| 1 | // Copyright (C) 2020 The Qt Company Ltd. |
| 2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only |
| 3 | |
| 4 | |
| 5 | // |
| 6 | // W A R N I N G |
| 7 | // ------------- |
| 8 | // |
| 9 | // This file is not part of the Qt API. It exists purely as an |
| 10 | // implementation detail. This header file may change from version to |
| 11 | // version without notice, or even be removed. |
| 12 | // |
| 13 | // We mean it. |
| 14 | // |
| 15 | |
| 16 | // |
| 17 | // W A R N I N G |
| 18 | // ------------- |
| 19 | // |
| 20 | // This file is automatically generated from qxmlstream.g. |
| 21 | // Changes should be made to that file, not here. Any change to this file will |
| 22 | // be lost! |
| 23 | // |
| 24 | // To regenerate this file, run: |
| 25 | // qlalr --no-debug --no-lines --qt qxmlstream.g |
| 26 | // |
| 27 | |
| 28 | #include <QtCore/private/qglobal_p.h> |
| 29 | #include <qxmlstream.h> |
| 30 | #include "qxmlstream_p.h" |
| 31 | #include "qxmlutils_p.h" |
| 32 | #include <qstringconverter.h> |
| 33 | |
| 34 | #include <memory> |
| 35 | |
| 36 | #ifndef QXMLSTREAMPARSER_P_H |
| 37 | #define QXMLSTREAMPARSER_P_H |
| 38 | |
| 39 | QT_BEGIN_NAMESPACE |
| 40 | |
| 41 | #if QT_CONFIG(xmlstreamreader) |
| 42 | |
| 43 | bool QXmlStreamReaderPrivate::parse() |
| 44 | { |
| 45 | // cleanup currently reported token |
| 46 | |
| 47 | using namespace Qt::StringLiterals; |
| 48 | |
| 49 | switch (type) { |
| 50 | case QXmlStreamReader::StartElement: |
| 51 | name.clear(); |
| 52 | prefix.clear(); |
| 53 | qualifiedName.clear(); |
| 54 | namespaceUri.clear(); |
| 55 | publicNamespaceDeclarations.clear(); |
| 56 | attributes.clear(); |
| 57 | if (isEmptyElement) { |
| 58 | setType(QXmlStreamReader::EndElement); |
| 59 | Tag tag = tagStack_pop(); |
| 60 | namespaceUri = tag.namespaceDeclaration.namespaceUri; |
| 61 | prefix = tag.namespaceDeclaration.prefix; |
| 62 | name = tag.name; |
| 63 | qualifiedName = tag.qualifiedName; |
| 64 | isEmptyElement = false; |
| 65 | return true; |
| 66 | } |
| 67 | clearTextBuffer(); |
| 68 | break; |
| 69 | case QXmlStreamReader::EndElement: |
| 70 | name.clear(); |
| 71 | prefix.clear(); |
| 72 | qualifiedName.clear(); |
| 73 | namespaceUri.clear(); |
| 74 | clearTextBuffer(); |
| 75 | break; |
| 76 | case QXmlStreamReader::DTD: |
| 77 | publicNotationDeclarations.clear(); |
| 78 | publicEntityDeclarations.clear(); |
| 79 | dtdName.clear(); |
| 80 | dtdPublicId.clear(); |
| 81 | dtdSystemId.clear(); |
| 82 | Q_FALLTHROUGH(); |
| 83 | case QXmlStreamReader::Comment: |
| 84 | case QXmlStreamReader::Characters: |
| 85 | isCDATA = false; |
| 86 | isWhitespace = true; |
| 87 | text.clear(); |
| 88 | clearTextBuffer(); |
| 89 | break; |
| 90 | case QXmlStreamReader::EntityReference: |
| 91 | text.clear(); |
| 92 | name.clear(); |
| 93 | clearTextBuffer(); |
| 94 | break; |
| 95 | case QXmlStreamReader::ProcessingInstruction: |
| 96 | processingInstructionTarget.clear(); |
| 97 | processingInstructionData.clear(); |
| 98 | clearTextBuffer(); |
| 99 | break; |
| 100 | case QXmlStreamReader::NoToken: |
| 101 | case QXmlStreamReader::Invalid: |
| 102 | break; |
| 103 | case QXmlStreamReader::StartDocument: |
| 104 | lockEncoding = true; |
| 105 | documentVersion.clear(); |
| 106 | documentEncoding.clear(); |
| 107 | if (decoder.isValid() && decoder.hasError()) { |
| 108 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Encountered incorrectly encoded content." )); |
| 109 | readBuffer.clear(); |
| 110 | return false; |
| 111 | } |
| 112 | Q_FALLTHROUGH(); |
| 113 | default: |
| 114 | clearTextBuffer(); |
| 115 | ; |
| 116 | } |
| 117 | |
| 118 | setType(QXmlStreamReader::NoToken); |
| 119 | |
| 120 | |
| 121 | // the main parse loop |
| 122 | int act, r; |
| 123 | |
| 124 | if (resumeReduction) { |
| 125 | act = state_stack[tos-1]; |
| 126 | r = resumeReduction; |
| 127 | resumeReduction = 0; |
| 128 | goto ResumeReduction; |
| 129 | } |
| 130 | |
| 131 | act = state_stack[tos]; |
| 132 | |
| 133 | while (true) { |
