| 1 | // Copyright (C) 2016 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 | // Qt-Security score:critical reason:data-parser |
| 4 | |
| 5 | #include "qglobal.h" |
| 6 | |
| 7 | // Disable warning about use of deprecated QXmlStreamLocator in QScopedPointer<> |
| 8 | QT_WARNING_DISABLE_MSVC(4996) |
| 9 | |
| 10 | #include "qxml.h" |
| 11 | #include "qxml_p.h" |
| 12 | #if QT_CONFIG(textcodec) |
| 13 | #include "qtextcodec.h" |
| 14 | #endif |
| 15 | #include "qbuffer.h" |
| 16 | #if QT_CONFIG(regularexpression) |
| 17 | #include "qregularexpression.h" |
| 18 | #endif |
| 19 | #include "qmap.h" |
| 20 | #include "qhash.h" |
| 21 | #include "qstack.h" |
| 22 | #include <qdebug.h> |
| 23 | |
| 24 | #ifdef Q_CC_BOR // borland 6 finds bogus warnings when building this file in uic3 |
| 25 | # pragma warn -8080 |
| 26 | #endif |
| 27 | |
| 28 | //#define QT_QXML_DEBUG |
| 29 | |
| 30 | // Error strings for the XML reader |
| 31 | #define XMLERR_OK QT_TRANSLATE_NOOP("QXml", "no error occurred") |
| 32 | #define XMLERR_ERRORBYCONSUMER QT_TRANSLATE_NOOP("QXml", "error triggered by consumer") |
| 33 | #define XMLERR_UNEXPECTEDEOF QT_TRANSLATE_NOOP("QXml", "unexpected end of file") |
| 34 | #define XMLERR_MORETHANONEDOCTYPE QT_TRANSLATE_NOOP("QXml", "more than one document type definition") |
| 35 | #define XMLERR_ERRORPARSINGELEMENT QT_TRANSLATE_NOOP("QXml", "error occurred while parsing element") |
| 36 | #define XMLERR_TAGMISMATCH QT_TRANSLATE_NOOP("QXml", "tag mismatch") |
| 37 | #define XMLERR_ERRORPARSINGCONTENT QT_TRANSLATE_NOOP("QXml", "error occurred while parsing content") |
| 38 | #define XMLERR_UNEXPECTEDCHARACTER QT_TRANSLATE_NOOP("QXml", "unexpected character") |
| 39 | #define XMLERR_INVALIDNAMEFORPI QT_TRANSLATE_NOOP("QXml", "invalid name for processing instruction") |
| 40 | #define XMLERR_VERSIONEXPECTED QT_TRANSLATE_NOOP("QXml", "version expected while reading the XML declaration") |
| 41 | #define XMLERR_WRONGVALUEFORSDECL QT_TRANSLATE_NOOP("QXml", "wrong value for standalone declaration") |
| 42 | #define XMLERR_EDECLORSDDECLEXPECTED QT_TRANSLATE_NOOP("QXml", "encoding declaration or standalone declaration expected while reading the XML declaration") |
| 43 | #define XMLERR_SDDECLEXPECTED QT_TRANSLATE_NOOP("QXml", "standalone declaration expected while reading the XML declaration") |
| 44 | #define XMLERR_ERRORPARSINGDOCTYPE QT_TRANSLATE_NOOP("QXml", "error occurred while parsing document type definition") |
| 45 | #define XMLERR_LETTEREXPECTED QT_TRANSLATE_NOOP("QXml", "letter is expected") |
| 46 | #define QT_TRANSLATE_NOOP("QXml", "error occurred while parsing comment") |
| 47 | #define XMLERR_ERRORPARSINGREFERENCE QT_TRANSLATE_NOOP("QXml", "error occurred while parsing reference") |
| 48 | #define XMLERR_INTERNALGENERALENTITYINDTD QT_TRANSLATE_NOOP("QXml", "internal general entity reference not allowed in DTD") |
| 49 | #define XMLERR_EXTERNALGENERALENTITYINAV QT_TRANSLATE_NOOP("QXml", "external parsed general entity reference not allowed in attribute value") |
| 50 | #define XMLERR_EXTERNALGENERALENTITYINDTD QT_TRANSLATE_NOOP("QXml", "external parsed general entity reference not allowed in DTD") |
| 51 | #define XMLERR_UNPARSEDENTITYREFERENCE QT_TRANSLATE_NOOP("QXml", "unparsed entity reference in wrong context") |
| 52 | #define XMLERR_RECURSIVEENTITIES QT_TRANSLATE_NOOP("QXml", "recursive entities") |
| 53 | #define XMLERR_ERRORINTEXTDECL QT_TRANSLATE_NOOP("QXml", "error in the text declaration of an external entity") |
| 54 | |
| 55 | QT_BEGIN_NAMESPACE |
| 56 | |
| 57 | namespace { |
| 58 | |
| 59 | // work around missing std::stack::clear() |
| 60 | template <typename Container> |
| 61 | void clear(Container &c) { c = Container(); } |
| 62 | |
| 63 | } |
| 64 | |
| 65 | // the constants for the lookup table |
| 66 | static const signed char cltWS = 0; // white space |
| 67 | static const signed char cltPer = 1; // % |
| 68 | static const signed char cltAmp = 2; // & |
| 69 | static const signed char cltGt = 3; // > |
| 70 | static const signed char cltLt = 4; // < |
| 71 | static const signed char cltSlash = 5; // / |
| 72 | static const signed char cltQm = 6; // ? |
| 73 | static const signed char cltEm = 7; // ! |
| 74 | static const signed char cltDash = 8; // - |
| 75 | static const signed char cltCB = 9; // ] |
| 76 | static const signed char cltOB = 10; // [ |
| 77 | static const signed char cltEq = 11; // = |
| 78 | static const signed char cltDq = 12; // " |
| 79 | static const signed char cltSq = 13; // ' |
| 80 | static const signed char cltUnknown = 14; |
| 81 | |
| 82 | // character lookup table |
| 83 | static const signed char charLookupTable[256]={ |
| 84 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x00 - 0x07 |
| 85 | cltUnknown, // 0x08 |
| 86 | cltWS, // 0x09 \t |
| 87 | cltWS, // 0x0A \n |
| 88 | cltUnknown, // 0x0B |
| 89 | cltUnknown, // 0x0C |
| 90 | cltWS, // 0x0D \r |
| 91 | cltUnknown, // 0x0E |
| 92 | cltUnknown, // 0x0F |
| 93 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x17 - 0x16 |
| 94 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x18 - 0x1F |
| 95 | cltWS, // 0x20 Space |
| 96 | cltEm, // 0x21 ! |
| 97 | cltDq, // 0x22 " |
| 98 | cltUnknown, // 0x23 |
| 99 | cltUnknown, // 0x24 |
| 100 | cltPer, // 0x25 % |
| 101 | cltAmp, // 0x26 & |
| 102 | cltSq, // 0x27 ' |
| 103 | cltUnknown, // 0x28 |
| 104 | cltUnknown, // 0x29 |
| 105 | cltUnknown, // 0x2A |
| 106 | cltUnknown, // 0x2B |
| 107 | cltUnknown, // 0x2C |
| 108 | cltDash, // 0x2D - |
| 109 | cltUnknown, // 0x2E |
| 110 | cltSlash, // 0x2F / |
| 111 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x30 - 0x37 |
| 112 | cltUnknown, // 0x38 |
| 113 | cltUnknown, // 0x39 |
| 114 | cltUnknown, // 0x3A |
| 115 | cltUnknown, // 0x3B |
| 116 | cltLt, // 0x3C < |
| 117 | cltEq, // 0x3D = |
| 118 | cltGt, // 0x3E > |
| 119 | cltQm, // 0x3F ? |
| 120 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x40 - 0x47 |
| 121 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x48 - 0x4F |
| 122 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x50 - 0x57 |
| 123 | cltUnknown, // 0x58 |
| 124 | cltUnknown, // 0x59 |
| 125 | cltUnknown, // 0x5A |
| 126 | cltOB, // 0x5B [ |
| 127 | cltUnknown, // 0x5C |
| 128 | cltCB, // 0x5D] |
| 129 | cltUnknown, // 0x5E |
| 130 | cltUnknown, // 0x5F |
| 131 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x60 - 0x67 |
| 132 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x68 - 0x6F |
| 133 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x70 - 0x77 |
| 134 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x78 - 0x7F |
| 135 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x80 - 0x87 |
| 136 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x88 - 0x8F |
| 137 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x90 - 0x97 |
| 138 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0x98 - 0x9F |
| 139 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xA0 - 0xA7 |
| 140 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xA8 - 0xAF |
| 141 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xB0 - 0xB7 |
| 142 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xB8 - 0xBF |
| 143 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xC0 - 0xC7 |
| 144 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xC8 - 0xCF |
| 145 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xD0 - 0xD7 |
| 146 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xD8 - 0xDF |
| 147 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xE0 - 0xE7 |
| 148 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xE8 - 0xEF |
| 149 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, // 0xF0 - 0xF7 |
| 150 | cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown, cltUnknown // 0xF8 - 0xFF |
| 151 | }; |
| 152 | |
| 153 | // |
| 154 | // local helper functions |
| 155 | // |
| 156 | |
| 157 | /* |
| 158 | This function strips the TextDecl [77] ("<?xml ...?>") from the string \a |
| 159 | str. The stripped version is stored in \a str. If this function finds an |
| 160 | invalid TextDecl, it returns \c false, otherwise true. |
| 161 | |
| 162 | This function is used for external entities since those can include an |
| 163 | TextDecl that must be stripped before inserting the entity. |
| 164 | */ |
| 165 | static bool stripTextDecl(QString& str) |
| 166 | { |
| 167 | QLatin1String textDeclStart("<?xml" ); |
| 168 | if (str.startsWith(s: textDeclStart)) { |
| 169 | #if QT_CONFIG(regularexpression) |
| 170 | QRegularExpression textDecl(QString::fromLatin1( |
| 171 | ba: "^<\\?xml\\s+" |
| 172 | "(version\\s*=\\s*((['\"])[-a-zA-Z0-9_.:]+\\3))?" |
| 173 | "\\s*" |
| 174 | "(encoding\\s*=\\s*((['\"])[A-Za-z][-a-zA-Z0-9_.]*\\6))?" |
| 175 | "\\s*\\?>" |
| 176 | )); |
| 177 | QString strTmp = str.replace(re: textDecl, after: QLatin1String("" )); |
| 178 | if (strTmp.size() != str.size()) |
| 179 | return false; // external entity has wrong TextDecl |
| 180 | str = strTmp; |
| 181 | #else |
| 182 | return false; |
| 183 | #endif |
| 184 | } |
| 185 | return true; |
| 186 | } |
| 187 | |
| 188 | class QXmlAttributesPrivate |
| 189 | { |
| 190 | }; |
| 191 | |
| 192 | /* \class QXmlInputSourcePrivate |
| 193 | \internal |
| 194 | |
| 195 | There's a slight misdesign in this class that can |
| 196 | be worth to keep in mind: the `str' member is |
| 197 | a buffer which QXmlInputSource::next() returns from, |
| 198 | and which is populated from the input device or input |
| 199 | stream. However, when the input is a QString(the user called |
| 200 | QXmlInputSource::setData()), `str' has two roles: it's the |
| 201 | buffer, but also the source. This /seems/ to be no problem |
| 202 | because in the case of having no device or stream, the QString |
| 203 | is read in one go. |
| 204 | */ |
| 205 | class QXmlInputSourcePrivate |
| 206 | { |
| 207 | public: |
| 208 | QIODevice *inputDevice; |
| 209 | QTextStream *inputStream; |
| 210 | |
| 211 | QString str; |
| 212 | const QChar *unicode; |
| 213 | int pos; |
| 214 | int length; |
| 215 | bool nextReturnedEndOfData; |
