
| 887 | } | |
| 888 | *nextTokPtr = ptr; | |
| 889 | return XML_TOK_DATA_CHARS; | |
| 890 | } | |
| 891 | ||
| 892 | /* ptr points to character following "%" */ | |
| 893 | ||
| 894 | static | |
| 895 | int PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end, | |
| 896 | const char **nextTokPtr) | |
| 897 | { | |
| 898 | if (ptr == end) | |
| 899 | return XML_TOK_PARTIAL; | |
| 900 | switch (BYTE_TYPE(enc, ptr)) { | |
| 901 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 902 | case BT_S: case BT_LF: case BT_CR: case BT_PERCNT: | |
| 903 | *nextTokPtr = ptr; | |
| 904 | return XML_TOK_PERCENT; | |
| 905 | default: | |
| 906 | *nextTokPtr = ptr; | |
| 907 | return XML_TOK_INVALID; | |
| 908 | } | |
| 909 | while (ptr != end) { | |
| 910 | switch (BYTE_TYPE(enc, ptr)) { | |
| 911 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 912 | case BT_SEMI: | |
| 913 | *nextTokPtr = ptr + MINBPC(enc); | |
| 914 | return XML_TOK_PARAM_ENTITY_REF; | |
| 915 | default: | |
| 916 | *nextTokPtr = ptr; | |
| 917 | return XML_TOK_INVALID; | |
| 918 | } | |
| 919 | } | |
| 920 | return XML_TOK_PARTIAL; | |
| 921 | } | |
| 922 | ||
| 923 | static | |
| 924 | int PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end, | |
| 925 | const char **nextTokPtr) | |
| 926 | { | |
| 927 | if (ptr == end) | |
| 928 | return XML_TOK_PARTIAL; | |
| 929 | switch (BYTE_TYPE(enc, ptr)) { | |
| 930 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) | |
| 931 | default: | |
| 932 | *nextTokPtr = ptr; | |
| 933 | return XML_TOK_INVALID; | |
| 934 | } | |
| 935 | while (ptr != end) { | |
| 936 | switch (BYTE_TYPE(enc, ptr)) { | |
| 937 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 938 | case BT_CR: case BT_LF: case BT_S: | |
| 939 | case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR: | |
| 940 | *nextTokPtr = ptr; | |
| 941 | return XML_TOK_POUND_NAME; | |
| 942 | default: | |
| 943 | *nextTokPtr = ptr; | |
| 944 | return XML_TOK_INVALID; | |
| 945 | } | |
| 946 | } | |
| 947 | return XML_TOK_PARTIAL; | |
| 948 | } | |
| 949 | ||
| 950 | static | |
| 951 | int PREFIX(scanLit)(int open, const ENCODING *enc, | |
| 952 | const char *ptr, const char *end, | |
| 953 | const char **nextTokPtr) | |
| 954 | { | |
| 955 | while (ptr != end) { | |
| 956 | int t = BYTE_TYPE(enc, ptr); | |
| 957 | switch (t) { | |
| 958 | INVALID_CASES(ptr, nextTokPtr) | |
| 959 | case BT_QUOT: | |
| 960 | case BT_APOS: | |
| 961 | ptr += MINBPC(enc); | |
| 962 | if (t != open) | |
| 963 | break; | |
| 964 | if (ptr == end) | |
| 965 | return XML_TOK_PARTIAL; | |
| 966 | *nextTokPtr = ptr; | |
| 967 | switch (BYTE_TYPE(enc, ptr)) { | |
| 968 | case BT_S: case BT_CR: case BT_LF: | |
| 969 | case BT_GT: case BT_PERCNT: case BT_LSQB: | |
| 970 | return XML_TOK_LITERAL; | |
| 971 | default: | |
| 972 | return XML_TOK_INVALID; | |
| 973 | } | |
| 974 | default: | |
| 975 | ptr += MINBPC(enc); | |
| 976 | break; | |
| 977 | } | |
| 978 | } | |
| 979 | return XML_TOK_PARTIAL; | |
| 980 | } | |
| 981 | ||
| 982 | static | |
| 983 | int PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end, | |
| 984 | const char **nextTokPtr) | |
| 985 | { | |
| 986 | int tok; | |
| 987 | if (ptr == end) | |
| 988 | return XML_TOK_NONE; | |
| 989 | if (MINBPC(enc) > 1) { | |
| 990 | size_t n = end - ptr; | |
| 991 | if (n & (MINBPC(enc) - 1)) { | |
| 992 | n &= ~(MINBPC(enc) - 1); | |
| 993 | if (n == 0) | |
| 994 | return XML_TOK_PARTIAL; | |
