Fri, 11 Oct 2002 03:14:01 +0000
[gaim-migrate @ 3753]
Yeah this will probably break a lot of shit knowing my luck. But hey, I really don't care what people thnk.
| 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 | #include "xmldef.h" | |
| 23 | #include "xmlrole.h" | |
| 24 | ||
| 25 | /* Doesn't check: | |
| 26 | ||
| 27 | that ,| are not mixed in a model group | |
| 28 | content of literals | |
| 29 | ||
| 30 | */ | |
| 31 | ||
| 32 | #ifndef MIN_BYTES_PER_CHAR | |
| 33 | #define MIN_BYTES_PER_CHAR(enc) ((enc)->minBytesPerChar) | |
| 34 | #endif | |
| 35 | ||
| 36 | typedef int PROLOG_HANDLER(struct prolog_state *state, | |
| 37 | int tok, | |
| 38 | const char *ptr, | |
| 39 | const char *end, | |
| 40 | const ENCODING *enc); | |
| 41 | ||
| 42 | static PROLOG_HANDLER | |
| 43 | prolog0, prolog1, prolog2, | |
| 44 | doctype0, doctype1, doctype2, doctype3, doctype4, doctype5, | |
| 45 | internalSubset, | |
| 46 | entity0, entity1, entity2, entity3, entity4, entity5, entity6, | |
| 47 | entity7, entity8, entity9, | |
| 48 | notation0, notation1, notation2, notation3, notation4, | |
| 49 | attlist0, attlist1, attlist2, attlist3, attlist4, attlist5, attlist6, | |
| 50 | attlist7, attlist8, attlist9, | |
| 51 | element0, element1, element2, element3, element4, element5, element6, | |
| 52 | element7, | |
| 53 | declClose, | |
| 54 | error; | |
| 55 | ||
| 56 | static | |
| 57 | int syntaxError(PROLOG_STATE *); | |
| 58 | ||
| 59 | static | |
| 60 | int prolog0(PROLOG_STATE *state, | |
| 61 | int tok, | |
| 62 | const char *ptr, | |
| 63 | const char *end, | |
| 64 | const ENCODING *enc) | |
| 65 | { | |
| 66 | switch (tok) { | |
| 67 | case XML_TOK_PROLOG_S: | |
| 68 | state->handler = prolog1; | |
| 69 | return XML_ROLE_NONE; | |
| 70 | case XML_TOK_XML_DECL: | |
| 71 | state->handler = prolog1; | |
| 72 | return XML_ROLE_XML_DECL; | |
| 73 | case XML_TOK_PI: | |
| 74 | state->handler = prolog1; | |
| 75 | return XML_ROLE_NONE; | |
| 76 | case XML_TOK_COMMENT: | |
| 77 | state->handler = prolog1; | |
| 78 | case XML_TOK_BOM: | |
| 79 | return XML_ROLE_NONE; | |
| 80 | case XML_TOK_DECL_OPEN: | |
| 81 | if (!XmlNameMatchesAscii(enc, | |
| 82 | ptr + 2 * MIN_BYTES_PER_CHAR(enc), | |
| 83 | "DOCTYPE")) | |
| 84 | break; | |
| 85 | state->handler = doctype0; | |
| 86 | return XML_ROLE_NONE; | |
| 87 | case XML_TOK_INSTANCE_START: | |
| 88 | state->handler = error; | |
| 89 | return XML_ROLE_INSTANCE_START; | |
| 90 | } | |
| 91 | return syntaxError(state); | |
| 92 | } | |
| 93 | ||
| 94 | static | |
| 95 | int prolog1(PROLOG_STATE *state, | |
| 96 | int tok, | |
| 97 | const char *ptr, | |
| 98 | const char *end, | |
| 99 | const ENCODING *enc) | |
| 100 | { | |
| 101 | switch (tok) { | |
| 102 | case XML_TOK_PROLOG_S: | |
| 103 | return XML_ROLE_NONE; | |
| 104 | case XML_TOK_PI: | |
| 105 | case XML_TOK_COMMENT: | |
| 106 | case XML_TOK_BOM: | |
| 107 | return XML_ROLE_NONE; | |
| 108 | case XML_TOK_DECL_OPEN: | |
| 109 | if (!XmlNameMatchesAscii(enc, | |
| 110 | ptr + 2 * MIN_BYTES_PER_CHAR(enc), | |
| 111 | "DOCTYPE")) | |
| 112 | break; | |
| 113 | state->handler = doctype0; | |
| 114 | return XML_ROLE_NONE; | |
| 115 | case XML_TOK_INSTANCE_START: | |
| 116 | state->handler = error; | |
| 117 | return XML_ROLE_INSTANCE_START; | |
| 118 | } | |
| 119 | return syntaxError(state); | |
| 120 | } | |
| 121 | ||
| 122 | static | |
| 123 | int prolog2(PROLOG_STATE *state, | |
| 124 | int tok, | |
| 125 | const char *ptr, | |
| 126 | const char *end, | |
| 127 | const ENCODING *enc) | |
| 128 | { | |
| 129 | switch (tok) { | |
| 130 | case XML_TOK_PROLOG_S: | |
| 131 | return XML_ROLE_NONE; | |
| 132 | case XML_TOK_PI: | |
| 133 | case XML_TOK_COMMENT: | |
| 134 | return XML_ROLE_NONE; | |
| 135 | case XML_TOK_INSTANCE_START: | |
| 136 | state->handler = error; | |
| 137 | return XML_ROLE_INSTANCE_START; | |
| 138 | } | |
| 139 | return syntaxError(state); | |
| 140 | } | |
| 141 | ||
| 142 | static | |
| 143 | int doctype0(PROLOG_STATE *state, | |
| 144 | int tok, | |
| 145 | const char *ptr, | |
| 146 | const char *end, | |