| 216 | #if QT_CONFIG(textcodec) |
| 217 | QTextDecoder *encMapper; |
| 218 | #endif |
| 219 | |
| 220 | QByteArray encodingDeclBytes; |
| 221 | QString encodingDeclChars; |
| 222 | bool lookingForEncodingDecl; |
| 223 | }; |
| 224 | class QXmlParseExceptionPrivate |
| 225 | { |
| 226 | public: |
| 227 | QXmlParseExceptionPrivate() |
| 228 | : column(-1), line(-1) |
| 229 | { |
| 230 | } |
| 231 | QXmlParseExceptionPrivate(const QXmlParseExceptionPrivate &other) |
| 232 | : msg(other.msg), column(other.column), line(other.line), |
| 233 | pub(other.pub), sys(other.sys) |
| 234 | { |
| 235 | } |
| 236 | |
| 237 | QString msg; |
| 238 | int column; |
| 239 | int line; |
| 240 | QString pub; |
| 241 | QString sys; |
| 242 | |
| 243 | }; |
| 244 | |
| 245 | class QXmlLocatorPrivate |
| 246 | { |
| 247 | }; |
| 248 | |
| 249 | class QXmlDefaultHandlerPrivate |
| 250 | { |
| 251 | }; |
| 252 | |
| 253 | /*! |
| 254 | \class QXmlParseException |
| 255 | \reentrant |
| 256 | \brief The QXmlParseException class is used to report errors with |
| 257 | the QXmlErrorHandler interface. |
| 258 | |
| 259 | \inmodule QtCore5Compat |
| 260 | \ingroup xml-tools |
| 261 | |
| 262 | The XML subsystem constructs an instance of this class when it |
| 263 | detects an error. You can retrieve the place where the error |
| 264 | occurred using systemId(), publicId(), lineNumber() and |
| 265 | columnNumber(), along with the error message(). The possible error |
| 266 | messages are: |
| 267 | |
| 268 | \list |
| 269 | \li "no error occurred" |
| 270 | \li "error triggered by consumer" |
| 271 | \li "unexpected end of file" |
| 272 | \li "more than one document type definition" |
| 273 | \li "error occurred while parsing element" |
| 274 | \li "tag mismatch" |
| 275 | \li "error occurred while parsing content" |
| 276 | \li "unexpected character" |
| 277 | \li "invalid name for processing instruction" |
| 278 | \li "version expected while reading the XML declaration" |
| 279 | \li "wrong value for standalone declaration" |
| 280 | \li "encoding declaration or standalone declaration expected while reading the XML declaration" |
| 281 | \li "standalone declaration expected while reading the XML declaration" |
| 282 | \li "error occurred while parsing document type definition" |
| 283 | \li "letter is expected" |
| 284 | \li "error occurred while parsing comment" |
| 285 | \li "error occurred while parsing reference" |
| 286 | \li "internal general entity reference not allowed in DTD" |
| 287 | \li "external parsed general entity reference not allowed in attribute value" |
| 288 | \li "external parsed general entity reference not allowed in DTD" |
| 289 | \li "unparsed entity reference n wrong context" |
| 290 | \li "recursive entities" |
| 291 | \li "error in the text declaration of an external entity" |
| 292 | \endlist |
| 293 | |
| 294 | Note that, if you want to display these error messages to your |
| 295 | application's users, they will be displayed in English unless |
| 296 | they are explicitly translated. |
| 297 | |
| 298 | \sa QXmlErrorHandler, QXmlReader |
| 299 | */ |
| 300 | |
| 301 | /*! |
| 302 | Constructs a parse exception with the error string \a name for |
| 303 | column \a c and line \a l for the public identifier \a p and the |
| 304 | system identifier \a s. |
| 305 | */ |
| 306 | |
| 307 | QXmlParseException::QXmlParseException(const QString& name, int c, int l, |
| 308 | const QString& p, const QString& s) |
| 309 | : d(new QXmlParseExceptionPrivate) |
| 310 | { |
| 311 | d->msg = name; |
| 312 | d->column = c; |
| 313 | d->line = l; |
| 314 | d->pub = p; |
| 315 | d->sys = s; |
| 316 | } |
| 317 | |
| 318 | /*! |
| 319 | Creates a copy of \a other. |
| 320 | */ |
| 321 | QXmlParseException::QXmlParseException(const QXmlParseException& other) : |
| 322 | d(new QXmlParseExceptionPrivate(*other.d)) |
| 323 | { |
| 324 | |
| 325 | } |
| 326 | |
| 327 | /*! |
| 328 | Destroys the QXmlParseException. |
| 329 | */ |
| 330 | QXmlParseException::~QXmlParseException() |
| 331 | { |
| 332 | } |
| 333 | |
| 334 | /*! |
| 335 | Returns the error message. |
| 336 | */ |
| 337 | QString QXmlParseException::message() const |
| 338 | { |
| 339 | return d->msg; |
| 340 | } |
| 341 | /*! |
| 342 | Returns the column number where the error occurred. |
| 343 | */ |
| 344 | int QXmlParseException::columnNumber() const |
| 345 | { |
| 346 | return d->column; |
| 347 | } |
| 348 | /*! |
| 349 | Returns the line number where the error occurred. |
| 350 | */ |
| 351 | int QXmlParseException::lineNumber() const |
| 352 | { |
| 353 | return d->line; |
| 354 | } |
| 355 | /*! |
| 356 | Returns the public identifier where the error occurred. |
| 357 | */ |
| 358 | QString QXmlParseException::publicId() const |
| 359 | { |
| 360 | return d->pub; |
| 361 | } |
| 362 | /*! |
| 363 | Returns the system identifier where the error occurred. |
| 364 | */ |
| 365 | QString QXmlParseException::systemId() const |
| 366 | { |
| 367 | return d->sys; |
| 368 | } |
| 369 | |
| 370 | /*! |
| 371 | \class QXmlLocator |
| 372 | \reentrant |
| 373 | \brief The QXmlLocator class provides the XML handler classes with |
| 374 | information about the parsing position within a file. |
| 375 | |
| 376 | \inmodule QtCore5Compat |
| 377 | \ingroup xml-tools |
| 378 | |
| 379 | The reader reports a QXmlLocator to the content handler before it |
| 380 | starts to parse the document. This is done with the |
| 381 | QXmlContentHandler::setDocumentLocator() function. The handler |
| 382 | classes can now use this locator to get the position (lineNumber() |
| 383 | and columnNumber()) that the reader has reached. |
| 384 | */ |
| 385 | |
| 386 | /*! |
| 387 | Constructor. |
| 388 | */ |
| 389 | QXmlLocator::QXmlLocator() |
| 390 | { |
| 391 | } |
| 392 | |
| 393 | /*! |
| 394 | Destructor. |
| 395 | */ |
| 396 | QXmlLocator::~QXmlLocator() |
| 397 | { |
| 398 | } |
| 399 | |
| 400 | /*! |
| 401 | \fn int QXmlLocator::columnNumber() const |
| 402 | |
| 403 | Returns the column number (starting at 1) or -1 if there is no |
| 404 | column number available. |
| 405 | */ |
| 406 | |
| 407 | /*! |
| 408 | \fn int QXmlLocator::lineNumber() const |
| 409 | |
| 410 | Returns the line number (starting at 1) or -1 if there is no line |
| 411 | number available. |
| 412 | */ |
| 413 | |
| 414 | class QXmlSimpleReaderLocator : public QXmlLocator |
| 415 | { |
| 416 | public: |
| 417 | QXmlSimpleReaderLocator(QXmlSimpleReader* parent) |
| 418 | { |
| 419 | reader = parent; |
| 420 | } |
| 421 | ~QXmlSimpleReaderLocator() |
| 422 | { |
| 423 | } |
| 424 | |
| 425 | int columnNumber() const override |
| 426 | { |
| 427 | return (reader->d_ptr->columnNr == -1 ? -1 : reader->d_ptr->columnNr + 1); |
| 428 | } |
| 429 | int lineNumber() const override |
| 430 | { |
| 431 | return (reader->d_ptr->lineNr == -1 ? -1 : reader->d_ptr->lineNr + 1); |
| 432 | } |
| 433 | |
| 434 | private: |
| 435 | QXmlSimpleReader *reader; |
| 436 | }; |
| 437 | |
| 438 | /********************************************* |
| 439 | * |
| 440 | * QXmlNamespaceSupport |
| 441 | * |
| 442 | *********************************************/ |
| 443 | |
| 444 | typedef QMap<QString, QString> NamespaceMap; |
| 445 | |
| 446 | class QXmlNamespaceSupportPrivate |
| 447 | { |
| 448 | public: |
| 449 | QXmlNamespaceSupportPrivate() |
| 450 | { |
| 451 | ns.insert(key: QLatin1String("xml" ), value: QLatin1String("http://www.w3.org/XML/1998/namespace" )); // the XML namespace |
| 452 | } |
| 453 | |
| 454 | ~QXmlNamespaceSupportPrivate() |
| 455 | { |
| 456 | } |
| 457 | |
| 458 | QStack<NamespaceMap> nsStack; |
| 459 | NamespaceMap ns; |
| 460 | }; |
| 461 | |
| 462 | /*! |
| 463 | \class QXmlNamespaceSupport |
| 464 | \since 4.4 |
| 465 | \reentrant |
| 466 | \brief The QXmlNamespaceSupport class is a helper class for XML |
| 467 | readers which want to include namespace support. |
| 468 | |
| 469 | \inmodule QtCore5Compat |
| 470 | \ingroup xml-tools |
| 471 | |
| 472 | You can set the prefix for the current namespace with setPrefix(), |
| 473 | and get the list of current prefixes (or those for a given URI) |
| 474 | with prefixes(). The namespace URI is available from uri(). Use |
| 475 | pushContext() to start a new namespace context, and popContext() |
| 476 | to return to the previous namespace context. Use splitName() or |
| 477 | processName() to split a name into its prefix and local name. |
| 478 | */ |
| 479 | |
| 480 | /*! |
| 481 | Constructs a QXmlNamespaceSupport. |
| 482 | */ |
| 483 | QXmlNamespaceSupport::QXmlNamespaceSupport() |
| 484 | { |
| 485 | d = new QXmlNamespaceSupportPrivate; |
| 486 | } |
| 487 | |
| 488 | /*! |
| 489 | Destroys a QXmlNamespaceSupport. |
| 490 | */ |
| 491 | QXmlNamespaceSupport::~QXmlNamespaceSupport() |
| 492 | { |
| 493 | delete d; |
| 494 | } |
| 495 | |
| 496 | /*! |
| 497 | This function declares a prefix \a pre in the current namespace |
| 498 | context to be the namespace URI \a uri. The prefix remains in |
| 499 | force until this context is popped, unless it is shadowed in a |
| 500 | descendant context. |
| 501 | |
| 502 | Note that there is an asymmetry in this library. prefix() does not |
| 503 | return the default "" prefix, even if you have declared one; to |
| 504 | check for a default prefix, you must look it up explicitly using |
| 505 | uri(). This asymmetry exists to make it easier to look up prefixes |
| 506 | for attribute names, where the default prefix is not allowed. |
| 507 | */ |
| 508 | void QXmlNamespaceSupport::setPrefix(const QString& pre, const QString& uri) |
| 509 | { |
| 510 | if (pre.isNull()) { |
| 511 | d->ns.insert(key: QLatin1String("" ), value: uri); |
| 512 | } else { |
| 513 | d->ns.insert(key: pre, value: uri); |
| 514 | } |
| 515 | } |
| 516 | |
| 517 | /*! |
| 518 | Returns one of the prefixes mapped to the namespace URI \a uri. |
| 519 | |
| 520 | If more than one prefix is currently mapped to the same URI, this |
| 521 | function makes an arbitrary selection; if you want all of the |
| 522 | prefixes, use prefixes() instead. |
| 523 | |