| 134 | if (token == -1 && - TERMINAL_COUNT != action_index[act]) { |
| 135 | uint cu = getChar(); |
| 136 | token = NOTOKEN; |
| 137 | token_char = cu == ~0U ? cu : ushort(cu); |
| 138 | if ((cu != ~0U) && (cu & 0xff0000)) { |
| 139 | token = cu >> 16; |
| 140 | } else switch (token_char) { |
| 141 | case 0xfffe: |
| 142 | case 0xffff: |
| 143 | token = XML_ERROR; |
| 144 | break; |
| 145 | case '\r': |
| 146 | token = SPACE; |
| 147 | if (cu == '\r') { |
| 148 | if ((token_char = filterCarriageReturn())) { |
| 149 | ++lineNumber; |
| 150 | lastLineStart = characterOffset + readBufferPos; |
| 151 | break; |
| 152 | } |
| 153 | } else { |
| 154 | break; |
| 155 | } |
| 156 | Q_FALLTHROUGH(); |
| 157 | case ~0U: { |
| 158 | token = EOF_SYMBOL; |
| 159 | if (!tagsDone && !inParseEntity) { |
| 160 | int a = t_action(state: act, token); |
| 161 | if (a < 0) { |
| 162 | raiseError(error: QXmlStreamReader::PrematureEndOfDocumentError); |
| 163 | return false; |
| 164 | } |
| 165 | } |
| 166 | |
| 167 | } break; |
| 168 | case '\n': |
| 169 | ++lineNumber; |
| 170 | lastLineStart = characterOffset + readBufferPos; |
| 171 | Q_FALLTHROUGH(); |
| 172 | case ' ': |
| 173 | case '\t': |
| 174 | token = SPACE; |
| 175 | break; |
| 176 | case '&': |
| 177 | token = AMPERSAND; |
| 178 | break; |
| 179 | case '#': |
| 180 | token = HASH; |
| 181 | break; |
| 182 | case '\'': |
| 183 | token = QUOTE; |
| 184 | break; |
| 185 | case '\"': |
| 186 | token = DBLQUOTE; |
| 187 | break; |
| 188 | case '<': |
| 189 | token = LANGLE; |
| 190 | break; |
| 191 | case '>': |
| 192 | token = RANGLE; |
| 193 | break; |
| 194 | case '[': |
| 195 | token = LBRACK; |
| 196 | break; |
| 197 | case ']': |
| 198 | token = RBRACK; |
| 199 | break; |
| 200 | case '(': |
| 201 | token = LPAREN; |
| 202 | break; |
| 203 | case ')': |
| 204 | token = RPAREN; |
| 205 | break; |
| 206 | case '|': |
| 207 | token = PIPE; |
| 208 | break; |
| 209 | case '=': |
| 210 | token = EQ; |
| 211 | break; |
| 212 | case '%': |
| 213 | token = PERCENT; |
| 214 | break; |
| 215 | case '/': |
| 216 | token = SLASH; |
| 217 | break; |
| 218 | case ':': |
| 219 | token = COLON; |
| 220 | break; |
| 221 | case ';': |
| 222 | token = SEMICOLON; |
| 223 | break; |
| 224 | case ',': |
| 225 | token = COMMA; |
| 226 | break; |
| 227 | case '-': |
| 228 | token = DASH; |
| 229 | break; |
| 230 | case '+': |
| 231 | token = PLUS; |
| 232 | break; |
| 233 | case '*': |
| 234 | token = STAR; |
| 235 | break; |
| 236 | case '.': |
| 237 | token = DOT; |
| 238 | break; |
| 239 | case '?': |
| 240 | token = QUESTIONMARK; |
| 241 | break; |
| 242 | case '!': |
| 243 | token = BANG; |
| 244 | break; |
| 245 | case '0': |
| 246 | case '1': |
| 247 | case '2': |
| 248 | case '3': |
| 249 | case '4': |
| 250 | case '5': |
| 251 | case '6': |
| 252 | case '7': |
| 253 | case '8': |
| 254 | case '9': |
| 255 | token = DIGIT; |
| 256 | break; |
| 257 | default: |
| 258 | if (cu < 0x20) |
| 259 | token = NOTOKEN; |
| 260 | else |
| 261 | token = LETTER; |
| 262 | break; |
| 263 | } |
| 264 | } |
| 265 | |
| 266 | act = t_action (state: act, token); |
| 267 | if (act == ACCEPT_STATE) { |
| 268 | // reset the parser in case someone resumes (process instructions can follow a valid document) |
| 269 | tos = 0; |
| 270 | state_stack[tos++] = 0; |
| 271 | state_stack[tos] = 0; |
| 272 | return true; |
| 273 | } else if (act > 0) { |
| 274 | if (++tos >= stack_size-1) |
| 275 | reallocateStack(); |
| 276 | |
| 277 | Value &val = sym_stack[tos]; |
| 278 | val.c = token_char; |
| 279 | val.pos = textBuffer.size(); |
| 280 | val.prefix = 0; |
| 281 | val.len = 1; |
| 282 | if (token_char) |
| 283 | textBuffer += QChar(token_char); |
| 284 | |
| 285 | state_stack[tos] = act; |
| 286 | token = -1; |
| 287 | |
| 288 | |
| 289 | } else if (act < 0) { |
| 290 | r = - act - 1; |
| 291 | |
| 292 | #if defined (QLALR_DEBUG) |
| 293 | int ridx = rule_index[r]; |
| 294 | printf ("%3d) %s ::=" , r + 1, spell[rule_info[ridx]]); |
| 295 | ++ridx; |
| 296 | for (int i = ridx; i < ridx + rhs[r]; ++i) { |
| 297 | int symbol = rule_info[i]; |