| 147 | const ENCODING *enc) | |
| 148 | { | |
| 149 | switch (tok) { | |
| 150 | case XML_TOK_PROLOG_S: | |
| 151 | return XML_ROLE_NONE; | |
| 152 | case XML_TOK_NAME: | |
| 153 | case XML_TOK_PREFIXED_NAME: | |
| 154 | state->handler = doctype1; | |
| 155 | return XML_ROLE_DOCTYPE_NAME; | |
| 156 | } | |
| 157 | return syntaxError(state); | |
| 158 | } | |
| 159 | ||
| 160 | static | |
| 161 | int doctype1(PROLOG_STATE *state, | |
| 162 | int tok, | |
| 163 | const char *ptr, | |
| 164 | const char *end, | |
| 165 | const ENCODING *enc) | |
| 166 | { | |
| 167 | switch (tok) { | |
| 168 | case XML_TOK_PROLOG_S: | |
| 169 | return XML_ROLE_NONE; | |
| 170 | case XML_TOK_OPEN_BRACKET: | |
| 171 | state->handler = internalSubset; | |
| 172 | return XML_ROLE_NONE; | |
| 173 | case XML_TOK_DECL_CLOSE: | |
| 174 | state->handler = prolog2; | |
| 175 | return XML_ROLE_DOCTYPE_CLOSE; | |
| 176 | case XML_TOK_NAME: | |
| 177 | if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) { | |
| 178 | state->handler = doctype3; | |
| 179 | return XML_ROLE_NONE; | |
| 180 | } | |
| 181 | if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) { | |
| 182 | state->handler = doctype2; | |
| 183 | return XML_ROLE_NONE; | |
| 184 | } | |
| 185 | break; | |
| 186 | } | |
| 187 | return syntaxError(state); | |
| 188 | } | |
| 189 | ||
| 190 | static | |
| 191 | int doctype2(PROLOG_STATE *state, | |
| 192 | int tok, | |
| 193 | const char *ptr, | |
| 194 | const char *end, | |
| 195 | const ENCODING *enc) | |
| 196 | { | |
| 197 | switch (tok) { | |
| 198 | case XML_TOK_PROLOG_S: | |
| 199 | return XML_ROLE_NONE; | |
| 200 | case XML_TOK_LITERAL: | |
| 201 | state->handler = doctype3; | |
| 202 | return XML_ROLE_DOCTYPE_PUBLIC_ID; | |
| 203 | } | |
| 204 | return syntaxError(state); | |
| 205 | } | |
| 206 | ||
| 207 | static | |
| 208 | int doctype3(PROLOG_STATE *state, | |
| 209 | int tok, | |
| 210 | const char *ptr, | |
| 211 | const char *end, | |
| 212 | const ENCODING *enc) | |
| 213 | { | |
| 214 | switch (tok) { | |
| 215 | case XML_TOK_PROLOG_S: | |
| 216 | return XML_ROLE_NONE; | |
| 217 | case XML_TOK_LITERAL: | |
| 218 | state->handler = doctype4; | |
| 219 | return XML_ROLE_DOCTYPE_SYSTEM_ID; | |
| 220 | } | |
| 221 | return syntaxError(state); | |
| 222 | } | |
| 223 | ||
| 224 | static | |
| 225 | int doctype4(PROLOG_STATE *state, | |
| 226 | int tok, | |
| 227 | const char *ptr, | |
| 228 | const char *end, | |
| 229 | const ENCODING *enc) | |
| 230 | { | |
| 231 | switch (tok) { | |
| 232 | case XML_TOK_PROLOG_S: | |
| 233 | return XML_ROLE_NONE; | |
| 234 | case XML_TOK_OPEN_BRACKET: | |
| 235 | state->handler = internalSubset; | |
| 236 | return XML_ROLE_NONE; | |
| 237 | case XML_TOK_DECL_CLOSE: | |
| 238 | state->handler = prolog2; | |
| 239 | return XML_ROLE_DOCTYPE_CLOSE; | |
| 240 | } | |
| 241 | return syntaxError(state); | |
| 242 | } | |
| 243 | ||
| 244 | static | |
| 245 | int doctype5(PROLOG_STATE *state, | |
| 246 | int tok, | |
| 247 | const char *ptr, | |
| 248 | const char *end, | |
| 249 | const ENCODING *enc) | |
| 250 | { | |
| 251 | switch (tok) { | |
| 252 | case XML_TOK_PROLOG_S: | |
| 253 | return XML_ROLE_NONE; | |
| 254 | case XML_TOK_DECL_CLOSE: | |
| 255 | state->handler = prolog2; | |
| 256 | return XML_ROLE_DOCTYPE_CLOSE; | |
| 257 | } | |
| 258 | return syntaxError(state); | |
| 259 | } | |
| 260 | ||
| 261 | static | |
| 262 | int internalSubset(PROLOG_STATE *state, | |
| 263 | int tok, | |
| 264 | const char *ptr, | |
| 265 | const char *end, | |
| 266 | const ENCODING *enc) | |
| 267 | { | |
| 268 | switch (tok) { | |
| 269 | case XML_TOK_PROLOG_S: | |
| 270 | return XML_ROLE_NONE; | |
| 271 | case XML_TOK_DECL_OPEN: | |
| 272 | if (XmlNameMatchesAscii(enc, | |
| 273 | ptr + 2 * MIN_BYTES_PER_CHAR(enc), | |
| 274 | "ENTITY")) { | |
| 275 | state->handler = entity0; | |
| 276 | return XML_ROLE_NONE; | |
| 277 | } | |
| 278 | if (XmlNameMatchesAscii(enc, | |
| 279 | ptr + 2 * MIN_BYTES_PER_CHAR(enc), | |
| 280 | "ATTLIST")) { | |
| 281 | state->handler = attlist0; | |
| 282 | return XML_ROLE_NONE; | |
| 283 | } | |