| 995 | end = ptr + n; | |
| 996 | } | |
| 997 | } | |
| 998 | switch (BYTE_TYPE(enc, ptr)) { | |
| 999 | case BT_QUOT: | |
| 1000 | return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1001 | case BT_APOS: | |
| 1002 | return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1003 | case BT_LT: | |
| 1004 | { | |
| 1005 | ptr += MINBPC(enc); | |
| 1006 | if (ptr == end) | |
| 1007 | return XML_TOK_PARTIAL; | |
| 1008 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1009 | case BT_EXCL: | |
| 1010 | return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1011 | case BT_QUEST: | |
| 1012 | return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1013 | case BT_NMSTRT: | |
| 1014 | case BT_HEX: | |
| 1015 | case BT_NONASCII: | |
| 1016 | case BT_LEAD2: | |
| 1017 | case BT_LEAD3: | |
| 1018 | case BT_LEAD4: | |
| 1019 | *nextTokPtr = ptr - MINBPC(enc); | |
| 1020 | return XML_TOK_INSTANCE_START; | |
| 1021 | } | |
| 1022 | *nextTokPtr = ptr; | |
| 1023 | return XML_TOK_INVALID; | |
| 1024 | } | |
| 1025 | case BT_CR: | |
| 1026 | if (ptr + MINBPC(enc) == end) | |
| 1027 | return XML_TOK_TRAILING_CR; | |
| 1028 | /* fall through */ | |
| 1029 | case BT_S: case BT_LF: | |
| 1030 | for (;;) { | |
| 1031 | ptr += MINBPC(enc); | |
| 1032 | if (ptr == end) | |
| 1033 | break; | |
| 1034 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1035 | case BT_S: case BT_LF: | |
| 1036 | break; | |
| 1037 | case BT_CR: | |
| 1038 | /* don't split CR/LF pair */ | |
| 1039 | if (ptr + MINBPC(enc) != end) | |
| 1040 | break; | |
| 1041 | /* fall through */ | |
| 1042 | default: | |
| 1043 | *nextTokPtr = ptr; | |
| 1044 | return XML_TOK_PROLOG_S; | |
| 1045 | } | |
| 1046 | } | |
| 1047 | *nextTokPtr = ptr; | |
| 1048 | return XML_TOK_PROLOG_S; | |
| 1049 | case BT_PERCNT: | |
| 1050 | return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1051 | case BT_COMMA: | |
| 1052 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1053 | return XML_TOK_COMMA; | |
| 1054 | case BT_LSQB: | |
| 1055 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1056 | return XML_TOK_OPEN_BRACKET; | |
| 1057 | case BT_RSQB: | |
| 1058 | ptr += MINBPC(enc); | |
| 1059 | if (ptr == end) | |
| 1060 | return XML_TOK_PARTIAL; | |
| 1061 | if (CHAR_MATCHES(enc, ptr, ']')) { | |
| 1062 | if (ptr + MINBPC(enc) == end) | |
| 1063 | return XML_TOK_PARTIAL; | |
| 1064 | if (CHAR_MATCHES(enc, ptr + MINBPC(enc), '>')) { | |
| 1065 | *nextTokPtr = ptr + 2*MINBPC(enc); | |
| 1066 | return XML_TOK_COND_SECT_CLOSE; | |
| 1067 | } | |
| 1068 | } | |
| 1069 | *nextTokPtr = ptr; | |
| 1070 | return XML_TOK_CLOSE_BRACKET; | |
| 1071 | case BT_LPAR: | |
| 1072 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1073 | return XML_TOK_OPEN_PAREN; | |
| 1074 | case BT_RPAR: | |
| 1075 | ptr += MINBPC(enc); | |
| 1076 | if (ptr == end) | |
| 1077 | return XML_TOK_PARTIAL; | |
| 1078 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1079 | case BT_AST: | |
| 1080 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1081 | return XML_TOK_CLOSE_PAREN_ASTERISK; | |
| 1082 | case BT_QUEST: | |
| 1083 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1084 | return XML_TOK_CLOSE_PAREN_QUESTION; | |
| 1085 | case BT_PLUS: | |
| 1086 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1087 | return XML_TOK_CLOSE_PAREN_PLUS; | |
| 1088 | case BT_CR: case BT_LF: case BT_S: | |
| 1089 | case BT_GT: case BT_COMMA: case BT_VERBAR: | |