| 298 | if (const char *name = spell[symbol]) |
| 299 | printf (" %s" , name); |
| 300 | else |
| 301 | printf (" #%d" , symbol); |
| 302 | } |
| 303 | printf ("\n" ); |
| 304 | #endif |
| 305 | |
| 306 | tos -= rhs[r]; |
| 307 | act = state_stack[tos++]; |
| 308 | ResumeReduction: |
| 309 | switch (r) { |
| 310 | |
| 311 | case 0: |
| 312 | setType(QXmlStreamReader::EndDocument); |
| 313 | break; |
| 314 | |
| 315 | case 1: |
| 316 | if (type != QXmlStreamReader::Invalid) { |
| 317 | if (hasSeenTag || inParseEntity) { |
| 318 | setType(QXmlStreamReader::EndDocument); |
| 319 | } else { |
| 320 | raiseError(error: QXmlStreamReader::NotWellFormedError, message: QXmlStream::tr(sourceText: "Start tag expected." )); |
| 321 | // reset the parser |
| 322 | tos = 0; |
| 323 | state_stack[tos++] = 0; |
| 324 | state_stack[tos] = 0; |
| 325 | return false; |
| 326 | } |
| 327 | } |
| 328 | break; |
| 329 | |
| 330 | case 10: { |
| 331 | auto reference = entityReferenceStack.pop(); |
| 332 | auto it = reference.hash->find(key: reference.name); |
| 333 | Q_ASSERT(it != reference.hash->end()); |
| 334 | it->isCurrentlyReferenced = false; |
| 335 | if (entityReferenceStack.isEmpty()) |
| 336 | entityLength = 0; |
| 337 | clearSym(); |
| 338 | } break; |
| 339 | |
| 340 | case 11: |
| 341 | if (!scanString(str: spell[VERSION], tokenToInject: VERSION, requireSpace: false) && atEnd) { |
| 342 | resume(rule: 11); |
| 343 | return false; |
| 344 | } |
| 345 | break; |
| 346 | |
| 347 | case 12: |
| 348 | setType(QXmlStreamReader::StartDocument); |
| 349 | documentVersion = symString(index: 6); |
| 350 | startDocument(); |
| 351 | break; |
| 352 | |
| 353 | case 13: |
| 354 | hasExternalDtdSubset = true; |
| 355 | dtdSystemId = symString(index: 2); |
| 356 | break; |
| 357 | |
| 358 | case 14: |
| 359 | checkPublicLiteral(publicId: symString(index: 2)); |
| 360 | dtdPublicId = symString(index: 2); |
| 361 | dtdSystemId = symString(index: 4); |
| 362 | hasExternalDtdSubset = true; |
| 363 | break; |
| 364 | |
| 365 | case 16: |
| 366 | if (!scanPublicOrSystem() && atEnd) { |
| 367 | resume(rule: 16); |
| 368 | return false; |
| 369 | } |
| 370 | dtdName = symString(index: 3); |
| 371 | break; |
| 372 | |
| 373 | case 17: |
| 374 | case 18: |
| 375 | dtdName = symString(index: 3); |
| 376 | Q_FALLTHROUGH(); |
| 377 | |
| 378 | case 19: |
| 379 | case 20: |
| 380 | setType(QXmlStreamReader::DTD); |
| 381 | text = &textBuffer; |
| 382 | break; |
| 383 | |
| 384 | case 21: |
| 385 | scanDtd = true; |
| 386 | break; |
| 387 | |
| 388 | case 22: |
| 389 | scanDtd = false; |
| 390 | break; |
| 391 | |
| 392 | case 37: |
| 393 | if (!scanString(str: spell[EMPTY], tokenToInject: EMPTY, requireSpace: false) |
| 394 | && !scanString(str: spell[ANY], tokenToInject: ANY, requireSpace: false) |
| 395 | && atEnd) { |
| 396 | resume(rule: 37); |
| 397 | return false; |
| 398 | } |
| 399 | break; |
| 400 | |
| 401 | case 43: |
| 402 | if (!scanString(str: spell[PCDATA], tokenToInject: PCDATA, requireSpace: false) && atEnd) { |
| 403 | resume(rule: 43); |
| 404 | return false; |
| 405 | } |
| 406 | break; |
| 407 | |
| 408 | case 68: { |
| 409 | lastAttributeIsCData = true; |
| 410 | } break; |
| 411 | |
| 412 | case 78: |
| 413 | if (!scanAfterDefaultDecl() && atEnd) { |
| 414 | resume(rule: 78); |
| 415 | return false; |
| 416 | } |
| 417 | break; |
| 418 | |
| 419 | case 83: |
| 420 | sym(index: 1) = sym(index: 2); |
| 421 | lastAttributeValue.clear(); |
| 422 | lastAttributeIsCData = false; |
| 423 | if (!scanAttType() && atEnd) { |
| 424 | resume(rule: 83); |
| 425 | return false; |
| 426 | } |
| 427 | break; |
| 428 | |
| 429 | case 84: { |
| 430 | DtdAttribute &dtdAttribute = dtdAttributes.push(); |
| 431 | dtdAttribute.tagName.clear(); |
| 432 | dtdAttribute.isCDATA = lastAttributeIsCData; |
| 433 | dtdAttribute.attributePrefix = addToStringStorage(s: symPrefix(index: 1)); |
| 434 | dtdAttribute.attributeName = addToStringStorage(s: symString(index: 1)); |
| 435 | dtdAttribute.attributeQualifiedName = addToStringStorage(s: symName(index: 1)); |