| 284 | if (XmlNameMatchesAscii(enc, | |
| 285 | ptr + 2 * MIN_BYTES_PER_CHAR(enc), | |
| 286 | "ELEMENT")) { | |
| 287 | state->handler = element0; | |
| 288 | return XML_ROLE_NONE; | |
| 289 | } | |
| 290 | if (XmlNameMatchesAscii(enc, | |
| 291 | ptr + 2 * MIN_BYTES_PER_CHAR(enc), | |
| 292 | "NOTATION")) { | |
| 293 | state->handler = notation0; | |
| 294 | return XML_ROLE_NONE; | |
| 295 | } | |
| 296 | break; | |
| 297 | case XML_TOK_PI: | |
| 298 | case XML_TOK_COMMENT: | |
| 299 | return XML_ROLE_NONE; | |
| 300 | case XML_TOK_PARAM_ENTITY_REF: | |
| 301 | return XML_ROLE_PARAM_ENTITY_REF; | |
| 302 | case XML_TOK_CLOSE_BRACKET: | |
| 303 | state->handler = doctype5; | |
| 304 | return XML_ROLE_NONE; | |
| 305 | } | |
| 306 | return syntaxError(state); | |
| 307 | } | |
| 308 | ||
| 309 | static | |
| 310 | int entity0(PROLOG_STATE *state, | |
| 311 | int tok, | |
| 312 | const char *ptr, | |
| 313 | const char *end, | |
| 314 | const ENCODING *enc) | |
| 315 | { | |
| 316 | switch (tok) { | |
| 317 | case XML_TOK_PROLOG_S: | |
| 318 | return XML_ROLE_NONE; | |
| 319 | case XML_TOK_PERCENT: | |
| 320 | state->handler = entity1; | |
| 321 | return XML_ROLE_NONE; | |
| 322 | case XML_TOK_NAME: | |
| 323 | state->handler = entity2; | |
| 324 | return XML_ROLE_GENERAL_ENTITY_NAME; | |
| 325 | } | |
| 326 | return syntaxError(state); | |
| 327 | } | |
| 328 | ||
| 329 | static | |
| 330 | int entity1(PROLOG_STATE *state, | |
| 331 | int tok, | |
| 332 | const char *ptr, | |
| 333 | const char *end, | |
| 334 | const ENCODING *enc) | |
| 335 | { | |
| 336 | switch (tok) { | |
| 337 | case XML_TOK_PROLOG_S: | |
| 338 | return XML_ROLE_NONE; | |
| 339 | case XML_TOK_NAME: | |
| 340 | state->handler = entity7; | |
| 341 | return XML_ROLE_PARAM_ENTITY_NAME; | |
| 342 | } | |
| 343 | return syntaxError(state); | |
| 344 | } | |
| 345 | ||
| 346 | static | |
| 347 | int entity2(PROLOG_STATE *state, | |
| 348 | int tok, | |
| 349 | const char *ptr, | |
| 350 | const char *end, | |
| 351 | const ENCODING *enc) | |
| 352 | { | |
| 353 | switch (tok) { | |
| 354 | case XML_TOK_PROLOG_S: | |
| 355 | return XML_ROLE_NONE; | |
| 356 | case XML_TOK_NAME: | |
| 357 | if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) { | |
| 358 | state->handler = entity4; | |
| 359 | return XML_ROLE_NONE; | |
| 360 | } | |
| 361 | if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) { | |
| 362 | state->handler = entity3; | |
| 363 | return XML_ROLE_NONE; | |
| 364 | } | |
| 365 | break; | |
| 366 | case XML_TOK_LITERAL: | |
| 367 | state->handler = declClose; | |
| 368 | return XML_ROLE_ENTITY_VALUE; | |
| 369 | } | |
| 370 | return syntaxError(state); | |
| 371 | } | |
| 372 | ||
| 373 | static | |
| 374 | int entity3(PROLOG_STATE *state, | |
| 375 | int tok, | |
| 376 | const char *ptr, | |
| 377 | const char *end, | |
| 378 | const ENCODING *enc) | |
| 379 | { | |
| 380 | switch (tok) { | |
| 381 | case XML_TOK_PROLOG_S: | |
| 382 | return XML_ROLE_NONE; | |
| 383 | case XML_TOK_LITERAL: | |
| 384 | state->handler = entity4; | |
| 385 | return XML_ROLE_ENTITY_PUBLIC_ID; | |
| 386 | } | |
| 387 | return syntaxError(state); | |
| 388 | } | |
| 389 | ||
| 390 | ||
| 391 | static | |
| 392 | int entity4(PROLOG_STATE *state, | |
| 393 | int tok, | |
| 394 | const char *ptr, | |
| 395 | const char *end, | |
| 396 | const ENCODING *enc) | |
| 397 | { | |
| 398 | switch (tok) { | |
| 399 | case XML_TOK_PROLOG_S: | |
| 400 | return XML_ROLE_NONE; | |
| 401 | case XML_TOK_LITERAL: | |
| 402 | state->handler = entity5; | |
| 403 | return XML_ROLE_ENTITY_SYSTEM_ID; | |
| 404 | } | |
| 405 | return syntaxError(state); | |
| 406 | } | |
| 407 | ||
| 408 | static | |
| 409 | int entity5(PROLOG_STATE *state, | |
| 410 | int tok, | |
| 411 | const char *ptr, | |
| 412 | const char *end, | |
| 413 | const ENCODING *enc) | |
| 414 | { | |
| 415 | switch (tok) { | |
| 416 | case XML_TOK_PROLOG_S: | |
| 417 | return XML_ROLE_NONE; | |
| 418 | case XML_TOK_DECL_CLOSE: | |
| 419 | state->handler = internalSubset; | |
| 420 | return XML_ROLE_NONE; | |
| 421 | case XML_TOK_NAME: | |