| 524 | Note: to check for a default prefix, use the uri() function with |
| 525 | an argument of "". |
| 526 | */ |
| 527 | QString QXmlNamespaceSupport::prefix(const QString& uri) const |
| 528 | { |
| 529 | NamespaceMap::const_iterator itc, it = d->ns.constBegin(); |
| 530 | while ((itc=it) != d->ns.constEnd()) { |
| 531 | ++it; |
| 532 | if (*itc == uri && !itc.key().isEmpty()) |
| 533 | return itc.key(); |
| 534 | } |
| 535 | return QLatin1String("" ); |
| 536 | } |
| 537 | |
| 538 | /*! |
| 539 | Looks up the prefix \a prefix in the current context and returns |
| 540 | the currently-mapped namespace URI. Use the empty string ("") for |
| 541 | the default namespace. |
| 542 | */ |
| 543 | QString QXmlNamespaceSupport::uri(const QString& prefix) const |
| 544 | { |
| 545 | return d->ns[prefix]; |
| 546 | } |
| 547 | |
| 548 | /*! |
| 549 | Splits the name \a qname at the ':' and returns the prefix in \a |
| 550 | prefix and the local name in \a localname. |
| 551 | |
| 552 | \sa processName() |
| 553 | */ |
| 554 | void QXmlNamespaceSupport::splitName(const QString& qname, QString& prefix, |
| 555 | QString& localname) const |
| 556 | { |
| 557 | int pos = qname.indexOf(ch: QLatin1Char(':')); |
| 558 | if (pos == -1) |
| 559 | pos = qname.size(); |
| 560 | |
| 561 | prefix = qname.left(n: pos); |
| 562 | localname = qname.mid(position: pos+1); |
| 563 | } |
| 564 | |
| 565 | /*! |
| 566 | Processes a raw XML 1.0 name in the current context by removing |
| 567 | the prefix and looking it up among the prefixes currently |
| 568 | declared. |
| 569 | |
| 570 | \a qname is the raw XML 1.0 name to be processed. \a isAttribute |
| 571 | is true if the name is an attribute name. |
| 572 | |
| 573 | This function stores the namespace URI in \a nsuri (which will be |
| 574 | set to an empty string if the raw name has an undeclared prefix), |
| 575 | and stores the local name (without prefix) in \a localname (which |
| 576 | will be set to an empty string if no namespace is in use). |
| 577 | |
| 578 | Note that attribute names are processed differently than element |
| 579 | names: an unprefixed element name gets the default namespace (if |
| 580 | any), while an unprefixed attribute name does not. |
| 581 | */ |
| 582 | void QXmlNamespaceSupport::processName(const QString& qname, |
| 583 | bool isAttribute, |
| 584 | QString& nsuri, QString& localname) const |
| 585 | { |
| 586 | int len = qname.size(); |
| 587 | const QChar *data = qname.constData(); |
| 588 | for (int pos = 0; pos < len; ++pos) { |
| 589 | if (data[pos] == QLatin1Char(':')) { |
| 590 | nsuri = uri(prefix: qname.left(n: pos)); |
| 591 | localname = qname.mid(position: pos + 1); |
| 592 | return; |
| 593 | } |
| 594 | } |
| 595 | |
| 596 | // there was no ':' |
| 597 | nsuri.clear(); |
| 598 | // attributes don't take default namespace |
| 599 | if (!isAttribute && !d->ns.isEmpty()) { |
| 600 | /* |
| 601 | We want to access d->ns.value(""), but as an optimization |
| 602 | we use the fact that "" compares less than any other |
| 603 | string, so it's either first in the map or not there. |
| 604 | */ |
| 605 | NamespaceMap::const_iterator first = d->ns.constBegin(); |
| 606 | if (first.key().isEmpty()) |
| 607 | nsuri = first.value(); // get default namespace |
| 608 | } |
| 609 | localname = qname; |
| 610 | } |
| 611 | |
| 612 | /*! |
| 613 | Returns a list of all the prefixes currently declared. |
| 614 | |
| 615 | If there is a default prefix, this function does not return it in |
| 616 | the list; check for the default prefix using uri() with an |
| 617 | argument of "". |
| 618 | */ |
| 619 | QStringList QXmlNamespaceSupport::prefixes() const |
| 620 | { |
| 621 | QStringList list; |
| 622 | |
| 623 | NamespaceMap::const_iterator itc, it = d->ns.constBegin(); |
| 624 | while ((itc=it) != d->ns.constEnd()) { |
| 625 | ++it; |
| 626 | if (!itc.key().isEmpty()) |
| 627 | list.append(t: itc.key()); |
| 628 | } |
| 629 | return list; |
| 630 | } |
| 631 | |
| 632 | /*! |
| 633 | \overload |
| 634 | |
| 635 | Returns a list of all prefixes currently declared for the |
| 636 | namespace URI \a uri. |
| 637 | |
| 638 | The "xml:" prefix is included. If you only want one prefix that is |
| 639 | mapped to the namespace URI, and you don't care which one you get, |
| 640 | use the prefix() function instead. |
| 641 | |
| 642 | Note: The empty (default) prefix is never included in this list; |
| 643 | to check for the presence of a default namespace, call uri() with |
| 644 | "" as the argument. |
| 645 | */ |
| 646 | QStringList QXmlNamespaceSupport::prefixes(const QString& uri) const |
| 647 | { |
| 648 | QStringList list; |
| 649 | |
| 650 | NamespaceMap::const_iterator itc, it = d->ns.constBegin(); |
| 651 | while ((itc=it) != d->ns.constEnd()) { |
| 652 | ++it; |
| 653 | if (*itc == uri && !itc.key().isEmpty()) |
| 654 | list.append(t: itc.key()); |
| 655 | } |
| 656 | return list; |
| 657 | } |
| 658 | |
| 659 | /*! |
| 660 | Starts a new namespace context. |
| 661 | |
| 662 | Normally, you should push a new context at the beginning of each |
| 663 | XML element: the new context automatically inherits the |
| 664 | declarations of its parent context, and it also keeps track of |
| 665 | which declarations were made within this context. |
| 666 | |
| 667 | \sa popContext() |
| 668 | */ |
| 669 | void QXmlNamespaceSupport::pushContext() |
| 670 | { |
| 671 | d->nsStack.push(t: d->ns); |
| 672 | } |
| 673 | |
| 674 | /*! |
| 675 | Reverts to the previous namespace context. |
| 676 | |
| 677 | Normally, you should pop the context at the end of each XML |
| 678 | element. After popping the context, all namespace prefix mappings |
| 679 | that were previously in force are restored. |
| 680 | |
| 681 | \sa pushContext() |
| 682 | */ |
| 683 | void QXmlNamespaceSupport::popContext() |
| 684 | { |
| 685 | d->ns.clear(); |
| 686 | if (!d->nsStack.isEmpty()) |
| 687 | d->ns = d->nsStack.pop(); |
| 688 | } |
| 689 | |
| 690 | /*! |
| 691 | Resets this namespace support object ready for reuse. |
| 692 | */ |
| 693 | void QXmlNamespaceSupport::reset() |
| 694 | { |
| 695 | QXmlNamespaceSupportPrivate *newD = new QXmlNamespaceSupportPrivate; |
| 696 | delete d; |
| 697 | d = newD; |
| 698 | } |
| 699 | |
| 700 | /********************************************* |
| 701 | * |
| 702 | * QXmlAttributes |
| 703 | * |
| 704 | *********************************************/ |
| 705 | |
| 706 | /*! |
| 707 | \class QXmlAttributes |
| 708 | \reentrant |
| 709 | \brief The QXmlAttributes class provides XML attributes. |
| 710 | |
| 711 | \inmodule QtCore5Compat |
| 712 | \ingroup xml-tools |
| 713 | |
| 714 | If attributes are reported by QXmlContentHandler::startElement() |
| 715 | this class is used to pass the attribute values. |
| 716 | |
| 717 | Use index() to locate the position of an attribute in the list, |
| 718 | count() to retrieve the number of attributes, and clear() to |
| 719 | remove the attributes. New attributes can be added with append(). |
| 720 | Use type() to get an attribute's type and value() to get its |
| 721 | value. The attribute's name is available from localName() or |
| 722 | qName(), and its namespace URI from uri(). |
| 723 | |
| 724 | */ |
| 725 | |
| 726 | /*! |
| 727 | \fn QXmlAttributes::QXmlAttributes() |
| 728 | |
| 729 | Constructs an empty attribute list. |
| 730 | */ |
| 731 | QXmlAttributes::QXmlAttributes() |
| 732 | { |
| 733 | // ### In Qt 5.0, this function was inlined and d was not initialized |
| 734 | // The member cannot be used until Qt 6.0 |
| 735 | Q_UNUSED(d); |
| 736 | } |
| 737 | |
| 738 | /*! |
| 739 | \fn QXmlAttributes::~QXmlAttributes() |
| 740 | |
| 741 | Destroys the attributes object. |
| 742 | */ |
| 743 | QXmlAttributes::~QXmlAttributes() |
| 744 | { |
| 745 | } |
| 746 | |
| 747 | /*! |
| 748 | \fn void QXmlAttributes::swap(QXmlAttributes &other) |
| 749 | |
| 750 | Swaps \c this with \a other. |
| 751 | */ |
| 752 | |
| 753 | /*! |
| 754 | Looks up the index of an attribute by the qualified name \a qName. |
| 755 | |
| 756 | Returns the index of the attribute or -1 if it wasn't found. |
| 757 | */ |
| 758 | int QXmlAttributes::index(const QString& qName) const |
| 759 | { |
| 760 | for (int i = 0; i < attList.size(); ++i) { |
| 761 | if (attList.at(i).qname == qName) |
| 762 | return i; |
| 763 | } |
| 764 | return -1; |
| 765 | } |
| 766 | |
| 767 | /*! \overload |
| 768 | */ |
| 769 | int QXmlAttributes::index(QLatin1String qName) const |
| 770 | { |
| 771 | for (int i = 0; i < attList.size(); ++i) { |
| 772 | if (attList.at(i).qname == qName) |
| 773 | return i; |
| 774 | } |
| 775 | return -1; |
| 776 | } |
| 777 | |
| 778 | /*! |
| 779 | \overload |
| 780 | |
| 781 | Looks up the index of an attribute by a namespace name. |
| 782 | |
| 783 | \a uri specifies the namespace URI, or an empty string if the name |
| 784 | has no namespace URI. \a localPart specifies the attribute's local |
| 785 | name. |
| 786 | |
| 787 | Returns the index of the attribute, or -1 if it wasn't found. |
| 788 | */ |
| 789 | int QXmlAttributes::index(const QString& uri, const QString& localPart) const |
| 790 | { |
| 791 | for (int i = 0; i < attList.size(); ++i) { |
| 792 | const Attribute &att = attList.at(i); |
| 793 | if (att.uri == uri && att.localname == localPart) |
| 794 | return i; |
| 795 | } |
| 796 | return -1; |
| 797 | } |
| 798 | |
| 799 | /*! |
| 800 | Returns the number of attributes in the list. |
| 801 | |
| 802 | \sa count() |
| 803 | */ |
| 804 | int QXmlAttributes::length() const |
| 805 | { |
| 806 | return attList.size(); |
| 807 | } |
| 808 | |
| 809 | /*! |
| 810 | \fn int QXmlAttributes::count() const |
| 811 | |
| 812 | Returns the number of attributes in the list. This function is |
| 813 | equivalent to length(). |
| 814 | */ |
| 815 | |
| 816 | /*! |
| 817 | Looks up an attribute's local name for the attribute at position |
| 818 | \a index. If no namespace processing is done, the local name is |
| 819 | an empty string. |
| 820 | */ |
| 821 | QString QXmlAttributes::localName(int index) const |
| 822 | { |
| 823 | return attList.at(i: index).localname; |