| 436 | dtdAttribute.isNamespaceAttribute = (dtdAttribute.attributePrefix == "xmlns"_L1 |
| 437 | || (dtdAttribute.attributePrefix.isEmpty() |
| 438 | && dtdAttribute.attributeName == "xmlns"_L1 )); |
| 439 | if (lastAttributeValue.isNull()) { |
| 440 | dtdAttribute.defaultValue.clear(); |
| 441 | } else { |
| 442 | if (dtdAttribute.isCDATA) |
| 443 | dtdAttribute.defaultValue = addToStringStorage(s: lastAttributeValue); |
| 444 | else |
| 445 | dtdAttribute.defaultValue = addToStringStorage(s: lastAttributeValue.toString().simplified()); |
| 446 | |
| 447 | } |
| 448 | } break; |
| 449 | |
| 450 | case 88: { |
| 451 | if (referenceToUnparsedEntityDetected && !standalone) |
| 452 | break; |
| 453 | qsizetype n = dtdAttributes.size(); |
| 454 | XmlStringRef tagName = addToStringStorage(s: symName(index: 3)); |
| 455 | while (n--) { |
| 456 | DtdAttribute &dtdAttribute = dtdAttributes[n]; |
| 457 | if (!dtdAttribute.tagName.isNull()) |
| 458 | break; |
| 459 | dtdAttribute.tagName = tagName; |
| 460 | for (qsizetype i = 0; i < n; ++i) { |
| 461 | if ((dtdAttributes[i].tagName.isNull() || dtdAttributes[i].tagName == tagName) |
| 462 | && dtdAttributes[i].attributeQualifiedName == dtdAttribute.attributeQualifiedName) { |
| 463 | dtdAttribute.attributeQualifiedName.clear(); // redefined, delete it |
| 464 | break; |
| 465 | } |
| 466 | } |
| 467 | } |
| 468 | } break; |
| 469 | |
| 470 | case 89: { |
| 471 | if (!scanPublicOrSystem() && atEnd) { |
| 472 | resume(rule: 89); |
| 473 | return false; |
| 474 | } |
| 475 | EntityDeclaration &entityDeclaration = entityDeclarations.push(); |
| 476 | entityDeclaration.clear(); |
| 477 | entityDeclaration.name = symString(index: 3); |
| 478 | } break; |
| 479 | |
| 480 | case 90: { |
| 481 | if (!scanPublicOrSystem() && atEnd) { |
| 482 | resume(rule: 90); |
| 483 | return false; |
| 484 | } |
| 485 | EntityDeclaration &entityDeclaration = entityDeclarations.push(); |
| 486 | entityDeclaration.clear(); |
| 487 | entityDeclaration.name = symString(index: 5); |
| 488 | entityDeclaration.parameter = true; |
| 489 | } break; |
| 490 | |
| 491 | case 91: { |
| 492 | if (!scanNData() && atEnd) { |
| 493 | resume(rule: 91); |
| 494 | return false; |
| 495 | } |
| 496 | EntityDeclaration &entityDeclaration = entityDeclarations.top(); |
| 497 | entityDeclaration.systemId = symString(index: 3); |
| 498 | entityDeclaration.external = true; |
| 499 | } break; |
| 500 | |
| 501 | case 92: { |
| 502 | if (!scanNData() && atEnd) { |
| 503 | resume(rule: 92); |
| 504 | return false; |
| 505 | } |
| 506 | EntityDeclaration &entityDeclaration = entityDeclarations.top(); |
| 507 | checkPublicLiteral(publicId: (entityDeclaration.publicId = symString(index: 3))); |
| 508 | entityDeclaration.systemId = symString(index: 5); |
| 509 | entityDeclaration.external = true; |
| 510 | } break; |
| 511 | |
| 512 | case 93: { |
| 513 | EntityDeclaration &entityDeclaration = entityDeclarations.top(); |
| 514 | entityDeclaration.notationName = symString(index: 3); |
| 515 | if (entityDeclaration.parameter) |
| 516 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "NDATA in parameter entity declaration." )); |
| 517 | } |
| 518 | Q_FALLTHROUGH(); |
| 519 | |
| 520 | case 94: |
| 521 | case 95: { |
| 522 | if (referenceToUnparsedEntityDetected && !standalone) { |
| 523 | entityDeclarations.pop(); |
| 524 | break; |
| 525 | } |
| 526 | EntityDeclaration &entityDeclaration = entityDeclarations.top(); |
| 527 | if (!entityDeclaration.external) |
| 528 | entityDeclaration.value = symString(index: 2); |
| 529 | auto &hash = entityDeclaration.parameter ? parameterEntityHash : entityHash; |
| 530 | if (!hash.contains(key: entityDeclaration.name)) { |
| 531 | Entity entity(entityDeclaration.name.toString(), |
| 532 | entityDeclaration.value.toString()); |
| 533 | entity.unparsed = (!entityDeclaration.notationName.isNull()); |
| 534 | entity.external = entityDeclaration.external; |
| 535 | hash.insert(key: qToStringViewIgnoringNull(s: entity.name), value: entity); |