| 1090 | case BT_RPAR: | |
| 1091 | *nextTokPtr = ptr; | |
| 1092 | return XML_TOK_CLOSE_PAREN; | |
| 1093 | } | |
| 1094 | *nextTokPtr = ptr; | |
| 1095 | return XML_TOK_INVALID; | |
| 1096 | case BT_VERBAR: | |
| 1097 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1098 | return XML_TOK_OR; | |
| 1099 | case BT_GT: | |
| 1100 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1101 | return XML_TOK_DECL_CLOSE; | |
| 1102 | case BT_NUM: | |
| 1103 | return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1104 | #define LEAD_CASE(n) \ | |
| 1105 | case BT_LEAD ## n: \ | |
| 1106 | if (end - ptr < n) \ | |
| 1107 | return XML_TOK_PARTIAL_CHAR; \ | |
| 1108 | if (IS_NMSTRT_CHAR(enc, ptr, n)) { \ | |
| 1109 | ptr += n; \ | |
| 1110 | tok = XML_TOK_NAME; \ | |
| 1111 | break; \ | |
| 1112 | } \ | |
| 1113 | if (IS_NAME_CHAR(enc, ptr, n)) { \ | |
| 1114 | ptr += n; \ | |
| 1115 | tok = XML_TOK_NMTOKEN; \ | |
| 1116 | break; \ | |
| 1117 | } \ | |
| 1118 | *nextTokPtr = ptr; \ | |
| 1119 | return XML_TOK_INVALID; | |
| 1120 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 1121 | #undef LEAD_CASE | |
| 1122 | case BT_NMSTRT: | |
| 1123 | case BT_HEX: | |
| 1124 | tok = XML_TOK_NAME; | |
| 1125 | ptr += MINBPC(enc); | |
| 1126 | break; | |
| 1127 | case BT_DIGIT: | |
| 1128 | case BT_NAME: | |
| 1129 | case BT_MINUS: | |
| 1130 | #ifdef XML_NS | |
| 1131 | case BT_COLON: | |
| 1132 | #endif | |
| 1133 | tok = XML_TOK_NMTOKEN; | |
| 1134 | ptr += MINBPC(enc); | |
| 1135 | break; | |
| 1136 | case BT_NONASCII: | |
| 1137 | if (IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { | |
| 1138 | ptr += MINBPC(enc); | |
| 1139 | tok = XML_TOK_NAME; | |
| 1140 | break; | |
| 1141 | } | |
| 1142 | if (IS_NAME_CHAR_MINBPC(enc, ptr)) { | |
| 1143 | ptr += MINBPC(enc); | |
| 1144 | tok = XML_TOK_NMTOKEN; | |
| 1145 | break; | |
| 1146 | } | |
| 1147 | /* fall through */ | |
| 1148 | default: | |
| 1149 | *nextTokPtr = ptr; | |
| 1150 | return XML_TOK_INVALID; | |
| 1151 | } | |
| 1152 | while (ptr != end) { | |
| 1153 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1154 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 1155 | case BT_GT: case BT_RPAR: case BT_COMMA: | |
| 1156 | case BT_VERBAR: case BT_LSQB: case BT_PERCNT: | |
| 1157 | case BT_S: case BT_CR: case BT_LF: | |
| 1158 | *nextTokPtr = ptr; | |
| 1159 | return tok; | |
| 1160 | #ifdef XML_NS | |
| 1161 | case BT_COLON: | |
| 1162 | ptr += MINBPC(enc); | |
| 1163 | switch (tok) { | |
| 1164 | case XML_TOK_NAME: | |
| 1165 | if (ptr == end) | |
| 1166 | return XML_TOK_PARTIAL; | |
| 1167 | tok = XML_TOK_PREFIXED_NAME; | |
| 1168 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1169 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) | |
| 1170 | default: | |
| 1171 | tok = XML_TOK_NMTOKEN; | |
| 1172 | break; | |
| 1173 | } | |
| 1174 | break; | |
| 1175 | case XML_TOK_PREFIXED_NAME: | |
| 1176 | tok = XML_TOK_NMTOKEN; | |
| 1177 | break; | |
| 1178 | } | |
| 1179 | break; | |
| 1180 | #endif | |
| 1181 | case BT_PLUS: | |
| 1182 | if (tok == XML_TOK_NMTOKEN) { | |
| 1183 | *nextTokPtr = ptr; | |
| 1184 | return XML_TOK_INVALID; | |
| 1185 | } | |
| 1186 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1187 | return XML_TOK_NAME_PLUS; | |
| 1188 | case BT_AST: | |
| 1189 | if (tok == XML_TOK_NMTOKEN) { | |
| 1190 | *nextTokPtr = ptr; | |
| 1191 | return XML_TOK_INVALID; | |
| 1192 | } | |