| 824 | } |
| 825 | |
| 826 | /*! |
| 827 | Looks up an attribute's XML 1.0 qualified name for the attribute |
| 828 | at position \a index. |
| 829 | */ |
| 830 | QString QXmlAttributes::qName(int index) const |
| 831 | { |
| 832 | return attList.at(i: index).qname; |
| 833 | } |
| 834 | |
| 835 | /*! |
| 836 | Looks up an attribute's namespace URI for the attribute at |
| 837 | position \a index. If no namespace processing is done or if the |
| 838 | attribute has no namespace, the namespace URI is an empty string. |
| 839 | */ |
| 840 | QString QXmlAttributes::uri(int index) const |
| 841 | { |
| 842 | return attList.at(i: index).uri; |
| 843 | } |
| 844 | |
| 845 | /*! |
| 846 | Looks up an attribute's type for the attribute at position \a |
| 847 | index. |
| 848 | |
| 849 | Currently only "CDATA" is returned. |
| 850 | */ |
| 851 | QString QXmlAttributes::type(int) const |
| 852 | { |
| 853 | return QLatin1String("CDATA" ); |
| 854 | } |
| 855 | |
| 856 | /*! |
| 857 | \overload |
| 858 | |
| 859 | Looks up an attribute's type for the qualified name \a qName. |
| 860 | |
| 861 | Currently only "CDATA" is returned. |
| 862 | */ |
| 863 | QString QXmlAttributes::type(const QString&) const |
| 864 | { |
| 865 | return QLatin1String("CDATA" ); |
| 866 | } |
| 867 | |
| 868 | /*! |
| 869 | \overload |
| 870 | |
| 871 | Looks up an attribute's type by namespace name. |
| 872 | |
| 873 | \a uri specifies the namespace URI and \a localName specifies the |
| 874 | local name. If the name has no namespace URI, use an empty string |
| 875 | for \a uri. |
| 876 | |
| 877 | Currently only "CDATA" is returned. |
| 878 | */ |
| 879 | QString QXmlAttributes::type(const QString&, const QString&) const |
| 880 | { |
| 881 | return QLatin1String("CDATA" ); |
| 882 | } |
| 883 | |
| 884 | /*! |
| 885 | Returns an attribute's value for the attribute at position \a |
| 886 | index. The index must be a valid position |
| 887 | (i.e., 0 <= \a index < count()). |
| 888 | */ |
| 889 | QString QXmlAttributes::value(int index) const |
| 890 | { |
| 891 | return attList.at(i: index).value; |
| 892 | } |
| 893 | |
| 894 | /*! |
| 895 | \overload |
| 896 | |
| 897 | Returns an attribute's value for the qualified name \a qName, or an |
| 898 | empty string if no attribute exists for the name given. |
| 899 | */ |
| 900 | QString QXmlAttributes::value(const QString& qName) const |
| 901 | { |
| 902 | int i = index(qName); |
| 903 | if (i == -1) |
| 904 | return QString(); |
| 905 | return attList.at(i).value; |
| 906 | } |
| 907 | |
| 908 | /*! |
| 909 | \overload |
| 910 | |
| 911 | Returns an attribute's value for the qualified name \a qName, or an |
| 912 | empty string if no attribute exists for the name given. |
| 913 | */ |
| 914 | QString QXmlAttributes::value(QLatin1String qName) const |
| 915 | { |
| 916 | int i = index(qName); |
| 917 | if (i == -1) |
| 918 | return QString(); |
| 919 | return attList.at(i).value; |
| 920 | } |
| 921 | |
| 922 | /*! |
| 923 | \overload |
| 924 | |
| 925 | Returns an attribute's value by namespace name. |
| 926 | |
| 927 | \a uri specifies the namespace URI, or an empty string if the name |
| 928 | has no namespace URI. \a localName specifies the attribute's local |
| 929 | name. |
| 930 | */ |
| 931 | QString QXmlAttributes::value(const QString& uri, const QString& localName) const |
| 932 | { |
| 933 | int i = index(uri, localPart: localName); |
| 934 | if (i == -1) |
| 935 | return QString(); |
| 936 | return attList.at(i).value; |
| 937 | } |
| 938 | |
| 939 | /*! |
| 940 | Clears the list of attributes. |
| 941 | |
| 942 | \sa append() |
| 943 | */ |
| 944 | void QXmlAttributes::clear() |
| 945 | { |
| 946 | attList.clear(); |
| 947 | } |
| 948 | |
| 949 | /*! |
| 950 | Appends a new attribute entry to the list of attributes. The |
| 951 | qualified name of the attribute is \a qName, the namespace URI is |
| 952 | \a uri and the local name is \a localPart. The value of the |
| 953 | attribute is \a value. |
| 954 | |
| 955 | \sa qName(), uri(), localName(), value() |
| 956 | */ |
| 957 | void QXmlAttributes::append(const QString &qName, const QString &uri, const QString &localPart, const QString &value) |
| 958 | { |
| 959 | Attribute att; |
| 960 | att.qname = qName; |
| 961 | att.uri = uri; |
| 962 | att.localname = localPart; |
| 963 | att.value = value; |
| 964 | |
| 965 | attList.append(t: att); |
| 966 | } |
| 967 | |
| 968 | /********************************************* |
| 969 | * |
| 970 | * QXmlInputSource |
| 971 | * |
| 972 | *********************************************/ |
| 973 | |
| 974 | /*! |
| 975 | \class QXmlInputSource |
| 976 | \reentrant |
| 977 | \brief The QXmlInputSource class provides the input data for the |
| 978 | QXmlReader subclasses. |
| 979 | |
| 980 | \inmodule QtCore5Compat |
| 981 | \ingroup xml-tools |
| 982 | |
| 983 | All subclasses of QXmlReader read the input XML document from this |
| 984 | class. |
| 985 | |
| 986 | This class recognizes the encoding of the data by reading the |
| 987 | encoding declaration in the XML file if it finds one, and reading |
| 988 | the data using the corresponding encoding. If it does not find an |
| 989 | encoding declaration, then it assumes that the data is either in |
| 990 | UTF-8 or UTF-16, depending on whether it can find a byte-order |
| 991 | mark. |
| 992 | |
| 993 | There are two ways to populate the input source with data: you can |
| 994 | construct it with a QIODevice* so that the input source reads the |
| 995 | data from that device. Or you can set the data explicitly with one |
| 996 | of the setData() functions. |
| 997 | |
| 998 | Usually you either construct a QXmlInputSource that works on a |
| 999 | QIODevice* or you construct an empty QXmlInputSource and set the |
| 1000 | data with setData(). There are only rare occasions where you would |
| 1001 | want to mix both methods. |
| 1002 | |
| 1003 | The QXmlReader subclasses use the next() function to read the |
| 1004 | input character by character. If you want to start from the |
| 1005 | beginning again, use reset(). |
| 1006 | |
| 1007 | The functions data() and fetchData() are useful if you want to do |
| 1008 | something with the data other than parsing, e.g. displaying the |
| 1009 | raw XML file. The benefit of using the QXmlInputClass in such |
| 1010 | cases is that it tries to use the correct encoding. |
| 1011 | |
| 1012 | \sa QXmlReader, QXmlSimpleReader |
| 1013 | */ |
| 1014 | |
| 1015 | // the following two are guaranteed not to be a character |
| 1016 | const char16_t QXmlInputSource::EndOfData = 0xfffe; |
| 1017 | const char16_t QXmlInputSource::EndOfDocument = 0xffff; |
| 1018 | |
| 1019 | /* |
| 1020 | Common part of the constructors. |
| 1021 | */ |
| 1022 | void QXmlInputSource::init() |
| 1023 | { |
| 1024 | d = new QXmlInputSourcePrivate; |
| 1025 | |
| 1026 | QT_TRY { |
| 1027 | d->inputDevice = nullptr; |
| 1028 | d->inputStream = nullptr; |
| 1029 | |
| 1030 | setData(QString()); |
| 1031 | #if QT_CONFIG(textcodec) |
| 1032 | d->encMapper = nullptr; |
| 1033 | #endif |
| 1034 | d->nextReturnedEndOfData = true; // first call to next() will call fetchData() |
| 1035 | |
| 1036 | d->encodingDeclBytes.clear(); |
| 1037 | d->encodingDeclChars.clear(); |
| 1038 | d->lookingForEncodingDecl = true; |
| 1039 | } QT_CATCH(...) { |
| 1040 | delete(d); |
| 1041 | QT_RETHROW; |
| 1042 | } |
| 1043 | } |
| 1044 | |
| 1045 | /*! |
| 1046 | Constructs an input source which contains no data. |
| 1047 | |
| 1048 | \sa setData() |
| 1049 | */ |
| 1050 | QXmlInputSource::QXmlInputSource() |
| 1051 | { |
| 1052 | init(); |
| 1053 | } |
| 1054 | |
| 1055 | /*! |
| 1056 | Constructs an input source and gets the data from device \a dev. |
| 1057 | If \a dev is not open, it is opened in read-only mode. If \a dev |
| 1058 | is 0 or it is not possible to read from the device, the input |
| 1059 | source will contain no data. |
| 1060 | |
| 1061 | \sa setData(), fetchData(), QIODevice |
| 1062 | */ |
| 1063 | QXmlInputSource::QXmlInputSource(QIODevice *dev) |
| 1064 | { |
| 1065 | init(); |
| 1066 | d->inputDevice = dev; |
| 1067 | if (dev->isOpen()) |
| 1068 | d->inputDevice->setTextModeEnabled(false); |
| 1069 | } |
| 1070 | |
| 1071 | /*! |
| 1072 | Destructor. |
| 1073 | */ |
| 1074 | QXmlInputSource::~QXmlInputSource() |
| 1075 | { |
| 1076 | // ### close the input device. |
| 1077 | #if QT_CONFIG(textcodec) |
| 1078 | delete d->encMapper; |
| 1079 | #endif |
| 1080 | delete d; |
| 1081 | } |
| 1082 | |
| 1083 | /*! |
| 1084 | Returns the next character of the input source. If this function |
| 1085 | reaches the end of available data, it returns |
| 1086 | QXmlInputSource::EndOfData. If you call next() after that, it |
| 1087 | tries to fetch more data by calling fetchData(). If the |
| 1088 | fetchData() call results in new data, this function returns the |
| 1089 | first character of that data; otherwise it returns |
| 1090 | QXmlInputSource::EndOfDocument. |
| 1091 | |
| 1092 | Readers, such as QXmlSimpleReader, will assume that the end of |
| 1093 | the XML document has been reached if the this function returns |
| 1094 | QXmlInputSource::EndOfDocument, and will check that the |
| 1095 | supplied input is well-formed. Therefore, when reimplementing |
| 1096 | this function, it is important to ensure that this behavior is |
| 1097 | duplicated. |
| 1098 | |
| 1099 | \sa reset(), fetchData(), QXmlSimpleReader::parse(), |
| 1100 | QXmlSimpleReader::parseContinue() |
| 1101 | */ |
| 1102 | QChar QXmlInputSource::next() |
| 1103 | { |
| 1104 | if (d->pos >= d->length) { |
| 1105 | if (d->nextReturnedEndOfData) { |
| 1106 | d->nextReturnedEndOfData = false; |
| 1107 | fetchData(); |
| 1108 | if (d->pos >= d->length) { |
| 1109 | return EndOfDocument; |
| 1110 | } |
| 1111 | return next(); |
| 1112 | } |
| 1113 | d->nextReturnedEndOfData = true; |
| 1114 | return EndOfData; |
| 1115 | } |
| 1116 | |
| 1117 | // QXmlInputSource has no way to signal encoding errors. The best we can do |
| 1118 | // is return EndOfDocument. We do *not* return EndOfData, because the reader |