| 536 | } |
| 537 | } break; |
| 538 | |
| 539 | case 96: { |
| 540 | setType(QXmlStreamReader::ProcessingInstruction); |
| 541 | const qsizetype pos = sym(index: 4).pos + sym(index: 4).len; |
| 542 | processingInstructionTarget = symString(index: 3); |
| 543 | if (scanUntil(str: "?>" )) { |
| 544 | processingInstructionData = XmlStringRef(&textBuffer, pos, textBuffer.size() - pos - 2); |
| 545 | if (!processingInstructionTarget.view().compare(s: "xml"_L1 , cs: Qt::CaseInsensitive)) { |
| 546 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "XML declaration not at start of document." )); |
| 547 | } |
| 548 | else if (!QXmlUtils::isNCName(ncName: processingInstructionTarget)) |
| 549 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "%1 is an invalid processing instruction name." ) |
| 550 | .arg(a: processingInstructionTarget)); |
| 551 | } else if (type != QXmlStreamReader::Invalid){ |
| 552 | resume(rule: 96); |
| 553 | return false; |
| 554 | } |
| 555 | } break; |
| 556 | |
| 557 | case 97: |
| 558 | setType(QXmlStreamReader::ProcessingInstruction); |
| 559 | processingInstructionTarget = symString(index: 3); |
| 560 | if (!processingInstructionTarget.view().compare(s: "xml"_L1 , cs: Qt::CaseInsensitive)) |
| 561 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Invalid processing instruction name." )); |
| 562 | break; |
| 563 | |
| 564 | case 98: |
| 565 | if (!scanAfterLangleBang() && atEnd) { |
| 566 | resume(rule: 98); |
| 567 | return false; |
| 568 | } |
| 569 | break; |
| 570 | |
| 571 | case 99: |
| 572 | if (!scanUntil(str: "--" )) { |
| 573 | resume(rule: 99); |
| 574 | return false; |
| 575 | } |
| 576 | break; |
| 577 | |
| 578 | case 100: { |
| 579 | setType(QXmlStreamReader::Comment); |
| 580 | const qsizetype pos = sym(index: 1).pos + 4; |
| 581 | text = XmlStringRef(&textBuffer, pos, textBuffer.size() - pos - 3); |
| 582 | } break; |
| 583 | |
| 584 | case 101: { |
| 585 | setType(QXmlStreamReader::Characters); |
| 586 | isCDATA = true; |
| 587 | isWhitespace = false; |
| 588 | const qsizetype pos = sym(index: 2).pos; |
| 589 | if (scanUntil(str: "]]>" , tokenToInject: -1)) { |
| 590 | text = XmlStringRef(&textBuffer, pos, textBuffer.size() - pos - 3); |
| 591 | } else { |
| 592 | resume(rule: 101); |
| 593 | return false; |
| 594 | } |
| 595 | } break; |
| 596 | |
| 597 | case 102: { |
| 598 | if (!scanPublicOrSystem() && atEnd) { |
| 599 | resume(rule: 102); |
| 600 | return false; |
| 601 | } |
| 602 | NotationDeclaration ¬ationDeclaration = notationDeclarations.push(); |
| 603 | notationDeclaration.name = symString(index: 3); |
| 604 | } break; |
| 605 | |
| 606 | case 103: { |
| 607 | NotationDeclaration ¬ationDeclaration = notationDeclarations.top(); |
| 608 | notationDeclaration.systemId = symString(index: 3); |
| 609 | notationDeclaration.publicId.clear(); |
| 610 | } break; |
| 611 | |
| 612 | case 104: { |
| 613 | NotationDeclaration ¬ationDeclaration = notationDeclarations.top(); |
| 614 | notationDeclaration.systemId.clear(); |
| 615 | checkPublicLiteral(publicId: (notationDeclaration.publicId = symString(index: 3))); |
| 616 | } break; |
| 617 | |
| 618 | case 105: { |
| 619 | NotationDeclaration ¬ationDeclaration = notationDeclarations.top(); |
| 620 | checkPublicLiteral(publicId: (notationDeclaration.publicId = symString(index: 3))); |
| 621 | notationDeclaration.systemId = symString(index: 5); |
| 622 | } break; |
| 623 | |
| 624 | case 129: |
| 625 | isWhitespace = false; |
| 626 | Q_FALLTHROUGH(); |
| 627 | |
| 628 | case 130: |
| 629 | sym(index: 1).len += fastScanContentCharList(); |
| 630 | if (atEnd && !inParseEntity) { |
| 631 | resume(rule: 130); |
| 632 | return false; |
| 633 | } |
| 634 | break; |
| 635 | |
| 636 | case 139: |
| 637 | if (!textBuffer.isEmpty()) { |
| 638 | setType(QXmlStreamReader::Characters); |
| 639 | text = &textBuffer; |
| 640 | } |
| 641 | break; |
| 642 | |
| 643 | case 140: |
| 644 | case 141: |
| 645 | clearSym(); |
| 646 | break; |
| 647 | |
| 648 | case 142: |
| 649 | case 143: |