| 1193 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1194 | return XML_TOK_NAME_ASTERISK; | |
| 1195 | case BT_QUEST: | |
| 1196 | if (tok == XML_TOK_NMTOKEN) { | |
| 1197 | *nextTokPtr = ptr; | |
| 1198 | return XML_TOK_INVALID; | |
| 1199 | } | |
| 1200 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1201 | return XML_TOK_NAME_QUESTION; | |
| 1202 | default: | |
| 1203 | *nextTokPtr = ptr; | |
| 1204 | return XML_TOK_INVALID; | |
| 1205 | } | |
| 1206 | } | |
| 1207 | return XML_TOK_PARTIAL; | |
| 1208 | } | |
| 1209 | ||
| 1210 | static | |
| 1211 | int PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr, const char *end, | |
| 1212 | const char **nextTokPtr) | |
| 1213 | { | |
| 1214 | const char *start; | |
| 1215 | if (ptr == end) | |
| 1216 | return XML_TOK_NONE; | |
| 1217 | start = ptr; | |
| 1218 | while (ptr != end) { | |
| 1219 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1220 | #define LEAD_CASE(n) \ | |
| 1221 | case BT_LEAD ## n: ptr += n; break; | |
| 1222 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 1223 | #undef LEAD_CASE | |
| 1224 | case BT_AMP: | |
| 1225 | if (ptr == start) | |
| 1226 | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1227 | *nextTokPtr = ptr; | |
| 1228 | return XML_TOK_DATA_CHARS; | |
| 1229 | case BT_LT: | |
| 1230 | /* this is for inside entity references */ | |
| 1231 | *nextTokPtr = ptr; | |
| 1232 | return XML_TOK_INVALID; | |
| 1233 | case BT_LF: | |
| 1234 | if (ptr == start) { | |
| 1235 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1236 | return XML_TOK_DATA_NEWLINE; | |
| 1237 | } | |
| 1238 | *nextTokPtr = ptr; | |
| 1239 | return XML_TOK_DATA_CHARS; | |
| 1240 | case BT_CR: | |
| 1241 | if (ptr == start) { | |
| 1242 | ptr += MINBPC(enc); | |
| 1243 | if (ptr == end) | |
| 1244 | return XML_TOK_TRAILING_CR; | |
| 1245 | if (BYTE_TYPE(enc, ptr) == BT_LF) | |
| 1246 | ptr += MINBPC(enc); | |
| 1247 | *nextTokPtr = ptr; | |
| 1248 | return XML_TOK_DATA_NEWLINE; | |
| 1249 | } | |
| 1250 | *nextTokPtr = ptr; | |
| 1251 | return XML_TOK_DATA_CHARS; | |
| 1252 | case BT_S: | |
| 1253 | if (ptr == start) { | |
| 1254 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1255 | return XML_TOK_ATTRIBUTE_VALUE_S; | |
| 1256 | } | |
| 1257 | *nextTokPtr = ptr; | |
| 1258 | return XML_TOK_DATA_CHARS; | |
| 1259 | default: | |
| 1260 | ptr += MINBPC(enc); | |
| 1261 | break; | |
| 1262 | } | |
| 1263 | } | |
| 1264 | *nextTokPtr = ptr; | |
| 1265 | return XML_TOK_DATA_CHARS; | |
| 1266 | } | |
| 1267 | ||
| 1268 | static | |
| 1269 | int PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr, const char *end, | |
| 1270 | const char **nextTokPtr) | |
| 1271 | { | |
| 1272 | const char *start; | |
| 1273 | if (ptr == end) | |
| 1274 | return XML_TOK_NONE; | |
| 1275 | start = ptr; | |
| 1276 | while (ptr != end) { | |
| 1277 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1278 | #define LEAD_CASE(n) \ | |
| 1279 | case BT_LEAD ## n: ptr += n; break; | |
| 1280 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 1281 | #undef LEAD_CASE | |
| 1282 | case BT_AMP: | |
| 1283 | if (ptr == start) | |
| 1284 | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1285 | *nextTokPtr = ptr; | |
| 1286 | return XML_TOK_DATA_CHARS; | |
| 1287 | case BT_PERCNT: | |
| 1288 | if (ptr == start) | |
| 1289 | return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr); | |
| 1290 | *nextTokPtr = ptr; | |
| 1291 | return XML_TOK_DATA_CHARS; | |