| 422 | if (XmlNameMatchesAscii(enc, ptr, "NDATA")) { | |
| 423 | state->handler = entity6; | |
| 424 | return XML_ROLE_NONE; | |
| 425 | } | |
| 426 | break; | |
| 427 | } | |
| 428 | return syntaxError(state); | |
| 429 | } | |
| 430 | ||
| 431 | static | |
| 432 | int entity6(PROLOG_STATE *state, | |
| 433 | int tok, | |
| 434 | const char *ptr, | |
| 435 | const char *end, | |
| 436 | const ENCODING *enc) | |
| 437 | { | |
| 438 | switch (tok) { | |
| 439 | case XML_TOK_PROLOG_S: | |
| 440 | return XML_ROLE_NONE; | |
| 441 | case XML_TOK_NAME: | |
| 442 | state->handler = declClose; | |
| 443 | return XML_ROLE_ENTITY_NOTATION_NAME; | |
| 444 | } | |
| 445 | return syntaxError(state); | |
| 446 | } | |
| 447 | ||
| 448 | static | |
| 449 | int entity7(PROLOG_STATE *state, | |
| 450 | int tok, | |
| 451 | const char *ptr, | |
| 452 | const char *end, | |
| 453 | const ENCODING *enc) | |
| 454 | { | |
| 455 | switch (tok) { | |
| 456 | case XML_TOK_PROLOG_S: | |
| 457 | return XML_ROLE_NONE; | |
| 458 | case XML_TOK_NAME: | |
| 459 | if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) { | |
| 460 | state->handler = entity9; | |
| 461 | return XML_ROLE_NONE; | |
| 462 | } | |
| 463 | if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) { | |
| 464 | state->handler = entity8; | |
| 465 | return XML_ROLE_NONE; | |
| 466 | } | |
| 467 | break; | |
| 468 | case XML_TOK_LITERAL: | |
| 469 | state->handler = declClose; | |
| 470 | return XML_ROLE_ENTITY_VALUE; | |
| 471 | } | |
| 472 | return syntaxError(state); | |
| 473 | } | |
| 474 | ||
| 475 | static | |
| 476 | int entity8(PROLOG_STATE *state, | |
| 477 | int tok, | |
| 478 | const char *ptr, | |
| 479 | const char *end, | |
| 480 | const ENCODING *enc) | |
| 481 | { | |
| 482 | switch (tok) { | |
| 483 | case XML_TOK_PROLOG_S: | |
| 484 | return XML_ROLE_NONE; | |
| 485 | case XML_TOK_LITERAL: | |
| 486 | state->handler = entity9; | |
| 487 | return XML_ROLE_ENTITY_PUBLIC_ID; | |
| 488 | } | |
| 489 | return syntaxError(state); | |
| 490 | } | |
| 491 | ||
| 492 | static | |
| 493 | int entity9(PROLOG_STATE *state, | |
| 494 | int tok, | |
| 495 | const char *ptr, | |
| 496 | const char *end, | |
| 497 | const ENCODING *enc) | |
| 498 | { | |
| 499 | switch (tok) { | |
| 500 | case XML_TOK_PROLOG_S: | |
| 501 | return XML_ROLE_NONE; | |
| 502 | case XML_TOK_LITERAL: | |
| 503 | state->handler = declClose; | |
| 504 | return XML_ROLE_ENTITY_SYSTEM_ID; | |
| 505 | } | |
| 506 | return syntaxError(state); | |
| 507 | } | |
| 508 | ||
| 509 | static | |
| 510 | int notation0(PROLOG_STATE *state, | |
| 511 | int tok, | |
| 512 | const char *ptr, | |
| 513 | const char *end, | |
| 514 | const ENCODING *enc) | |
| 515 | { | |
| 516 | switch (tok) { | |
| 517 | case XML_TOK_PROLOG_S: | |
| 518 | return XML_ROLE_NONE; | |
| 519 | case XML_TOK_NAME: | |
| 520 | state->handler = notation1; | |
| 521 | return XML_ROLE_NOTATION_NAME; | |
| 522 | } | |
| 523 | return syntaxError(state); | |
| 524 | } | |
| 525 | ||
| 526 | static | |
| 527 | int notation1(PROLOG_STATE *state, | |
| 528 | int tok, | |
| 529 | const char *ptr, | |
| 530 | const char *end, | |
| 531 | const ENCODING *enc) | |
| 532 | { | |
| 533 | switch (tok) { | |
| 534 | case XML_TOK_PROLOG_S: | |
| 535 | return XML_ROLE_NONE; | |
| 536 | case XML_TOK_NAME: | |
| 537 | if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) { | |
| 538 | state->handler = notation3; | |
| 539 | return XML_ROLE_NONE; | |
| 540 | } | |
| 541 | if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) { | |
| 542 | state->handler = notation2; | |
| 543 | return XML_ROLE_NONE; | |
| 544 | } | |
| 545 | break; | |
| 546 | } | |
| 547 | return syntaxError(state); | |
| 548 | } | |
| 549 | ||
| 550 | static | |
| 551 | int notation2(PROLOG_STATE *state, | |
| 552 | int tok, | |
| 553 | const char *ptr, | |
| 554 | const char *end, | |
| 555 | const ENCODING *enc) | |
| 556 | { | |
| 557 | switch (tok) { | |
| 558 | case XML_TOK_PROLOG_S: | |
| 559 | return XML_ROLE_NONE; | |
| 560 | case XML_TOK_LITERAL: | |
| 561 | state->handler = notation4; | |