| 650 | sym(index: 1) = sym(index: 2); |
| 651 | break; |
| 652 | |
| 653 | case 144: |
| 654 | case 145: |
| 655 | case 146: |
| 656 | case 147: |
| 657 | sym(index: 1).len += sym(index: 2).len; |
| 658 | break; |
| 659 | |
| 660 | case 173: |
| 661 | if (normalizeLiterals) |
| 662 | textBuffer.data()[textBuffer.size()-1] = u' '; |
| 663 | break; |
| 664 | |
| 665 | case 174: |
| 666 | sym(index: 1).len += fastScanLiteralContent(); |
| 667 | if (atEnd) { |
| 668 | resume(rule: 174); |
| 669 | return false; |
| 670 | } |
| 671 | break; |
| 672 | |
| 673 | case 175: { |
| 674 | if (!QXmlUtils::isPublicID(candidate: symString(index: 1))) { |
| 675 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "%1 is an invalid PUBLIC identifier." ).arg(a: symString(index: 1))); |
| 676 | resume(rule: 175); |
| 677 | return false; |
| 678 | } |
| 679 | } break; |
| 680 | |
| 681 | case 176: |
| 682 | case 177: |
| 683 | clearSym(); |
| 684 | break; |
| 685 | |
| 686 | case 178: |
| 687 | case 179: |
| 688 | sym(index: 1) = sym(index: 2); |
| 689 | break; |
| 690 | |
| 691 | case 180: |
| 692 | case 181: |
| 693 | case 182: |
| 694 | case 183: |
| 695 | sym(index: 1).len += sym(index: 2).len; |
| 696 | break; |
| 697 | |
| 698 | case 213: |
| 699 | case 214: |
| 700 | clearSym(); |
| 701 | break; |
| 702 | |
| 703 | case 215: |
| 704 | case 216: |
| 705 | sym(index: 1) = sym(index: 2); |
| 706 | lastAttributeValue = symString(index: 1); |
| 707 | break; |
| 708 | |
| 709 | case 217: |
| 710 | case 218: |
| 711 | case 219: |
| 712 | case 220: |
| 713 | sym(index: 1).len += sym(index: 2).len; |
| 714 | break; |
| 715 | |
| 716 | case 229: { |
| 717 | const XmlStringRef prfx = symPrefix(index: 1); |
| 718 | if (prfx.isEmpty() && symString(index: 1) == "xmlns"_L1 && namespaceProcessing) { |
| 719 | NamespaceDeclaration &namespaceDeclaration = namespaceDeclarations.push(); |
| 720 | namespaceDeclaration.prefix.clear(); |
| 721 | |
| 722 | const XmlStringRef ns(symString(index: 5)); |
| 723 | if (ns.view() == "http://www.w3.org/2000/xmlns/"_L1 || |
| 724 | ns.view() == "http://www.w3.org/XML/1998/namespace"_L1 ) |
| 725 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Illegal namespace declaration." )); |
| 726 | else |
| 727 | namespaceDeclaration.namespaceUri = addToStringStorage(s: ns); |
| 728 | } else { |
| 729 | Attribute &attribute = attributeStack.push(); |
| 730 | attribute.key = sym(index: 1); |
| 731 | attribute.value = sym(index: 5); |
| 732 | |
| 733 | XmlStringRef attributeQualifiedName = symName(index: 1); |
| 734 | bool normalize = false; |
| 735 | for (const DtdAttribute &dtdAttribute : std::as_const(t&: dtdAttributes)) { |
| 736 | if (!dtdAttribute.isCDATA |
| 737 | && dtdAttribute.tagName == qualifiedName |
| 738 | && dtdAttribute.attributeQualifiedName == attributeQualifiedName |
| 739 | ) { |
| 740 | normalize = true; |
| 741 | break; |
| 742 | } |
| 743 | } |
| 744 | if (normalize) { |
| 745 | // normalize attribute value (simplify and trim) |
| 746 | const qsizetype pos = textBuffer.size(); |
| 747 | qsizetype n = 0; |
| 748 | bool wasSpace = true; |
| 749 | for (qsizetype i = 0; i < attribute.value.len; ++i) { |
| 750 | QChar c = textBuffer.at(i: attribute.value.pos + i); |
| 751 | if (c.unicode() == ' ') { |
| 752 | if (wasSpace) |
| 753 | continue; |
| 754 | wasSpace = true; |
| 755 | } else { |
| 756 | wasSpace = false; |
| 757 | } |
| 758 | textBuffer += textBuffer.at(i: attribute.value.pos + i); |
| 759 | ++n; |
| 760 | } |
| 761 | if (wasSpace) |
| 762 | while (n && textBuffer.at(i: pos + n - 1).unicode() == ' ') |
| 763 | --n; |
| 764 | attribute.value.pos = pos; |
| 765 | attribute.value.len = n; |
| 766 | } |
| 767 | if (prfx == "xmlns"_L1 && namespaceProcessing) { |
| 768 | NamespaceDeclaration &namespaceDeclaration = namespaceDeclarations.push(); |
| 769 | XmlStringRef namespacePrefix = symString(symbol: attribute.key); |
| 770 | XmlStringRef namespaceUri = symString(symbol: attribute.value); |
| 771 | attributeStack.pop(); |
| 772 | if (((namespacePrefix == "xml"_L1 ) |