| 1292 | case BT_LF: | |
| 1293 | if (ptr == start) { | |
| 1294 | *nextTokPtr = ptr + MINBPC(enc); | |
| 1295 | return XML_TOK_DATA_NEWLINE; | |
| 1296 | } | |
| 1297 | *nextTokPtr = ptr; | |
| 1298 | return XML_TOK_DATA_CHARS; | |
| 1299 | case BT_CR: | |
| 1300 | if (ptr == start) { | |
| 1301 | ptr += MINBPC(enc); | |
| 1302 | if (ptr == end) | |
| 1303 | return XML_TOK_TRAILING_CR; | |
| 1304 | if (BYTE_TYPE(enc, ptr) == BT_LF) | |
| 1305 | ptr += MINBPC(enc); | |
| 1306 | *nextTokPtr = ptr; | |
| 1307 | return XML_TOK_DATA_NEWLINE; | |
| 1308 | } | |
| 1309 | *nextTokPtr = ptr; | |
| 1310 | return XML_TOK_DATA_CHARS; | |
| 1311 | default: | |
| 1312 | ptr += MINBPC(enc); | |
| 1313 | break; | |
| 1314 | } | |
| 1315 | } | |
| 1316 | *nextTokPtr = ptr; | |
| 1317 | return XML_TOK_DATA_CHARS; | |
| 1318 | } | |
| 1319 | ||
| 1320 | static | |
| 1321 | int PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end, | |
| 1322 | const char **badPtr) | |
| 1323 | { | |
| 1324 | ptr += MINBPC(enc); | |
| 1325 | end -= MINBPC(enc); | |
| 1326 | for (; ptr != end; ptr += MINBPC(enc)) { | |
| 1327 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1328 | case BT_DIGIT: | |
| 1329 | case BT_HEX: | |
| 1330 | case BT_MINUS: | |
| 1331 | case BT_APOS: | |
| 1332 | case BT_LPAR: | |
| 1333 | case BT_RPAR: | |
| 1334 | case BT_PLUS: | |
| 1335 | case BT_COMMA: | |
| 1336 | case BT_SOL: | |
| 1337 | case BT_EQUALS: | |
| 1338 | case BT_QUEST: | |
| 1339 | case BT_CR: | |
| 1340 | case BT_LF: | |
| 1341 | case BT_SEMI: | |
| 1342 | case BT_EXCL: | |
| 1343 | case BT_AST: | |
| 1344 | case BT_PERCNT: | |
| 1345 | case BT_NUM: | |
| 1346 | #ifdef XML_NS | |
| 1347 | case BT_COLON: | |
| 1348 | #endif | |
| 1349 | break; | |
| 1350 | case BT_S: | |
| 1351 | if (CHAR_MATCHES(enc, ptr, '\t')) { | |
| 1352 | *badPtr = ptr; | |
| 1353 | return 0; | |
| 1354 | } | |
| 1355 | break; | |
| 1356 | case BT_NAME: | |
| 1357 | case BT_NMSTRT: | |
| 1358 | if (!(BYTE_TO_ASCII(enc, ptr) & ~0x7f)) | |
| 1359 | break; | |
| 1360 | default: | |
| 1361 | switch (BYTE_TO_ASCII(enc, ptr)) { | |
| 1362 | case 0x24: /* $ */ | |
| 1363 | case 0x40: /* @ */ | |
| 1364 | break; | |
| 1365 | default: | |
| 1366 | *badPtr = ptr; | |
| 1367 | return 0; | |
| 1368 | } | |
| 1369 | break; | |
| 1370 | } | |
| 1371 | } | |
| 1372 | return 1; | |
| 1373 | } | |
| 1374 | ||
| 1375 | /* This must only be called for a well-formed start-tag or empty element tag. | |
| 1376 | Returns the number of attributes. Pointers to the first attsMax attributes | |
| 1377 | are stored in atts. */ | |
| 1378 | ||
| 1379 | static | |
| 1380 | int PREFIX(getAtts)(const ENCODING *enc, const char *ptr, | |
| 1381 | int attsMax, ATTRIBUTE *atts) | |
| 1382 | { | |
| 1383 | enum { other, inName, inValue } state = inName; | |
| 1384 | int nAtts = 0; | |
| 3127 | 1385 | int open = 0; |
| 2086 | 1386 | |
| 1387 | for (ptr += MINBPC(enc);; ptr += MINBPC(enc)) { | |
| 1388 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1389 | #define START_NAME \ | |
| 1390 | if (state == other) { \ | |
| 1391 | if (nAtts < attsMax) { \ | |
| 1392 | atts[nAtts].name = ptr; \ | |
| 1393 | atts[nAtts].normalized = 1; \ | |
| 1394 | } \ | |
| 1395 | state = inName; \ | |
| 1396 | } | |
| 1397 | #define LEAD_CASE(n) \ | |
| 1398 | case BT_LEAD ## n: START_NAME ptr += (n - MINBPC(enc)); break; | |
| 1399 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 1400 | #undef LEAD_CASE | |