| 1119 | // will then just call this function again to get the next char. |
| 1120 | QChar c = d->unicode[d->pos++]; |
| 1121 | if (c == EndOfData) |
| 1122 | c = EndOfDocument; |
| 1123 | return c; |
| 1124 | } |
| 1125 | |
| 1126 | /*! |
| 1127 | This function sets the position used by next() to the beginning of |
| 1128 | the data returned by data(). This is useful if you want to use the |
| 1129 | input source for more than one parse. |
| 1130 | |
| 1131 | \note In the case that the underlying data source is a QIODevice, |
| 1132 | the current position in the device is not automatically set to the |
| 1133 | start of input. Call QIODevice::seek(0) on the device to do this. |
| 1134 | |
| 1135 | \sa next() |
| 1136 | */ |
| 1137 | void QXmlInputSource::reset() |
| 1138 | { |
| 1139 | d->nextReturnedEndOfData = false; |
| 1140 | d->pos = 0; |
| 1141 | } |
| 1142 | |
| 1143 | /*! |
| 1144 | Returns the data the input source contains or an empty string if the |
| 1145 | input source does not contain any data. |
| 1146 | |
| 1147 | \sa setData(), QXmlInputSource(), fetchData() |
| 1148 | */ |
| 1149 | QString QXmlInputSource::data() const |
| 1150 | { |
| 1151 | if (d->nextReturnedEndOfData) { |
| 1152 | QXmlInputSource *that = const_cast<QXmlInputSource*>(this); |
| 1153 | that->d->nextReturnedEndOfData = false; |
| 1154 | that->fetchData(); |
| 1155 | } |
| 1156 | return d->str; |
| 1157 | } |
| 1158 | |
| 1159 | /*! |
| 1160 | Sets the data of the input source to \a dat. |
| 1161 | |
| 1162 | If the input source already contains data, this function deletes |
| 1163 | that data first. |
| 1164 | |
| 1165 | \sa data() |
| 1166 | */ |
| 1167 | void QXmlInputSource::setData(const QString& dat) |
| 1168 | { |
| 1169 | d->str = dat; |
| 1170 | d->unicode = dat.unicode(); |
| 1171 | d->pos = 0; |
| 1172 | d->length = d->str.size(); |
| 1173 | d->nextReturnedEndOfData = false; |
| 1174 | } |
| 1175 | |
| 1176 | /*! |
| 1177 | \overload |
| 1178 | |
| 1179 | The data \a dat is passed through the correct text-codec, before |
| 1180 | it is set. |
| 1181 | */ |
| 1182 | void QXmlInputSource::setData(const QByteArray& dat) |
| 1183 | { |
| 1184 | setData(fromRawData(data: dat)); |
| 1185 | } |
| 1186 | |
| 1187 | /*! |
| 1188 | This function reads more data from the device that was set during |
| 1189 | construction. If the input source already contained data, this |
| 1190 | function deletes that data first. |
| 1191 | |
| 1192 | This object contains no data after a call to this function if the |
| 1193 | object was constructed without a device to read data from or if |
| 1194 | this function was not able to get more data from the device. |
| 1195 | |
| 1196 | There are two occasions where a fetch is done implicitly by |
| 1197 | another function call: during construction (so that the object |
| 1198 | starts out with some initial data where available), and during a |
| 1199 | call to next() (if the data had run out). |
| 1200 | |
| 1201 | You don't normally need to use this function if you use next(). |
| 1202 | |
| 1203 | \sa data(), next(), QXmlInputSource() |
| 1204 | */ |
| 1205 | |
| 1206 | void QXmlInputSource::fetchData() |
| 1207 | { |
| 1208 | enum |
| 1209 | { |
| 1210 | BufferSize = 1024 |
| 1211 | }; |
| 1212 | |
| 1213 | QByteArray rawData; |
| 1214 | |
| 1215 | if (d->inputDevice || d->inputStream) { |
| 1216 | QIODevice *device = d->inputDevice ? d->inputDevice : d->inputStream->device(); |
| 1217 | |
| 1218 | if (!device) { |
| 1219 | if (d->inputStream && d->inputStream->string()) { |
| 1220 | QString *s = d->inputStream->string(); |
| 1221 | rawData = QByteArray(reinterpret_cast<const char *>(s->constData()), |
| 1222 | int(s->size() * sizeof(QChar))); |
| 1223 | } |
| 1224 | } else if (device->isOpen() || device->open(mode: QIODevice::ReadOnly)) { |
| 1225 | rawData.resize(size: BufferSize); |
| 1226 | qint64 size = device->read(data: rawData.data(), maxlen: BufferSize); |
| 1227 | if (size == 0 && device->waitForReadyRead(msecs: -1)) |
| 1228 | size = device->read(data: rawData.data(), maxlen: BufferSize); |
| 1229 | |
| 1230 | rawData.resize(size: qMax(a: qint64(0), b: size)); |
| 1231 | } |
| 1232 | |
| 1233 | /* We do this inside the "if (d->inputDevice ..." scope |
| 1234 | * because if we're not using a stream or device, that is, |
| 1235 | * the user set a QString manually, we don't want to set |
| 1236 | * d->str. */ |
| 1237 | setData(fromRawData(data: rawData)); |
| 1238 | } |
| 1239 | } |
| 1240 | |
| 1241 | #if QT_CONFIG(textcodec) |
| 1242 | static QString (const QString &text, bool *needMoreText) |
| 1243 | { |
| 1244 | *needMoreText = false; |
| 1245 | |
| 1246 | int l = text.size(); |
| 1247 | const QLatin1String snip("<?xml" , std::min(a: l, b: 5)); |
| 1248 | if (l > 0 && !text.startsWith(s: snip)) |
| 1249 | return QString(); |
| 1250 | |
| 1251 | int endPos = text.indexOf(ch: QLatin1Char('>')); |
| 1252 | if (endPos == -1) { |
| 1253 | *needMoreText = l < 255; // we won't look forever |
| 1254 | return QString(); |
| 1255 | } |
| 1256 | |
| 1257 | int pos = text.indexOf(s: QLatin1String("encoding" )); |
| 1258 | if (pos == -1 || pos >= endPos) |
| 1259 | return QString(); |
| 1260 | |
| 1261 | while (pos < endPos) { |
| 1262 | QChar uc = text.at(i: pos); |
| 1263 | if (uc == u'\'' || uc == u'"') |
| 1264 | break; |
| 1265 | ++pos; |
| 1266 | } |
| 1267 | |
| 1268 | if (pos == endPos) |
| 1269 | return QString(); |
| 1270 | |
| 1271 | QString encoding; |
| 1272 | ++pos; |
| 1273 | while (pos < endPos) { |
| 1274 | QChar uc = text.at(i: pos); |
| 1275 | if (uc == u'\'' || uc == u'"') |
| 1276 | break; |
| 1277 | encoding.append(c: uc); |
| 1278 | ++pos; |
| 1279 | } |
| 1280 | |
| 1281 | return encoding; |
| 1282 | } |
| 1283 | #endif // textcodec |
| 1284 | |
| 1285 | /*! |
| 1286 | This function reads the XML file from \a data and tries to |
| 1287 | recognize the encoding. It converts the raw data \a data into a |
| 1288 | QString and returns it. It tries its best to get the correct |
| 1289 | encoding for the XML file. |
| 1290 | |
| 1291 | If \a beginning is true, this function assumes that the data |
| 1292 | starts at the beginning of a new XML document and looks for an |
| 1293 | encoding declaration. If \a beginning is false, it converts the |
| 1294 | raw data using the encoding determined from prior calls. |
| 1295 | */ |
| 1296 | QString QXmlInputSource::fromRawData(const QByteArray &data, bool beginning) |
| 1297 | { |
| 1298 | #if !QT_CONFIG(textcodec) |
| 1299 | Q_UNUSED(beginning); |
| 1300 | return QString::fromLatin1(data.constData(), data.size()); |
| 1301 | #else |
| 1302 | if (data.size() == 0) |
| 1303 | return QString(); |
| 1304 | if (beginning) { |
| 1305 | delete d->encMapper; |
| 1306 | d->encMapper = nullptr; |
| 1307 | } |
| 1308 | |
| 1309 | int mib = 106; // UTF-8 |
| 1310 | |
| 1311 | // This is the initial UTF codec we will read the encoding declaration with |
| 1312 | if (d->encMapper == nullptr) { |
| 1313 | d->encodingDeclBytes.clear(); |
| 1314 | d->encodingDeclChars.clear(); |
| 1315 | d->lookingForEncodingDecl = true; |
| 1316 | |
| 1317 | // look for byte order mark and read the first 5 characters |
| 1318 | if (data.size() >= 4) { |
| 1319 | uchar ch1 = data.at(i: 0); |
| 1320 | uchar ch2 = data.at(i: 1); |
| 1321 | uchar ch3 = data.at(i: 2); |
| 1322 | uchar ch4 = data.at(i: 3); |
| 1323 | |
| 1324 | if ((ch1 == 0 && ch2 == 0 && ch3 == 0xfe && ch4 == 0xff) || |
| 1325 | (ch1 == 0xff && ch2 == 0xfe && ch3 == 0 && ch4 == 0)) |
| 1326 | mib = 1017; // UTF-32 with byte order mark |
| 1327 | else if (ch1 == 0x3c && ch2 == 0x00 && ch3 == 0x00 && ch4 == 0x00) |
| 1328 | mib = 1019; // UTF-32LE |
| 1329 | else if (ch1 == 0x00 && ch2 == 0x00 && ch3 == 0x00 && ch4 == 0x3c) |
| 1330 | mib = 1018; // UTF-32BE |
| 1331 | } |
| 1332 | if (mib == 106 && data.size() >= 2) { |
| 1333 | uchar ch1 = data.at(i: 0); |
| 1334 | uchar ch2 = data.at(i: 1); |
| 1335 | |
| 1336 | if ((ch1 == 0xfe && ch2 == 0xff) || (ch1 == 0xff && ch2 == 0xfe)) |
| 1337 | mib = 1015; // UTF-16 with byte order mark |
| 1338 | else if (ch1 == 0x3c && ch2 == 0x00) |
| 1339 | mib = 1014; // UTF-16LE |
| 1340 | else if (ch1 == 0x00 && ch2 == 0x3c) |
| 1341 | mib = 1013; // UTF-16BE |
| 1342 | } |
| 1343 | |
| 1344 | QTextCodec *codec = QTextCodec::codecForMib(mib); |
| 1345 | Q_ASSERT(codec); |
| 1346 | |
| 1347 | d->encMapper = codec->makeDecoder(); |
| 1348 | } |
| 1349 | |
| 1350 | QString input = d->encMapper->toUnicode(chars: data.constData(), len: data.size()); |
| 1351 | |
| 1352 | if (d->lookingForEncodingDecl) { |
| 1353 | d->encodingDeclChars += input; |
| 1354 | |
| 1355 | bool needMoreText; |
| 1356 | QString encoding = extractEncodingDecl(text: d->encodingDeclChars, needMoreText: &needMoreText); |
| 1357 | |
| 1358 | if (!encoding.isEmpty()) { |
| 1359 | if (QTextCodec *codec = QTextCodec::codecForName(name: std::move(encoding).toLatin1())) { |
| 1360 | /* If the encoding is the same, we don't have to do toUnicode() all over again. */ |
| 1361 | if(codec->mibEnum() != mib) { |
| 1362 | delete d->encMapper; |
| 1363 | d->encMapper = codec->makeDecoder(); |
| 1364 | |
| 1365 | /* The variable input can potentially be large, so we deallocate |
| 1366 | * it before calling toUnicode() in order to avoid having two |
| 1367 | * large QStrings in memory simultaneously. */ |
| 1368 | input.clear(); |
| 1369 | |
| 1370 | // prime the decoder with the data so far |
| 1371 | d->encMapper->toUnicode(chars: d->encodingDeclBytes.constData(), len: d->encodingDeclBytes.size()); |
| 1372 | // now feed it the new data |
| 1373 | input = d->encMapper->toUnicode(chars: data.constData(), len: data.size()); |
| 1374 | } |
| 1375 | } |
| 1376 | } |
| 1377 | |
| 1378 | d->encodingDeclBytes += data; |
| 1379 | d->lookingForEncodingDecl = needMoreText; |
| 1380 | } |
| 1381 | |
| 1382 | return input; |
| 1383 | #endif |
| 1384 | } |
| 1385 | |
| 1386 | /********************************************* |
| 1387 | * |
| 1388 | * QXmlDefaultHandler |
| 1389 | * |
| 1390 | *********************************************/ |