| 562 | return XML_ROLE_NOTATION_PUBLIC_ID; | |
| 563 | } | |
| 564 | return syntaxError(state); | |
| 565 | } | |
| 566 | ||
| 567 | static | |
| 568 | int notation3(PROLOG_STATE *state, | |
| 569 | int tok, | |
| 570 | const char *ptr, | |
| 571 | const char *end, | |
| 572 | const ENCODING *enc) | |
| 573 | { | |
| 574 | switch (tok) { | |
| 575 | case XML_TOK_PROLOG_S: | |
| 576 | return XML_ROLE_NONE; | |
| 577 | case XML_TOK_LITERAL: | |
| 578 | state->handler = declClose; | |
| 579 | return XML_ROLE_NOTATION_SYSTEM_ID; | |
| 580 | } | |
| 581 | return syntaxError(state); | |
| 582 | } | |
| 583 | ||
| 584 | static | |
| 585 | int notation4(PROLOG_STATE *state, | |
| 586 | int tok, | |
| 587 | const char *ptr, | |
| 588 | const char *end, | |
| 589 | const ENCODING *enc) | |
| 590 | { | |
| 591 | switch (tok) { | |
| 592 | case XML_TOK_PROLOG_S: | |
| 593 | return XML_ROLE_NONE; | |
| 594 | case XML_TOK_LITERAL: | |
| 595 | state->handler = declClose; | |
| 596 | return XML_ROLE_NOTATION_SYSTEM_ID; | |
| 597 | case XML_TOK_DECL_CLOSE: | |
| 598 | state->handler = internalSubset; | |
| 599 | return XML_ROLE_NOTATION_NO_SYSTEM_ID; | |
| 600 | } | |
| 601 | return syntaxError(state); | |
| 602 | } | |
| 603 | ||
| 604 | static | |
| 605 | int attlist0(PROLOG_STATE *state, | |
| 606 | int tok, | |
| 607 | const char *ptr, | |
| 608 | const char *end, | |
| 609 | const ENCODING *enc) | |
| 610 | { | |
| 611 | switch (tok) { | |
| 612 | case XML_TOK_PROLOG_S: | |
| 613 | return XML_ROLE_NONE; | |
| 614 | case XML_TOK_NAME: | |
| 615 | case XML_TOK_PREFIXED_NAME: | |
| 616 | state->handler = attlist1; | |
| 617 | return XML_ROLE_ATTLIST_ELEMENT_NAME; | |
| 618 | } | |
| 619 | return syntaxError(state); | |
| 620 | } | |
| 621 | ||
| 622 | static | |
| 623 | int attlist1(PROLOG_STATE *state, | |
| 624 | int tok, | |
| 625 | const char *ptr, | |
| 626 | const char *end, | |
| 627 | const ENCODING *enc) | |
| 628 | { | |
| 629 | switch (tok) { | |
| 630 | case XML_TOK_PROLOG_S: | |
| 631 | return XML_ROLE_NONE; | |
| 632 | case XML_TOK_DECL_CLOSE: | |
| 633 | state->handler = internalSubset; | |
| 634 | return XML_ROLE_NONE; | |
| 635 | case XML_TOK_NAME: | |
| 636 | case XML_TOK_PREFIXED_NAME: | |
| 637 | state->handler = attlist2; | |
| 638 | return XML_ROLE_ATTRIBUTE_NAME; | |
| 639 | } | |
| 640 | return syntaxError(state); | |
| 641 | } | |
| 642 | ||
| 643 | static | |
| 644 | int attlist2(PROLOG_STATE *state, | |
| 645 | int tok, | |
| 646 | const char *ptr, | |
| 647 | const char *end, | |
| 648 | const ENCODING *enc) | |
| 649 | { | |
| 650 | switch (tok) { | |
| 651 | case XML_TOK_PROLOG_S: | |
| 652 | return XML_ROLE_NONE; | |
| 653 | case XML_TOK_NAME: | |
| 654 | { | |
| 655 | static const char *types[] = { | |
| 656 | "CDATA", | |
| 657 | "ID", | |
| 658 | "IDREF", | |
| 659 | "IDREFS", | |
| 660 | "ENTITY", | |
| 661 | "ENTITIES", | |
| 662 | "NMTOKEN", | |
| 663 | "NMTOKENS", | |
| 664 | }; | |
| 665 | int i; | |
| 666 | for (i = 0; i < (int)(sizeof(types)/sizeof(types[0])); i++) | |
| 667 | if (XmlNameMatchesAscii(enc, ptr, types[i])) { | |
| 668 | state->handler = attlist8; | |
| 669 | return XML_ROLE_ATTRIBUTE_TYPE_CDATA + i; | |
| 670 | } | |
| 671 | } | |
| 672 | if (XmlNameMatchesAscii(enc, ptr, "NOTATION")) { | |
| 673 | state->handler = attlist5; | |
| 674 | return XML_ROLE_NONE; | |
| 675 | } | |
| 676 | break; | |
| 677 | case XML_TOK_OPEN_PAREN: | |
| 678 | state->handler = attlist3; | |
| 679 | return XML_ROLE_NONE; | |
| 680 | } | |
| 681 | return syntaxError(state); | |
| 682 | } | |
| 683 | ||
| 684 | static | |
| 685 | int attlist3(PROLOG_STATE *state, | |
| 686 | int tok, | |
| 687 | const char *ptr, | |
| 688 | const char *end, | |
| 689 | const ENCODING *enc) | |
| 690 | { | |
| 691 | switch (tok) { | |
| 692 | case XML_TOK_PROLOG_S: | |
| 693 | return XML_ROLE_NONE; | |
| 694 | case XML_TOK_NMTOKEN: | |
| 695 | case XML_TOK_NAME: | |
| 696 | case XML_TOK_PREFIXED_NAME: | |
| 697 | state->handler = attlist4; | |
| 698 | return XML_ROLE_ATTRIBUTE_ENUM_VALUE; | |
| 699 | } | |