| 1401 | case BT_NONASCII: | |
| 1402 | case BT_NMSTRT: | |
| 1403 | case BT_HEX: | |
| 1404 | START_NAME | |
| 1405 | break; | |
| 1406 | #undef START_NAME | |
| 1407 | case BT_QUOT: | |
| 1408 | if (state != inValue) { | |
| 1409 | if (nAtts < attsMax) | |
| 1410 | atts[nAtts].valuePtr = ptr + MINBPC(enc); | |
| 1411 | state = inValue; | |
| 1412 | open = BT_QUOT; | |
| 1413 | } | |
| 1414 | else if (open == BT_QUOT) { | |
| 1415 | state = other; | |
| 1416 | if (nAtts < attsMax) | |
| 1417 | atts[nAtts].valueEnd = ptr; | |
| 1418 | nAtts++; | |
| 1419 | } | |
| 1420 | break; | |
| 1421 | case BT_APOS: | |
| 1422 | if (state != inValue) { | |
| 1423 | if (nAtts < attsMax) | |
| 1424 | atts[nAtts].valuePtr = ptr + MINBPC(enc); | |
| 1425 | state = inValue; | |
| 1426 | open = BT_APOS; | |
| 1427 | } | |
| 1428 | else if (open == BT_APOS) { | |
| 1429 | state = other; | |
| 1430 | if (nAtts < attsMax) | |
| 1431 | atts[nAtts].valueEnd = ptr; | |
| 1432 | nAtts++; | |
| 1433 | } | |
| 1434 | break; | |
| 1435 | case BT_AMP: | |
| 1436 | if (nAtts < attsMax) | |
| 1437 | atts[nAtts].normalized = 0; | |
| 1438 | break; | |
| 1439 | case BT_S: | |
| 1440 | if (state == inName) | |
| 1441 | state = other; | |
| 1442 | else if (state == inValue | |
| 1443 | && nAtts < attsMax | |
| 1444 | && atts[nAtts].normalized | |
| 1445 | && (ptr == atts[nAtts].valuePtr | |
| 1446 | || BYTE_TO_ASCII(enc, ptr) != ' ' | |
| 1447 | || BYTE_TO_ASCII(enc, ptr + MINBPC(enc)) == ' ' | |
| 1448 | || BYTE_TYPE(enc, ptr + MINBPC(enc)) == open)) | |
| 1449 | atts[nAtts].normalized = 0; | |
| 1450 | break; | |
| 1451 | case BT_CR: case BT_LF: | |
| 1452 | /* This case ensures that the first attribute name is counted | |
| 1453 | Apart from that we could just change state on the quote. */ | |
| 1454 | if (state == inName) | |
| 1455 | state = other; | |
| 1456 | else if (state == inValue && nAtts < attsMax) | |
| 1457 | atts[nAtts].normalized = 0; | |
| 1458 | break; | |
| 1459 | case BT_GT: | |
| 1460 | case BT_SOL: | |
| 1461 | if (state != inValue) | |
| 1462 | return nAtts; | |
| 1463 | break; | |
| 1464 | default: | |
| 1465 | break; | |
| 1466 | } | |
| 1467 | } | |
| 1468 | /* not reached */ | |
| 1469 | } | |
| 1470 | ||
| 1471 | static | |
| 1472 | int PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr) | |
| 1473 | { | |
| 1474 | int result = 0; | |
| 1475 | /* skip &# */ | |
| 1476 | ptr += 2*MINBPC(enc); | |
| 1477 | if (CHAR_MATCHES(enc, ptr, 'x')) { | |
| 1478 | for (ptr += MINBPC(enc); !CHAR_MATCHES(enc, ptr, ';'); ptr += MINBPC(enc)) { | |
| 1479 | int c = BYTE_TO_ASCII(enc, ptr); | |
| 1480 | switch (c) { | |
| 1481 | case '0': case '1': case '2': case '3': case '4': | |
| 1482 | case '5': case '6': case '7': case '8': case '9': | |
| 1483 | result <<= 4; | |
| 1484 | result |= (c - '0'); | |
| 1485 | break; | |
| 1486 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
| 1487 | result <<= 4; | |
| 1488 | result += 10 + (c - 'A'); | |
| 1489 | break; | |
| 1490 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': | |
| 1491 | result <<= 4; | |
| 1492 | result += 10 + (c - 'a'); | |
| 1493 | break; | |
| 1494 | } | |
| 1495 | if (result >= 0x110000) | |
| 1496 | return -1; | |
| 1497 | } | |
| 1498 | } | |
| 1499 | else { | |
| 1500 | for (; !CHAR_MATCHES(enc, ptr, ';'); ptr += MINBPC(enc)) { | |
| 1501 | int c = BYTE_TO_ASCII(enc, ptr); | |
| 1502 | result *= 10; | |