| 1391 | |
| 1392 | /*! |
| 1393 | \class QXmlContentHandler |
| 1394 | \reentrant |
| 1395 | \brief The QXmlContentHandler class provides an interface to |
| 1396 | report the logical content of XML data. |
| 1397 | |
| 1398 | \inmodule QtCore5Compat |
| 1399 | \ingroup xml-tools |
| 1400 | |
| 1401 | If the application needs to be informed of basic parsing events, |
| 1402 | it can implement this interface and activate it using |
| 1403 | QXmlReader::setContentHandler(). The reader can then report basic |
| 1404 | document-related events like the start and end of elements and |
| 1405 | character data through this interface. |
| 1406 | |
| 1407 | The order of events in this interface is very important, and |
| 1408 | mirrors the order of information in the document itself. For |
| 1409 | example, all of an element's content (character data, processing |
| 1410 | instructions, and sub-elements) appears, in order, between the |
| 1411 | startElement() event and the corresponding endElement() event. |
| 1412 | |
| 1413 | The class QXmlDefaultHandler provides a default implementation for |
| 1414 | this interface; subclassing from the QXmlDefaultHandler class is |
| 1415 | very convenient if you only want to be informed of some parsing |
| 1416 | events. |
| 1417 | |
| 1418 | The startDocument() function is called at the start of the |
| 1419 | document, and endDocument() is called at the end. Before parsing |
| 1420 | begins setDocumentLocator() is called. For each element |
| 1421 | startElement() is called, with endElement() being called at the |
| 1422 | end of each element. The characters() function is called with |
| 1423 | chunks of character data; ignorableWhitespace() is called with |
| 1424 | chunks of whitespace and processingInstruction() is called with |
| 1425 | processing instructions. If an entity is skipped skippedEntity() |
| 1426 | is called. At the beginning of prefix-URI scopes |
| 1427 | startPrefixMapping() is called. |
| 1428 | |
| 1429 | \sa QXmlDTDHandler, QXmlDeclHandler, QXmlEntityResolver, QXmlErrorHandler, |
| 1430 | QXmlLexicalHandler |
| 1431 | */ |
| 1432 | |
| 1433 | /*! |
| 1434 | \fn QXmlContentHandler::~QXmlContentHandler() |
| 1435 | |
| 1436 | Destroys the content handler. |
| 1437 | */ |
| 1438 | QXmlContentHandler::~QXmlContentHandler() |
| 1439 | = default; |
| 1440 | |
| 1441 | /*! |
| 1442 | \fn void QXmlContentHandler::setDocumentLocator(QXmlLocator* locator) |
| 1443 | |
| 1444 | The reader calls this function before it starts parsing the |
| 1445 | document. The argument \a locator is a pointer to a QXmlLocator |
| 1446 | which allows the application to get the parsing position within |
| 1447 | the document. |
| 1448 | |
| 1449 | Do not destroy the \a locator; it is destroyed when the reader is |
| 1450 | destroyed. (Do not use the \a locator after the reader is |
| 1451 | destroyed). |
| 1452 | */ |
| 1453 | |
| 1454 | /*! |
| 1455 | \fn bool QXmlContentHandler::startDocument() |
| 1456 | |
| 1457 | The reader calls this function when it starts parsing the |
| 1458 | document. The reader calls this function just once, after the call |
| 1459 | to setDocumentLocator(), and before any other functions in this |
| 1460 | class or in the QXmlDTDHandler class are called. |
| 1461 | |
| 1462 | If this function returns \c false the reader stops parsing and |
| 1463 | reports an error. The reader uses the function errorString() to |
| 1464 | get the error message. |
| 1465 | |
| 1466 | \sa endDocument() |
| 1467 | */ |
| 1468 | |
| 1469 | /*! |
| 1470 | \fn bool QXmlContentHandler::endDocument() |
| 1471 | |
| 1472 | The reader calls this function after it has finished parsing. It |
| 1473 | is called just once, and is the last handler function called. It |
| 1474 | is called after the reader has read all input or has abandoned |
| 1475 | parsing because of a fatal error. |
| 1476 | |
| 1477 | If this function returns \c false the reader stops parsing and |
| 1478 | reports an error. The reader uses the function errorString() to |
| 1479 | get the error message. |
| 1480 | |
| 1481 | \sa startDocument() |
| 1482 | */ |
| 1483 | |
| 1484 | /*! |
| 1485 | \fn bool QXmlContentHandler::startPrefixMapping(const QString& prefix, const QString& uri) |
| 1486 | |
| 1487 | The reader calls this function to signal the begin of a prefix-URI |
| 1488 | namespace mapping scope. This information is not necessary for |
| 1489 | normal namespace processing since the reader automatically |
| 1490 | replaces prefixes for element and attribute names. |
| 1491 | |
| 1492 | Note that startPrefixMapping() and endPrefixMapping() calls are |
| 1493 | not guaranteed to be properly nested relative to each other: all |
| 1494 | startPrefixMapping() events occur before the corresponding |
| 1495 | startElement() event, and all endPrefixMapping() events occur |
| 1496 | after the corresponding endElement() event, but their order is not |
| 1497 | otherwise guaranteed. |
| 1498 | |
| 1499 | The argument \a prefix is the namespace prefix being declared and |
| 1500 | the argument \a uri is the namespace URI the prefix is mapped to. |
| 1501 | |
| 1502 | If this function returns \c false the reader stops parsing and |
| 1503 | reports an error. The reader uses the function errorString() to |
| 1504 | get the error message. |
| 1505 | |
| 1506 | \sa endPrefixMapping() |
| 1507 | */ |
| 1508 | |
| 1509 | /*! |
| 1510 | \fn bool QXmlContentHandler::endPrefixMapping(const QString& prefix) |
| 1511 | |
| 1512 | The reader calls this function to signal the end of a prefix |
| 1513 | mapping for the prefix \a prefix. |
| 1514 | |
| 1515 | If this function returns \c false the reader stops parsing and |
| 1516 | reports an error. The reader uses the function errorString() to |
| 1517 | get the error message. |
| 1518 | |
| 1519 | \sa startPrefixMapping() |
| 1520 | */ |
| 1521 | |
| 1522 | /*! |
| 1523 | \fn bool QXmlContentHandler::startElement(const QString& namespaceURI, const QString& localName, const QString& qName, const QXmlAttributes& atts) |
| 1524 | |
| 1525 | The reader calls this function when it has parsed a start element |
| 1526 | tag. |
| 1527 | |
| 1528 | There is a corresponding endElement() call when the corresponding |
| 1529 | end element tag is read. The startElement() and endElement() calls |
| 1530 | are always nested correctly. Empty element tags (e.g. \c{<x/>}) |
| 1531 | cause a startElement() call to be immediately followed by an |
| 1532 | endElement() call. |
| 1533 | |
| 1534 | The attribute list provided only contains attributes with explicit |
| 1535 | values. The attribute list contains attributes used for namespace |
| 1536 | declaration (i.e. attributes starting with xmlns) only if the |
| 1537 | namespace-prefix property of the reader is true. |
| 1538 | |
| 1539 | The argument \a namespaceURI is the namespace URI, or |
| 1540 | an empty string if the element has no namespace URI or if no |
| 1541 | namespace processing is done. \a localName is the local name |
| 1542 | (without prefix), or an empty string if no namespace processing is |
| 1543 | done, \a qName is the qualified name (with prefix) and \a atts are |
| 1544 | the attributes attached to the element. If there are no |
| 1545 | attributes, \a atts is an empty attributes object. |
| 1546 | |
| 1547 | If this function returns \c false the reader stops parsing and |
| 1548 | reports an error. The reader uses the function errorString() to |
| 1549 | get the error message. |
| 1550 | |
| 1551 | \sa endElement() |
| 1552 | */ |
| 1553 | |
| 1554 | /*! |
| 1555 | \fn bool QXmlContentHandler::endElement(const QString& namespaceURI, const QString& localName, const QString& qName) |
| 1556 | |
| 1557 | The reader calls this function when it has parsed an end element |
| 1558 | tag with the qualified name \a qName, the local name \a localName |
| 1559 | and the namespace URI \a namespaceURI. |
| 1560 | |
| 1561 | If this function returns \c false the reader stops parsing and |
| 1562 | reports an error. The reader uses the function errorString() to |
| 1563 | get the error message. |
| 1564 | |
| 1565 | \sa startElement() |
| 1566 | */ |
| 1567 | |
| 1568 | /*! |
| 1569 | \fn bool QXmlContentHandler::characters(const QString& ch) |
| 1570 | |
| 1571 | The reader calls this function when it has parsed a chunk of |
| 1572 | character data (either normal character data or character data |
| 1573 | inside a CDATA section; if you need to distinguish between those |
| 1574 | two types you must use QXmlLexicalHandler::startCDATA() and |
| 1575 | QXmlLexicalHandler::endCDATA()). The character data is reported in |
| 1576 | \a ch. |
| 1577 | |
| 1578 | Some readers report whitespace in element content using the |
| 1579 | ignorableWhitespace() function rather than using this one. |
| 1580 | |
| 1581 | A reader may report the character data of an element in more than |
| 1582 | one chunk; e.g. a reader might want to report "a\<b" in three |
| 1583 | characters() events ("a ", "\<" and " b"). |
| 1584 | |
| 1585 | If this function returns \c false the reader stops parsing and |
| 1586 | reports an error. The reader uses the function errorString() to |
| 1587 | get the error message. |
| 1588 | */ |
| 1589 | |
| 1590 | /*! |
| 1591 | \fn bool QXmlContentHandler::ignorableWhitespace(const QString& ch) |
| 1592 | |
| 1593 | Some readers may use this function to report each chunk of |
| 1594 | whitespace in element content. The whitespace is reported in \a ch. |
| 1595 | |
| 1596 | If this function returns \c false the reader stops parsing and |
| 1597 | reports an error. The reader uses the function errorString() to |
| 1598 | get the error message. |
| 1599 | */ |
| 1600 | |
| 1601 | /*! |
| 1602 | \fn bool QXmlContentHandler::processingInstruction(const QString& target, const QString& data) |
| 1603 | |
| 1604 | The reader calls this function when it has parsed a processing |
| 1605 | instruction. |
| 1606 | |
| 1607 | \a target is the target name of the processing instruction and \a |
| 1608 | data is the data in the processing instruction. |