| 700 | return syntaxError(state); | |
| 701 | } | |
| 702 | ||
| 703 | static | |
| 704 | int attlist4(PROLOG_STATE *state, | |
| 705 | int tok, | |
| 706 | const char *ptr, | |
| 707 | const char *end, | |
| 708 | const ENCODING *enc) | |
| 709 | { | |
| 710 | switch (tok) { | |
| 711 | case XML_TOK_PROLOG_S: | |
| 712 | return XML_ROLE_NONE; | |
| 713 | case XML_TOK_CLOSE_PAREN: | |
| 714 | state->handler = attlist8; | |
| 715 | return XML_ROLE_NONE; | |
| 716 | case XML_TOK_OR: | |
| 717 | state->handler = attlist3; | |
| 718 | return XML_ROLE_NONE; | |
| 719 | } | |
| 720 | return syntaxError(state); | |
| 721 | } | |
| 722 | ||
| 723 | static | |
| 724 | int attlist5(PROLOG_STATE *state, | |
| 725 | int tok, | |
| 726 | const char *ptr, | |
| 727 | const char *end, | |
| 728 | const ENCODING *enc) | |
| 729 | { | |
| 730 | switch (tok) { | |
| 731 | case XML_TOK_PROLOG_S: | |
| 732 | return XML_ROLE_NONE; | |
| 733 | case XML_TOK_OPEN_PAREN: | |
| 734 | state->handler = attlist6; | |
| 735 | return XML_ROLE_NONE; | |
| 736 | } | |
| 737 | return syntaxError(state); | |
| 738 | } | |
| 739 | ||
| 740 | ||
| 741 | static | |
| 742 | int attlist6(PROLOG_STATE *state, | |
| 743 | int tok, | |
| 744 | const char *ptr, | |
| 745 | const char *end, | |
| 746 | const ENCODING *enc) | |
| 747 | { | |
| 748 | switch (tok) { | |
| 749 | case XML_TOK_PROLOG_S: | |
| 750 | return XML_ROLE_NONE; | |
| 751 | case XML_TOK_NAME: | |
| 752 | state->handler = attlist7; | |
| 753 | return XML_ROLE_ATTRIBUTE_NOTATION_VALUE; | |
| 754 | } | |
| 755 | return syntaxError(state); | |
| 756 | } | |
| 757 | ||
| 758 | static | |
| 759 | int attlist7(PROLOG_STATE *state, | |
| 760 | int tok, | |
| 761 | const char *ptr, | |
| 762 | const char *end, | |
| 763 | const ENCODING *enc) | |
| 764 | { | |
| 765 | switch (tok) { | |
| 766 | case XML_TOK_PROLOG_S: | |
| 767 | return XML_ROLE_NONE; | |
| 768 | case XML_TOK_CLOSE_PAREN: | |
| 769 | state->handler = attlist8; | |
| 770 | return XML_ROLE_NONE; | |
| 771 | case XML_TOK_OR: | |
| 772 | state->handler = attlist6; | |
| 773 | return XML_ROLE_NONE; | |
| 774 | } | |
| 775 | return syntaxError(state); | |
| 776 | } | |
| 777 | ||
| 778 | /* default value */ | |
| 779 | static | |
| 780 | int attlist8(PROLOG_STATE *state, | |
| 781 | int tok, | |
| 782 | const char *ptr, | |
| 783 | const char *end, | |
| 784 | const ENCODING *enc) | |
| 785 | { | |
| 786 | switch (tok) { | |
| 787 | case XML_TOK_PROLOG_S: | |
| 788 | return XML_ROLE_NONE; | |
| 789 | case XML_TOK_POUND_NAME: | |
| 790 | if (XmlNameMatchesAscii(enc, | |
| 791 | ptr + MIN_BYTES_PER_CHAR(enc), | |
| 792 | "IMPLIED")) { | |
| 793 | state->handler = attlist1; | |
| 794 | return XML_ROLE_IMPLIED_ATTRIBUTE_VALUE; | |
| 795 | } | |
| 796 | if (XmlNameMatchesAscii(enc, | |
| 797 | ptr + MIN_BYTES_PER_CHAR(enc), | |
| 798 | "REQUIRED")) { | |
| 799 | state->handler = attlist1; | |
| 800 | return XML_ROLE_REQUIRED_ATTRIBUTE_VALUE; | |
| 801 | } | |
| 802 | if (XmlNameMatchesAscii(enc, | |
| 803 | ptr + MIN_BYTES_PER_CHAR(enc), | |
| 804 | "FIXED")) { | |
| 805 | state->handler = attlist9; | |
| 806 | return XML_ROLE_NONE; | |
| 807 | } | |
| 808 | break; | |
| 809 | case XML_TOK_LITERAL: | |
| 810 | state->handler = attlist1; | |
| 811 | return XML_ROLE_DEFAULT_ATTRIBUTE_VALUE; | |
| 812 | } | |
| 813 | return syntaxError(state); | |
| 814 | } | |
| 815 | ||
| 816 | static | |
| 817 | int attlist9(PROLOG_STATE *state, | |
| 818 | int tok, | |
| 819 | const char *ptr, | |
| 820 | const char *end, | |
| 821 | const ENCODING *enc) | |
| 822 | { | |
| 823 | switch (tok) { | |
| 824 | case XML_TOK_PROLOG_S: | |
| 825 | return XML_ROLE_NONE; | |
| 826 | case XML_TOK_LITERAL: | |
| 827 | state->handler = attlist1; | |
| 828 | return XML_ROLE_FIXED_ATTRIBUTE_VALUE; | |
| 829 | } | |
| 830 | return syntaxError(state); | |
| 831 | } | |
| 832 | ||
| 833 | static | |
| 834 | int element0(PROLOG_STATE *state, | |
| 835 | int tok, | |
| 836 | const char *ptr, | |
| 837 | const char *end, | |
| 838 | const ENCODING *enc) | |
| 839 | { | |
| 840 | switch (tok) { | |