| 1503 | result += (c - '0'); | |
| 1504 | if (result >= 0x110000) | |
| 1505 | return -1; | |
| 1506 | } | |
| 1507 | } | |
| 1508 | return checkCharRefNumber(result); | |
| 1509 | } | |
| 1510 | ||
| 1511 | static | |
| 1512 | int PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr, const char *end) | |
| 1513 | { | |
| 1514 | switch ((end - ptr)/MINBPC(enc)) { | |
| 1515 | case 2: | |
| 1516 | if (CHAR_MATCHES(enc, ptr + MINBPC(enc), 't')) { | |
| 1517 | switch (BYTE_TO_ASCII(enc, ptr)) { | |
| 1518 | case 'l': | |
| 1519 | return '<'; | |
| 1520 | case 'g': | |
| 1521 | return '>'; | |
| 1522 | } | |
| 1523 | } | |
| 1524 | break; | |
| 1525 | case 3: | |
| 1526 | if (CHAR_MATCHES(enc, ptr, 'a')) { | |
| 1527 | ptr += MINBPC(enc); | |
| 1528 | if (CHAR_MATCHES(enc, ptr, 'm')) { | |
| 1529 | ptr += MINBPC(enc); | |
| 1530 | if (CHAR_MATCHES(enc, ptr, 'p')) | |
| 1531 | return '&'; | |
| 1532 | } | |
| 1533 | } | |
| 1534 | break; | |
| 1535 | case 4: | |
| 1536 | switch (BYTE_TO_ASCII(enc, ptr)) { | |
| 1537 | case 'q': | |
| 1538 | ptr += MINBPC(enc); | |
| 1539 | if (CHAR_MATCHES(enc, ptr, 'u')) { | |
| 1540 | ptr += MINBPC(enc); | |
| 1541 | if (CHAR_MATCHES(enc, ptr, 'o')) { | |
| 1542 | ptr += MINBPC(enc); | |
| 1543 | if (CHAR_MATCHES(enc, ptr, 't')) | |
| 1544 | return '"'; | |
| 1545 | } | |
| 1546 | } | |
| 1547 | break; | |
| 1548 | case 'a': | |
| 1549 | ptr += MINBPC(enc); | |
| 1550 | if (CHAR_MATCHES(enc, ptr, 'p')) { | |
| 1551 | ptr += MINBPC(enc); | |
| 1552 | if (CHAR_MATCHES(enc, ptr, 'o')) { | |
| 1553 | ptr += MINBPC(enc); | |
| 1554 | if (CHAR_MATCHES(enc, ptr, 's')) | |
| 1555 | return '\''; | |
| 1556 | } | |
| 1557 | } | |
| 1558 | break; | |
| 1559 | } | |
| 1560 | } | |
| 1561 | return 0; | |
| 1562 | } | |
| 1563 | ||
| 1564 | static | |
| 1565 | int PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2) | |
| 1566 | { | |
| 1567 | for (;;) { | |
| 1568 | switch (BYTE_TYPE(enc, ptr1)) { | |
| 1569 | #define LEAD_CASE(n) \ | |
| 1570 | case BT_LEAD ## n: \ | |
| 1571 | if (*ptr1++ != *ptr2++) \ | |
| 1572 | return 0; | |
| 1573 | LEAD_CASE(4) LEAD_CASE(3) LEAD_CASE(2) | |
| 1574 | #undef LEAD_CASE | |
| 1575 | /* fall through */ | |
| 1576 | if (*ptr1++ != *ptr2++) | |
| 1577 | return 0; | |
| 1578 | break; | |
| 1579 | case BT_NONASCII: | |
| 1580 | case BT_NMSTRT: | |
| 1581 | #ifdef XML_NS | |
| 1582 | case BT_COLON: | |
| 1583 | #endif | |
| 1584 | case BT_HEX: | |
| 1585 | case BT_DIGIT: | |
| 1586 | case BT_NAME: | |
| 1587 | case BT_MINUS: | |
| 1588 | if (*ptr2++ != *ptr1++) | |
| 1589 | return 0; | |
| 1590 | if (MINBPC(enc) > 1) { | |
| 1591 | if (*ptr2++ != *ptr1++) | |
| 1592 | return 0; | |
| 1593 | if (MINBPC(enc) > 2) { | |
| 1594 | if (*ptr2++ != *ptr1++) | |
| 1595 | return 0; | |
| 1596 | if (MINBPC(enc) > 3) { | |
| 1597 | if (*ptr2++ != *ptr1++) | |
| 1598 | return 0; | |
| 1599 | } | |
| 1600 | } | |
| 1601 | } | |
| 1602 | break; | |
| 1603 | default: | |
| 1604 | if (MINBPC(enc) == 1 && *ptr1 == *ptr2) | |
| 1605 | return 1; | |
| 1606 | switch (BYTE_TYPE(enc, ptr2)) { | |
| 1607 | case BT_LEAD2: | |
| 1608 | case BT_LEAD3: | |
| 1609 | case BT_LEAD4: | |
| 1610 | case BT_NONASCII: | |
| 1611 | case BT_NMSTRT: | |
| 1612 | #ifdef XML_NS | |
| 1613 | case BT_COLON: | |
| 1614 | #endif | |
| 1615 | case BT_HEX: | |
| 1616 | case BT_DIGIT: | |
| 1617 | case BT_NAME: | |
| 1618 | case BT_MINUS: | |
| 1619 | return 0; | |