| 773 | ^ (namespaceUri == "http://www.w3.org/XML/1998/namespace"_L1 )) |
| 774 | || namespaceUri == "http://www.w3.org/2000/xmlns/"_L1 |
| 775 | || namespaceUri.isEmpty() |
| 776 | || namespacePrefix == "xmlns"_L1 ) |
| 777 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Illegal namespace declaration." )); |
| 778 | |
| 779 | namespaceDeclaration.prefix = addToStringStorage(s: namespacePrefix); |
| 780 | namespaceDeclaration.namespaceUri = addToStringStorage(s: namespaceUri); |
| 781 | } |
| 782 | } |
| 783 | } break; |
| 784 | |
| 785 | case 235: { |
| 786 | normalizeLiterals = true; |
| 787 | Tag &tag = tagStack_push(); |
| 788 | prefix = tag.namespaceDeclaration.prefix = addToStringStorage(s: symPrefix(index: 2)); |
| 789 | name = tag.name = addToStringStorage(s: symString(index: 2)); |
| 790 | qualifiedName = tag.qualifiedName = addToStringStorage(s: symName(index: 2)); |
| 791 | if ((!prefix.isEmpty() && !QXmlUtils::isNCName(ncName: prefix)) || !QXmlUtils::isNCName(ncName: name)) |
| 792 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Invalid XML name." )); |
| 793 | } break; |
| 794 | |
| 795 | case 236: |
| 796 | isEmptyElement = true; |
| 797 | Q_FALLTHROUGH(); |
| 798 | |
| 799 | case 237: |
| 800 | setType(QXmlStreamReader::StartElement); |
| 801 | resolveTag(); |
| 802 | if (tagStack.size() == 1 && hasSeenTag && !inParseEntity) |
| 803 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Extra content at end of document." )); |
| 804 | hasSeenTag = true; |
| 805 | break; |
| 806 | |
| 807 | case 238: { |
| 808 | setType(QXmlStreamReader::EndElement); |
| 809 | Tag tag = tagStack_pop(); |
| 810 | |
| 811 | namespaceUri = tag.namespaceDeclaration.namespaceUri; |
| 812 | prefix = tag.namespaceDeclaration.prefix; |
| 813 | name = tag.name; |
| 814 | qualifiedName = tag.qualifiedName; |
| 815 | if (qualifiedName != symName(index: 3)) |
| 816 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Opening and ending tag mismatch." )); |
| 817 | } break; |
| 818 | |
| 819 | case 239: |
| 820 | if (entitiesMustBeDeclared()) { |
| 821 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Entity '%1' not declared." ).arg(a: unresolvedEntity)); |
| 822 | break; |
| 823 | } |
| 824 | setType(QXmlStreamReader::EntityReference); |
| 825 | name = &unresolvedEntity; |
| 826 | break; |
| 827 | |
| 828 | case 240: { |
| 829 | sym(index: 1).len += sym(index: 2).len + 1; |
| 830 | QStringView reference = symView(index: 2); |
| 831 | if (const auto it = entityHash.find(key: reference); it != entityHash.end()) { |
| 832 | Entity &entity = *it; |
| 833 | if (entity.unparsed) { |
| 834 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Reference to unparsed entity '%1'." ).arg(a: reference)); |
| 835 | } else { |
| 836 | if (!entity.hasBeenParsed) { |
| 837 | parseEntity(value: entity.value); |
| 838 | entity.hasBeenParsed = true; |
| 839 | } |
| 840 | if (entity.literal) |
| 841 | putStringLiteral(s: entity.value); |
| 842 | else if (referenceEntity(hash: &entityHash, entity)) |
| 843 | putReplacement(s: entity.value); |
| 844 | textBuffer.chop(n: 2 + sym(index: 2).len); |
| 845 | clearSym(); |
| 846 | } |
| 847 | break; |
| 848 | } |
| 849 | |
| 850 | if (entityResolver) { |
| 851 | QString replacementText = resolveUndeclaredEntity(name: reference.toString()); |
| 852 | if (!replacementText.isNull()) { |
| 853 | putReplacement(s: replacementText); |
| 854 | textBuffer.chop(n: 2 + sym(index: 2).len); |
| 855 | clearSym(); |
| 856 | break; |
| 857 | } |
| 858 | } |
| 859 | |
| 860 | injectToken(tokenToInject: UNRESOLVED_ENTITY); |
| 861 | unresolvedEntity = symString(index: 2).toString(); |
| 862 | textBuffer.chop(n: 2 + sym(index: 2).len); |
| 863 | clearSym(); |
| 864 | |
| 865 | } break; |
| 866 | |
| 867 | case 241: { |
| 868 | sym(index: 1).len += sym(index: 2).len + 1; |
| 869 | QStringView reference = symView(index: 2); |
| 870 | if (const auto it = parameterEntityHash.find(key: reference); it != parameterEntityHash.end()) { |
| 871 | referenceToParameterEntityDetected = true; |
| 872 | Entity &entity = *it; |