| 841 | case XML_TOK_PROLOG_S: | |
| 842 | return XML_ROLE_NONE; | |
| 843 | case XML_TOK_NAME: | |
| 844 | case XML_TOK_PREFIXED_NAME: | |
| 845 | state->handler = element1; | |
| 846 | return XML_ROLE_ELEMENT_NAME; | |
| 847 | } | |
| 848 | return syntaxError(state); | |
| 849 | } | |
| 850 | ||
| 851 | static | |
| 852 | int element1(PROLOG_STATE *state, | |
| 853 | int tok, | |
| 854 | const char *ptr, | |
| 855 | const char *end, | |
| 856 | const ENCODING *enc) | |
| 857 | { | |
| 858 | switch (tok) { | |
| 859 | case XML_TOK_PROLOG_S: | |
| 860 | return XML_ROLE_NONE; | |
| 861 | case XML_TOK_NAME: | |
| 862 | if (XmlNameMatchesAscii(enc, ptr, "EMPTY")) { | |
| 863 | state->handler = declClose; | |
| 864 | return XML_ROLE_CONTENT_EMPTY; | |
| 865 | } | |
| 866 | if (XmlNameMatchesAscii(enc, ptr, "ANY")) { | |
| 867 | state->handler = declClose; | |
| 868 | return XML_ROLE_CONTENT_ANY; | |
| 869 | } | |
| 870 | break; | |
| 871 | case XML_TOK_OPEN_PAREN: | |
| 872 | state->handler = element2; | |
| 873 | state->level = 1; | |
| 874 | return XML_ROLE_GROUP_OPEN; | |
| 875 | } | |
| 876 | return syntaxError(state); | |
| 877 | } | |
| 878 | ||
| 879 | static | |
| 880 | int element2(PROLOG_STATE *state, | |
| 881 | int tok, | |
| 882 | const char *ptr, | |
| 883 | const char *end, | |
| 884 | const ENCODING *enc) | |
| 885 | { | |
| 886 | switch (tok) { | |
| 887 | case XML_TOK_PROLOG_S: | |
| 888 | return XML_ROLE_NONE; | |
| 889 | case XML_TOK_POUND_NAME: | |
| 890 | if (XmlNameMatchesAscii(enc, | |
| 891 | ptr + MIN_BYTES_PER_CHAR(enc), | |
| 892 | "PCDATA")) { | |
| 893 | state->handler = element3; | |
| 894 | return XML_ROLE_CONTENT_PCDATA; | |
| 895 | } | |
| 896 | break; | |
| 897 | case XML_TOK_OPEN_PAREN: | |
| 898 | state->level = 2; | |
| 899 | state->handler = element6; | |
| 900 | return XML_ROLE_GROUP_OPEN; | |
| 901 | case XML_TOK_NAME: | |
| 902 | case XML_TOK_PREFIXED_NAME: | |
| 903 | state->handler = element7; | |
| 904 | return XML_ROLE_CONTENT_ELEMENT; | |
| 905 | case XML_TOK_NAME_QUESTION: | |
| 906 | state->handler = element7; | |
| 907 | return XML_ROLE_CONTENT_ELEMENT_OPT; | |
| 908 | case XML_TOK_NAME_ASTERISK: | |
| 909 | state->handler = element7; | |
| 910 | return XML_ROLE_CONTENT_ELEMENT_REP; | |
| 911 | case XML_TOK_NAME_PLUS: | |
| 912 | state->handler = element7; | |
| 913 | return XML_ROLE_CONTENT_ELEMENT_PLUS; | |
| 914 | } | |
| 915 | return syntaxError(state); | |
| 916 | } | |
| 917 | ||
| 918 | static | |
| 919 | int element3(PROLOG_STATE *state, | |
| 920 | int tok, | |
| 921 | const char *ptr, | |
| 922 | const char *end, | |
| 923 | const ENCODING *enc) | |
| 924 | { | |
| 925 | switch (tok) { | |
| 926 | case XML_TOK_PROLOG_S: | |
| 927 | return XML_ROLE_NONE; | |
| 928 | case XML_TOK_CLOSE_PAREN: | |
| 929 | case XML_TOK_CLOSE_PAREN_ASTERISK: | |
| 930 | state->handler = declClose; | |
| 931 | return XML_ROLE_GROUP_CLOSE_REP; | |
| 932 | case XML_TOK_OR: | |
| 933 | state->handler = element4; | |
| 934 | return XML_ROLE_NONE; | |
| 935 | } | |
| 936 | return syntaxError(state); | |
| 937 | } | |
| 938 | ||
| 939 | static | |
| 940 | int element4(PROLOG_STATE *state, | |
| 941 | int tok, | |
| 942 | const char *ptr, | |
| 943 | const char *end, | |
| 944 | const ENCODING *enc) | |
| 945 | { | |
| 946 | switch (tok) { | |
| 947 | case XML_TOK_PROLOG_S: | |
| 948 | return XML_ROLE_NONE; | |
| 949 | case XML_TOK_NAME: | |
| 950 | case XML_TOK_PREFIXED_NAME: | |
| 951 | state->handler = element5; | |
| 952 | return XML_ROLE_CONTENT_ELEMENT; | |
| 953 | } | |
| 954 | return syntaxError(state); | |
| 955 | } | |
| 956 | ||
| 957 | static | |
| 958 | int element5(PROLOG_STATE *state, | |
| 959 | int tok, | |
| 960 | const char *ptr, | |
| 961 | const char *end, | |
| 962 | const ENCODING *enc) | |
| 963 | { | |
| 964 | switch (tok) { | |
| 965 | case XML_TOK_PROLOG_S: | |
| 966 | return XML_ROLE_NONE; | |
| 967 | case XML_TOK_CLOSE_PAREN_ASTERISK: | |
| 968 | state->handler = declClose; | |
| 969 | return XML_ROLE_GROUP_CLOSE_REP; | |
| 970 | case XML_TOK_OR: | |
| 971 | state->handler = element4; | |