| 1620 | default: | |
| 1621 | return 1; | |
| 1622 | } | |
| 1623 | } | |
| 1624 | } | |
| 1625 | /* not reached */ | |
| 1626 | } | |
| 1627 | ||
| 1628 | static | |
| 1629 | int PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1, const char *ptr2) | |
| 1630 | { | |
| 1631 | for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) { | |
| 1632 | if (!CHAR_MATCHES(enc, ptr1, *ptr2)) | |
| 1633 | return 0; | |
| 1634 | } | |
| 1635 | switch (BYTE_TYPE(enc, ptr1)) { | |
| 1636 | case BT_LEAD2: | |
| 1637 | case BT_LEAD3: | |
| 1638 | case BT_LEAD4: | |
| 1639 | case BT_NONASCII: | |
| 1640 | case BT_NMSTRT: | |
| 1641 | #ifdef XML_NS | |
| 1642 | case BT_COLON: | |
| 1643 | #endif | |
| 1644 | case BT_HEX: | |
| 1645 | case BT_DIGIT: | |
| 1646 | case BT_NAME: | |
| 1647 | case BT_MINUS: | |
| 1648 | return 0; | |
| 1649 | default: | |
| 1650 | return 1; | |
| 1651 | } | |
| 1652 | } | |
| 1653 | ||
| 1654 | static | |
| 1655 | int PREFIX(nameLength)(const ENCODING *enc, const char *ptr) | |
| 1656 | { | |
| 1657 | const char *start = ptr; | |
| 1658 | for (;;) { | |
| 1659 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1660 | #define LEAD_CASE(n) \ | |
| 1661 | case BT_LEAD ## n: ptr += n; break; | |
| 1662 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 1663 | #undef LEAD_CASE | |
| 1664 | case BT_NONASCII: | |
| 1665 | case BT_NMSTRT: | |
| 1666 | #ifdef XML_NS | |
| 1667 | case BT_COLON: | |
| 1668 | #endif | |
| 1669 | case BT_HEX: | |
| 1670 | case BT_DIGIT: | |
| 1671 | case BT_NAME: | |
| 1672 | case BT_MINUS: | |
| 1673 | ptr += MINBPC(enc); | |
| 1674 | break; | |
| 1675 | default: | |
| 1676 | return ptr - start; | |
| 1677 | } | |
| 1678 | } | |
| 1679 | } | |
| 1680 | ||
| 1681 | static | |
| 1682 | const char *PREFIX(skipS)(const ENCODING *enc, const char *ptr) | |
| 1683 | { | |
| 1684 | for (;;) { | |
| 1685 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1686 | case BT_LF: | |
| 1687 | case BT_CR: | |
| 1688 | case BT_S: | |
| 1689 | ptr += MINBPC(enc); | |
| 1690 | break; | |
| 1691 | default: | |
| 1692 | return ptr; | |
| 1693 | } | |
| 1694 | } | |
| 1695 | } | |
| 1696 | ||
| 1697 | static | |
| 1698 | void PREFIX(updatePosition)(const ENCODING *enc, | |
| 1699 | const char *ptr, | |
| 1700 | const char *end, | |
| 1701 | POSITION *pos) | |
| 1702 | { | |
| 1703 | while (ptr != end) { | |
| 1704 | switch (BYTE_TYPE(enc, ptr)) { | |
| 1705 | #define LEAD_CASE(n) \ | |
| 1706 | case BT_LEAD ## n: \ | |
| 1707 | ptr += n; \ | |
| 1708 | break; | |
| 1709 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) | |
| 1710 | #undef LEAD_CASE | |
| 1711 | case BT_LF: | |
| 1712 | pos->columnNumber = (unsigned)-1; | |
| 1713 | pos->lineNumber++; | |
| 1714 | ptr += MINBPC(enc); | |
| 1715 | break; | |
| 1716 | case BT_CR: | |
| 1717 | pos->lineNumber++; | |
| 1718 | ptr += MINBPC(enc); | |
| 1719 | if (ptr != end && BYTE_TYPE(enc, ptr) == BT_LF) | |
| 1720 | ptr += MINBPC(enc); | |
| 1721 | pos->columnNumber = (unsigned)-1; | |
| 1722 | break; | |
| 1723 | default: | |
| 1724 | ptr += MINBPC(enc); | |
| 1725 | break; | |
| 1726 | } | |
| 1727 | pos->columnNumber++; | |
| 1728 | } | |
| 1729 | } | |
| 1730 | ||
| 1731 | #undef DO_LEAD_CASE | |
| 1732 | #undef MULTIBYTE_CASES | |
| 1733 | #undef INVALID_CASES | |
| 1734 | #undef CHECK_NAME_CASE | |
| 1735 | #undef CHECK_NAME_CASES | |
| 1736 | #undef CHECK_NMSTRT_CASE | |
| 1737 | #undef CHECK_NMSTRT_CASES |