| 873 | if (entity.unparsed || entity.external) { |
| 874 | referenceToUnparsedEntityDetected = true; |
| 875 | } else { |
| 876 | if (referenceEntity(hash: ¶meterEntityHash, entity)) |
| 877 | putString(s: entity.value); |
| 878 | textBuffer.chop(n: 2 + sym(index: 2).len); |
| 879 | clearSym(); |
| 880 | } |
| 881 | } else if (entitiesMustBeDeclared()) { |
| 882 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Entity '%1' not declared." ).arg(a: symString(index: 2))); |
| 883 | } |
| 884 | } break; |
| 885 | |
| 886 | case 242: |
| 887 | sym(index: 1).len += sym(index: 2).len + 1; |
| 888 | break; |
| 889 | |
| 890 | case 243: { |
| 891 | sym(index: 1).len += sym(index: 2).len + 1; |
| 892 | QStringView reference = symView(index: 2); |
| 893 | if (const auto it = entityHash.find(key: reference); it != entityHash.end()) { |
| 894 | Entity &entity = *it; |
| 895 | if (entity.unparsed || entity.value.isNull()) { |
| 896 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Reference to external entity '%1' in attribute value." ).arg(a: reference)); |
| 897 | break; |
| 898 | } |
| 899 | if (!entity.hasBeenParsed) { |
| 900 | parseEntity(value: entity.value); |
| 901 | entity.hasBeenParsed = true; |
| 902 | } |
| 903 | if (entity.literal) |
| 904 | putStringLiteral(s: entity.value); |
| 905 | else if (referenceEntity(hash: &entityHash, entity)) |
| 906 | putReplacementInAttributeValue(s: entity.value); |
| 907 | textBuffer.chop(n: 2 + sym(index: 2).len); |
| 908 | clearSym(); |
| 909 | break; |
| 910 | } |
| 911 | |
| 912 | if (entityResolver) { |
| 913 | QString replacementText = resolveUndeclaredEntity(name: reference.toString()); |
| 914 | if (!replacementText.isNull()) { |
| 915 | putReplacement(s: replacementText); |
| 916 | textBuffer.chop(n: 2 + sym(index: 2).len); |
| 917 | clearSym(); |
| 918 | break; |
| 919 | } |
| 920 | } |
| 921 | if (entitiesMustBeDeclared()) { |
| 922 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Entity '%1' not declared." ).arg(a: reference)); |
| 923 | } |
| 924 | } break; |
| 925 | |
| 926 | case 244: { |
| 927 | if (char32_t s = resolveCharRef(symbolIndex: 3)) { |
| 928 | putStringLiteral(s: QChar::fromUcs4(c: s)); |
| 929 | textBuffer.chop(n: 3 + sym(index: 3).len); |
| 930 | clearSym(); |
| 931 | } else { |
| 932 | raiseWellFormedError(message: QXmlStream::tr(sourceText: "Invalid character reference." )); |
| 933 | } |
| 934 | } break; |
| 935 | |
| 936 | case 247: |
| 937 | case 248: |
| 938 | sym(index: 1).len += sym(index: 2).len; |
| 939 | break; |
| 940 | |
| 941 | case 259: |
| 942 | sym(index: 1).len += fastScanSpace(); |
| 943 | if (atEnd) { |
| 944 | resume(rule: 259); |
| 945 | return false; |
| 946 | } |
| 947 | break; |
| 948 | |
| 949 | case 262: { |
| 950 | Value &val = sym(index: 1); |
| 951 | if (auto res = fastScanName(val: &val)) |
| 952 | val.len += *res; |
| 953 | else |
| 954 | return false; |
| 955 | |
| 956 | if (atEnd) { |
| 957 | resume(rule: 262); |
| 958 | return false; |
| 959 | } |
| 960 | } break; |
| 961 | |
| 962 | case 263: |
| 963 | if (auto res = fastScanName()) |
| 964 | sym(index: 1).len += *res; |
| 965 | else |
| 966 | return false; |
| 967 | |
| 968 | if (atEnd) { |
| 969 | resume(rule: 263); |
| 970 | return false; |
| 971 | } |
| 972 | break; |
| 973 | |
| 974 | case 264: |
| 975 | case 265: |
| 976 | case 266: |
| 977 | case 267: |
| 978 | case 268: |
| 979 | sym(index: 1).len += fastScanNMTOKEN(); |
| 980 | if (atEnd) { |
| 981 | resume(rule: 268); |
| 982 | return false; |
| 983 | } |
| 984 | |
| 985 | break; |
| 986 | |
| 987 | default: |
| 988 | ; |
| 989 | } // switch |
| 990 | act = state_stack[tos] = nt_action (state: act, nt: lhs[r] - TERMINAL_COUNT); |
| 991 | if (type != QXmlStreamReader::NoToken) |
| 992 | return true; |
| 993 | } else { |
| 994 | parseError(); |
| 995 | break; |
| 996 | } |
| 997 | } |
| 998 | return false; |
| 999 | } |
| 1000 | |
| 1001 | #endif // feature xmlstreamreader |
| 1002 | |
| 1003 | QT_END_NAMESPACE |
| 1004 | |
| 1005 | #endif |
| 1006 | |
| 1007 | |