| 972 | return XML_ROLE_NONE; | |
| 973 | } | |
| 974 | return syntaxError(state); | |
| 975 | } | |
| 976 | ||
| 977 | static | |
| 978 | int element6(PROLOG_STATE *state, | |
| 979 | int tok, | |
| 980 | const char *ptr, | |
| 981 | const char *end, | |
| 982 | const ENCODING *enc) | |
| 983 | { | |
| 984 | switch (tok) { | |
| 985 | case XML_TOK_PROLOG_S: | |
| 986 | return XML_ROLE_NONE; | |
| 987 | case XML_TOK_OPEN_PAREN: | |
| 988 | state->level += 1; | |
| 989 | return XML_ROLE_GROUP_OPEN; | |
| 990 | case XML_TOK_NAME: | |
| 991 | case XML_TOK_PREFIXED_NAME: | |
| 992 | state->handler = element7; | |
| 993 | return XML_ROLE_CONTENT_ELEMENT; | |
| 994 | case XML_TOK_NAME_QUESTION: | |
| 995 | state->handler = element7; | |
| 996 | return XML_ROLE_CONTENT_ELEMENT_OPT; | |
| 997 | case XML_TOK_NAME_ASTERISK: | |
| 998 | state->handler = element7; | |
| 999 | return XML_ROLE_CONTENT_ELEMENT_REP; | |
| 1000 | case XML_TOK_NAME_PLUS: | |
| 1001 | state->handler = element7; | |
| 1002 | return XML_ROLE_CONTENT_ELEMENT_PLUS; | |
| 1003 | } | |
| 1004 | return syntaxError(state); | |
| 1005 | } | |
| 1006 | ||
| 1007 | static | |
| 1008 | int element7(PROLOG_STATE *state, | |
| 1009 | int tok, | |
| 1010 | const char *ptr, | |
| 1011 | const char *end, | |
| 1012 | const ENCODING *enc) | |
| 1013 | { | |
| 1014 | switch (tok) { | |
| 1015 | case XML_TOK_PROLOG_S: | |
| 1016 | return XML_ROLE_NONE; | |
| 1017 | case XML_TOK_CLOSE_PAREN: | |
| 1018 | state->level -= 1; | |
| 1019 | if (state->level == 0) | |
| 1020 | state->handler = declClose; | |
| 1021 | return XML_ROLE_GROUP_CLOSE; | |
| 1022 | case XML_TOK_CLOSE_PAREN_ASTERISK: | |
| 1023 | state->level -= 1; | |
| 1024 | if (state->level == 0) | |
| 1025 | state->handler = declClose; | |
| 1026 | return XML_ROLE_GROUP_CLOSE_REP; | |
| 1027 | case XML_TOK_CLOSE_PAREN_QUESTION: | |
| 1028 | state->level -= 1; | |
| 1029 | if (state->level == 0) | |
| 1030 | state->handler = declClose; | |
| 1031 | return XML_ROLE_GROUP_CLOSE_OPT; | |
| 1032 | case XML_TOK_CLOSE_PAREN_PLUS: | |
| 1033 | state->level -= 1; | |
| 1034 | if (state->level == 0) | |
| 1035 | state->handler = declClose; | |
| 1036 | return XML_ROLE_GROUP_CLOSE_PLUS; | |
| 1037 | case XML_TOK_COMMA: | |
| 1038 | state->handler = element6; | |
| 1039 | return XML_ROLE_GROUP_SEQUENCE; | |
| 1040 | case XML_TOK_OR: | |
| 1041 | state->handler = element6; | |
| 1042 | return XML_ROLE_GROUP_CHOICE; | |
| 1043 | } | |
| 1044 | return syntaxError(state); | |
| 1045 | } | |
| 1046 | ||
| 1047 | static | |
| 1048 | int declClose(PROLOG_STATE *state, | |
| 1049 | int tok, | |
| 1050 | const char *ptr, | |
| 1051 | const char *end, | |
| 1052 | const ENCODING *enc) | |
| 1053 | { | |
| 1054 | switch (tok) { | |
| 1055 | case XML_TOK_PROLOG_S: | |
| 1056 | return XML_ROLE_NONE; | |
| 1057 | case XML_TOK_DECL_CLOSE: | |
| 1058 | state->handler = internalSubset; | |
| 1059 | return XML_ROLE_NONE; | |
| 1060 | } | |
| 1061 | return syntaxError(state); | |
| 1062 | } | |
| 1063 | ||
| 1064 | #if 0 | |
| 1065 | ||
| 1066 | static | |
| 1067 | int ignore(PROLOG_STATE *state, | |
| 1068 | int tok, | |
| 1069 | const char *ptr, | |
| 1070 | const char *end, | |
| 1071 | const ENCODING *enc) | |
| 1072 | { | |
| 1073 | switch (tok) { | |
| 1074 | case XML_TOK_DECL_CLOSE: | |
| 1075 | state->handler = internalSubset; | |
| 1076 | return 0; | |
| 1077 | default: | |
| 1078 | return XML_ROLE_NONE; | |
| 1079 | } | |
| 1080 | return syntaxError(state); | |
| 1081 | } | |
| 1082 | #endif | |
| 1083 | ||
| 1084 | static | |
| 1085 | int error(PROLOG_STATE *state, | |
| 1086 | int tok, | |
| 1087 | const char *ptr, | |
| 1088 | const char *end, | |
| 1089 | const ENCODING *enc) | |
| 1090 | { | |
| 1091 | return XML_ROLE_NONE; | |
| 1092 | } | |
| 1093 | ||
| 1094 | static | |
| 1095 | int syntaxError(PROLOG_STATE *state) | |
| 1096 | { | |
| 1097 | state->handler = error; | |
| 1098 | return XML_ROLE_ERROR; | |
| 1099 | } | |
| 1100 | ||
| 1101 | void XmlPrologStateInit(PROLOG_STATE *state) | |
| 1102 | { | |
| 1103 | state->handler = prolog0; | |
| 1104 | } |