| 1609 | |
| 1610 | If this function returns \c false the reader stops parsing and |
| 1611 | reports an error. The reader uses the function errorString() to |
| 1612 | get the error message. |
| 1613 | */ |
| 1614 | |
| 1615 | /*! |
| 1616 | \fn bool QXmlContentHandler::skippedEntity(const QString& name) |
| 1617 | |
| 1618 | Some readers may skip entities if they have not seen the |
| 1619 | declarations (e.g. because they are in an external DTD). If they |
| 1620 | do so they report that they skipped the entity called \a name by |
| 1621 | calling this function. |
| 1622 | |
| 1623 | If this function returns \c false the reader stops parsing and |
| 1624 | reports an error. The reader uses the function errorString() to |
| 1625 | get the error message. |
| 1626 | */ |
| 1627 | |
| 1628 | /*! |
| 1629 | \fn QString QXmlContentHandler::errorString() const |
| 1630 | |
| 1631 | The reader calls this function to get an error string, e.g. if any |
| 1632 | of the handler functions returns \c false. |
| 1633 | */ |
| 1634 | |
| 1635 | /*! |
| 1636 | \class QXmlErrorHandler |
| 1637 | \reentrant |
| 1638 | \brief The QXmlErrorHandler class provides an interface to report |
| 1639 | errors in XML data. |
| 1640 | |
| 1641 | \inmodule QtCore5Compat |
| 1642 | \ingroup xml-tools |
| 1643 | |
| 1644 | If you want your application to report errors to the user or to |
| 1645 | perform customized error handling, you should subclass this class. |
| 1646 | |
| 1647 | You can set the error handler with QXmlReader::setErrorHandler(). |
| 1648 | |
| 1649 | Errors can be reported using warning(), error() and fatalError(), |
| 1650 | with the error text being reported with errorString(). |
| 1651 | |
| 1652 | \sa QXmlDTDHandler, QXmlDeclHandler, QXmlContentHandler, QXmlEntityResolver, |
| 1653 | QXmlLexicalHandler |
| 1654 | */ |
| 1655 | |
| 1656 | /*! |
| 1657 | \fn QXmlErrorHandler::~QXmlErrorHandler() |
| 1658 | |
| 1659 | Destroys the error handler. |
| 1660 | */ |
| 1661 | QXmlErrorHandler::~QXmlErrorHandler() |
| 1662 | = default; |
| 1663 | |
| 1664 | /*! |
| 1665 | \fn bool QXmlErrorHandler::warning(const QXmlParseException& exception) |
| 1666 | |
| 1667 | A reader might use this function to report a warning. Warnings are |
| 1668 | conditions that are not errors or fatal errors as defined by the |
| 1669 | XML 1.0 specification. Details of the warning are stored in \a |
| 1670 | exception. |
| 1671 | |
| 1672 | If this function returns \c false the reader stops parsing and |
| 1673 | reports an error. The reader uses the function errorString() to |
| 1674 | get the error message. |
| 1675 | */ |
| 1676 | |
| 1677 | /*! |
| 1678 | \fn bool QXmlErrorHandler::error(const QXmlParseException& exception) |
| 1679 | |
| 1680 | A reader might use this function to report a recoverable error. A |
| 1681 | recoverable error corresponds to the definiton of "error" in |
| 1682 | section 1.2 of the XML 1.0 specification. Details of the error are |
| 1683 | stored in \a exception. |
| 1684 | |
| 1685 | The reader must continue to provide normal parsing events after |
| 1686 | invoking this function. |
| 1687 | |
| 1688 | If this function returns \c false the reader stops parsing and |
| 1689 | reports an error. The reader uses the function errorString() to |
| 1690 | get the error message. |
| 1691 | */ |
| 1692 | |
| 1693 | /*! |
| 1694 | \fn bool QXmlErrorHandler::fatalError(const QXmlParseException& exception) |
| 1695 | |
| 1696 | A reader must use this function to report a non-recoverable error. |
| 1697 | Details of the error are stored in \a exception. |
| 1698 | |
| 1699 | If this function returns \c true the reader might try to go on |
| 1700 | parsing and reporting further errors, but no regular parsing |
| 1701 | events are reported. |
| 1702 | */ |
| 1703 | |
| 1704 | /*! |
| 1705 | \fn QString QXmlErrorHandler::errorString() const |
| 1706 | |
| 1707 | The reader calls this function to get an error string if any of |
| 1708 | the handler functions returns \c false. |
| 1709 | */ |
| 1710 | |
| 1711 | /*! |
| 1712 | \class QXmlDTDHandler |
| 1713 | \reentrant |
| 1714 | \brief The QXmlDTDHandler class provides an interface to report |
| 1715 | DTD content of XML data. |
| 1716 | |
| 1717 | \inmodule QtCore5Compat |
| 1718 | \ingroup xml-tools |
| 1719 | |
| 1720 | If an application needs information about notations and unparsed |
| 1721 | entities, it can implement this interface and register an instance |
| 1722 | with QXmlReader::setDTDHandler(). |
| 1723 | |
| 1724 | Note that this interface includes only those DTD events that the |
| 1725 | XML recommendation requires processors to report, i.e. notation |
| 1726 | and unparsed entity declarations using notationDecl() and |
| 1727 | unparsedEntityDecl() respectively. |
| 1728 | |
| 1729 | \sa QXmlDeclHandler, QXmlContentHandler, QXmlEntityResolver, QXmlErrorHandler, |
| 1730 | QXmlLexicalHandler |
| 1731 | */ |
| 1732 | |
| 1733 | /*! |
| 1734 | \fn QXmlDTDHandler::~QXmlDTDHandler() |
| 1735 | |
| 1736 | Destroys the DTD handler. |
| 1737 | */ |
| 1738 | QXmlDTDHandler::~QXmlDTDHandler() |
| 1739 | = default; |
| 1740 | |
| 1741 | /*! |
| 1742 | \fn bool QXmlDTDHandler::notationDecl(const QString& name, const QString& publicId, const QString& systemId) |
| 1743 | |
| 1744 | The reader calls this function when it has parsed a notation |
| 1745 | declaration. |
| 1746 | |
| 1747 | The argument \a name is the notation name, \a publicId is the |
| 1748 | notation's public identifier and \a systemId is the notation's |
| 1749 | system identifier. |
| 1750 | |
| 1751 | If this function returns \c false the reader stops parsing and |
| 1752 | reports an error. The reader uses the function errorString() to |
| 1753 | get the error message. |
| 1754 | */ |
| 1755 | |
| 1756 | /*! |
| 1757 | \fn bool QXmlDTDHandler::unparsedEntityDecl(const QString& name, const QString& publicId, const QString& systemId, const QString& notationName) |
| 1758 | |
| 1759 | The reader calls this function when it finds an unparsed entity |
| 1760 | declaration. |
| 1761 | |
| 1762 | The argument \a name is the unparsed entity's name, \a publicId is |
| 1763 | the entity's public identifier, \a systemId is the entity's system |
| 1764 | identifier and \a notationName is the name of the associated |
| 1765 | notation. |
| 1766 | |
| 1767 | If this function returns \c false the reader stops parsing and |
| 1768 | reports an error. The reader uses the function errorString() to |
| 1769 | get the error message. |
| 1770 | */ |
| 1771 | |
| 1772 | /*! |
| 1773 | \fn QString QXmlDTDHandler::errorString() const |
| 1774 | |
| 1775 | The reader calls this function to get an error string if any of |
| 1776 | the handler functions returns \c false. |
| 1777 | */ |
| 1778 | |
| 1779 | /*! |
| 1780 | \class QXmlEntityResolver |
| 1781 | \reentrant |
| 1782 | \brief The QXmlEntityResolver class provides an interface to |
| 1783 | resolve external entities contained in XML data. |
| 1784 | |
| 1785 | \inmodule QtCore5Compat |
| 1786 | \ingroup xml-tools |
| 1787 | |
| 1788 | If an application needs to implement customized handling for |
| 1789 | external entities, it must implement this interface, i.e. |
| 1790 | resolveEntity(), and register it with |
| 1791 | QXmlReader::setEntityResolver(). |
| 1792 | |
| 1793 | \sa QXmlDTDHandler, QXmlDeclHandler, QXmlContentHandler, QXmlErrorHandler, |
| 1794 | QXmlLexicalHandler |
| 1795 | */ |
| 1796 | |
| 1797 | /*! |
| 1798 | \fn QXmlEntityResolver::~QXmlEntityResolver() |
| 1799 | |
| 1800 | Destroys the entity resolver. |
| 1801 | */ |
| 1802 | QXmlEntityResolver::~QXmlEntityResolver() |
| 1803 | = default; |
| 1804 | |
| 1805 | /*! |
| 1806 | \fn bool QXmlEntityResolver::resolveEntity(const QString& publicId, const QString& systemId, QXmlInputSource*& ret) |
| 1807 | |
| 1808 | The reader calls this function before it opens any external |
| 1809 | entity, except the top-level document entity. The application may |
| 1810 | request the reader to resolve the entity itself (\a ret is 0) or |
| 1811 | to use an entirely different input source (\a ret points to the |
| 1812 | input source). |
| 1813 | |
| 1814 | The reader deletes the input source \a ret when it no longer needs |
| 1815 | it, so you should allocate it on the heap with \c new. |
| 1816 | |
| 1817 | The argument \a publicId is the public identifier of the external |
| 1818 | entity, \a systemId is the system identifier of the external |
| 1819 | entity and \a ret is the return value of this function. If \a ret |
| 1820 | is 0 the reader should resolve the entity itself, if it is |
| 1821 | non-zero it must point to an input source which the reader uses |
| 1822 | instead. |
| 1823 | |
| 1824 | If this function returns \c false the reader stops parsing and |
| 1825 | reports an error. The reader uses the function errorString() to |
| 1826 | get the error message. |
| 1827 | */ |
| 1828 | |
| 1829 | /*! |
| 1830 | \fn QString QXmlEntityResolver::errorString() const |
| 1831 | |
| 1832 | The reader calls this function to get an error string if any of |
| 1833 | the handler functions returns \c false. |
| 1834 | */ |
| 1835 | |
| 1836 | /*! |
| 1837 | \class QXmlLexicalHandler |
| 1838 | \reentrant |
| 1839 | \brief The QXmlLexicalHandler class provides an interface to |
| 1840 | report the lexical content of XML data. |
| 1841 | |
| 1842 | \inmodule QtCore5Compat |
| 1843 | \ingroup xml-tools |
| 1844 | |
| 1845 | The events in the lexical handler apply to the entire document, |
| 1846 | not just to the document element, and all lexical handler events |
| 1847 | appear between the content handler's startDocument and endDocument |
| 1848 | events. |
| 1849 | |
| 1850 | You can set the lexical handler with |
| 1851 | QXmlReader::setLexicalHandler(). |
| 1852 | |
| 1853 | This interface's design is based on the SAX2 extension |
| 1854 | LexicalHandler. |
| 1855 | |
| 1856 | The interface provides the startDTD(), endDTD(), startEntity(), |
| 1857 | endEntity(), startCDATA(), endCDATA() and comment() functions. |
| 1858 | |
| 1859 | \sa QXmlDTDHandler, QXmlDeclHandler, QXmlContentHandler, QXmlEntityResolver, |
| 1860 | QXmlErrorHandler |
| 1861 | */ |
| 1862 | |
| 1863 | /*! |
| 1864 | \fn QXmlLexicalHandler::~QXmlLexicalHandler() |
| 1865 | |
| 1866 | Destroys the lexical handler. |
| 1867 | */ |
| 1868 | QXmlLexicalHandler::~QXmlLexicalHandler() |
| 1869 | = default; |
| 1870 | |
| 1871 | /*! |
| 1872 | \fn bool QXmlLexicalHandler::startDTD(const QString& name, const QString& publicId, const QString& systemId) |
| 1873 | |
| 1874 | The reader calls this function to report the start of a DTD |
| 1875 | declaration, if any. It reports the name of the document type in |
| 1876 | \a name, the public identifier in \a publicId and the system |
| 1877 | identifier in \a systemId. |
| 1878 | |
| 1879 | If the public identifier is missing, \a publicId is set to |
| 1880 | an empty string. If the system identifier is missing, \a systemId is |
| 1881 | set to an empty string. Note that it is not valid XML to have a |
| 1882 | public identifier but no system identifier; in such cases a parse |
| 1883 | error will occur. |
| 1884 | |
| 1885 | All declarations reported through QXmlDTDHandler or |
| 1886 | QXmlDeclHandler appear between the startDTD() and endDTD() calls. |
| 1887 | |
| 1888 | If this function returns \c false the reader stops parsing and |
| 1889 | reports an error. The reader uses the function errorString() to |
| 1890 | get the error message. |
| 1891 | |
| 1892 | \sa endDTD() |
| 1893 | */ |
| 1894 | |
| 1895 | /*! |
| 1896 | \fn bool QXmlLexicalHandler::endDTD() |
| 1897 | |
| 1898 | The reader calls this function to report the end of a DTD |
| 1899 | declaration, if any. |
| 1900 | |
| 1901 | If this function returns \c false the reader stops parsing and |
| 1902 | reports an error. The reader uses the function errorString() to |
| 1903 | get the error message. |
| 1904 | |
| 1905 | \sa startDTD() |
| 1906 | */ |
| 1907 | |
| 1908 | /*! |
| 1909 | \fn bool QXmlLexicalHandler::startEntity(const QString& name) |
| 1910 | |
| 1911 | The reader calls this function to report the start of an entity |
| 1912 | called \a name. |
| 1913 | |
| 1914 | Note that if the entity is unknown, the reader reports it through |
| 1915 | QXmlContentHandler::skippedEntity() and not through this |
| 1916 | function. |
| 1917 | |
| 1918 | If this function returns \c false the reader stops parsing and |
| 1919 | reports an error. The reader uses the function errorString() to |
| 1920 | get the error message. |
| 1921 | |
| 1922 | \sa endEntity(), QXmlSimpleReader::setFeature() |
| 1923 | */ |
| 1924 | |
| 1925 | /*! |
| 1926 | \fn bool QXmlLexicalHandler::endEntity(const QString& name) |
| 1927 | |
| 1928 | The reader calls this function to report the end of an entity |
| 1929 | called \a name. |
| 1930 | |
| 1931 | For every startEntity() call, there is a corresponding endEntity() |
| 1932 | call. The calls to startEntity() and endEntity() are properly |
| 1933 | nested. |
| 1934 | |
| 1935 | If this function returns \c false the reader stops parsing and |
| 1936 | reports an error. The reader uses the function errorString() to |
| 1937 | get the error message. |
| 1938 | |
| 1939 | \sa startEntity(), QXmlContentHandler::skippedEntity(), QXmlSimpleReader::setFeature() |
| 1940 | */ |
| 1941 | |
| 1942 | /*! |
| 1943 | \fn bool QXmlLexicalHandler::startCDATA() |
| 1944 | |
| 1945 | The reader calls this function to report the start of a CDATA |
| 1946 | section. The content of the CDATA section is reported through the |
| 1947 | QXmlContentHandler::characters() function. This function is |
| 1948 | intended only to report the boundary. |
| 1949 | |
| 1950 | If this function returns \c false the reader stops parsing and |
| 1951 | reports an error. The reader uses the function errorString() to |
| 1952 | get the error message. |
| 1953 | |
| 1954 | \sa endCDATA() |
| 1955 | */ |
| 1956 | |
| 1957 | /*! |
| 1958 | \fn bool QXmlLexicalHandler::endCDATA() |
| 1959 | |
| 1960 | The reader calls this function to report the end of a CDATA |
| 1961 | section. |
| 1962 | |
| 1963 | If this function returns \c false the reader stops parsing and reports |
| 1964 | an error. The reader uses the function errorString() to get the error |
| 1965 | message. |
| 1966 | |
| 1967 | \sa startCDATA(), QXmlContentHandler::characters() |
| 1968 | */ |
| 1969 | |
| 1970 | /*! |
| 1971 | \fn bool QXmlLexicalHandler::comment(const QString& ch) |
| 1972 | |
| 1973 | The reader calls this function to report an XML comment anywhere |
| 1974 | in the document. It reports the text of the comment in \a ch. |
| 1975 | |
| 1976 | If this function returns \c false the reader stops parsing and |
| 1977 | reports an error. The reader uses the function errorString() to |
| 1978 | get the error message. |
| 1979 | */ |
| 1980 | |
| 1981 | /*! |
| 1982 | \fn QString QXmlLexicalHandler::errorString() const |
| 1983 | |
| 1984 | The reader calls this function to get an error string if any of |
| 1985 | the handler functions returns \c false. |
| 1986 | */ |
| 1987 | |
| 1988 | /*! |
| 1989 | \class QXmlDeclHandler |
| 1990 | \reentrant |
| 1991 | \brief The QXmlDeclHandler class provides an interface to report declaration |
| 1992 | content of XML data. |
| 1993 | |
| 1994 | \inmodule QtCore5Compat |
| 1995 | \ingroup xml-tools |
| 1996 | |
| 1997 | You can set the declaration handler with |
| 1998 | QXmlReader::setDeclHandler(). |
| 1999 | |
| 2000 | This interface is based on the SAX2 extension DeclHandler. |
| 2001 | |
| 2002 | The interface provides attributeDecl(), internalEntityDecl() and |
| 2003 | externalEntityDecl() functions. |
| 2004 | |
| 2005 | \sa QXmlDTDHandler, QXmlContentHandler, QXmlEntityResolver, QXmlErrorHandler, |
| 2006 | QXmlLexicalHandler |
| 2007 | */ |
| 2008 | |
| 2009 | /*! |
| 2010 | \fn QXmlDeclHandler::~QXmlDeclHandler() |
| 2011 | |
| 2012 | Destroys the declaration handler. |
| 2013 | */ |
| 2014 | QXmlDeclHandler::~QXmlDeclHandler() |
| 2015 | = default; |
| 2016 | |
| 2017 | /*! |
| 2018 | \fn bool QXmlDeclHandler::attributeDecl(const QString& eName, const QString& aName, const QString& type, const QString& valueDefault, const QString& value) |
| 2019 | |
| 2020 | The reader calls this function to report an attribute type |
| 2021 | declaration. Only the effective (first) declaration for an |
| 2022 | attribute is reported. |
| 2023 | |
| 2024 | The reader passes the name of the associated element in \a eName |
| 2025 | and the name of the attribute in \a aName. It passes a string that |
| 2026 | represents the attribute type in \a type and a string that |
| 2027 | represents the attribute default in \a valueDefault. This string |
| 2028 | is one of "#IMPLIED", "#REQUIRED", "#FIXED" or an empty string (if |
| 2029 | none of the others applies). The reader passes the attribute's |
| 2030 | default value in \a value. If no default value is specified in the |
| 2031 | XML file, \a value is an empty string. |
| 2032 | |
| 2033 | If this function returns \c false the reader stops parsing and |
| 2034 | reports an error. The reader uses the function errorString() to |
| 2035 | get the error message. |
| 2036 | */ |
| 2037 | |
| 2038 | /*! |
| 2039 | \fn bool QXmlDeclHandler::internalEntityDecl(const QString& name, const QString& value) |
| 2040 | |
| 2041 | The reader calls this function to report an internal entity |
| 2042 | declaration. Only the effective (first) declaration is reported. |
| 2043 | |
| 2044 | The reader passes the name of the entity in \a name and the value |
| 2045 | of the entity in \a value. |
| 2046 | |
| 2047 | If this function returns \c false the reader stops parsing and |
| 2048 | reports an error. The reader uses the function errorString() to |
| 2049 | get the error message. |
| 2050 | */ |
| 2051 | |
| 2052 | /*! |
| 2053 | \fn bool QXmlDeclHandler::externalEntityDecl(const QString& name, const QString& publicId, const QString& systemId) |
| 2054 | |
| 2055 | The reader calls this function to report a parsed external entity |
| 2056 | declaration. Only the effective (first) declaration for each |
| 2057 | entity is reported. |
| 2058 | |
| 2059 | The reader passes the name of the entity in \a name, the public |
| 2060 | identifier in \a publicId and the system identifier in \a |
| 2061 | systemId. If there is no public identifier specified, it passes |
| 2062 | an empty string in \a publicId. |
| 2063 | |
| 2064 | If this function returns \c false the reader stops parsing and |
| 2065 | reports an error. The reader uses the function errorString() to |
| 2066 | get the error message. |
| 2067 | */ |
| 2068 | |
| 2069 | /*! |
| 2070 | \fn QString QXmlDeclHandler::errorString() const |
| 2071 | |
| 2072 | The reader calls this function to get an error string if any of |
| 2073 | the handler functions returns \c false. |
| 2074 | */ |
| 2075 | |
| 2076 | /*! |
| 2077 | \class QXmlDefaultHandler |
| 2078 | \reentrant |
| 2079 | \brief The QXmlDefaultHandler class provides a default implementation of all |
| 2080 | the XML handler classes. |
| 2081 | |
| 2082 | \inmodule QtCore5Compat |
| 2083 | \ingroup xml-tools |
| 2084 | |
| 2085 | This class gathers together the features of |
| 2086 | the specialized handler classes, making it a convenient |
| 2087 | starting point when implementing custom handlers for |
| 2088 | subclasses of QXmlReader, particularly QXmlSimpleReader. |
| 2089 | The virtual functions from each of the base classes are |
| 2090 | reimplemented in this class, providing sensible default behavior |
| 2091 | for many common cases. By subclassing this class, and |
| 2092 | overriding these functions, you can concentrate |
| 2093 | on implementing the parts of the handler relevant to your |
| 2094 | application. |
| 2095 | |
| 2096 | The XML reader must be told which handler to use for different |
| 2097 | kinds of events during parsing. This means that, although |
| 2098 | QXmlDefaultHandler provides default implementations of functions |
| 2099 | inherited from all its base classes, we can still use specialized |
| 2100 | handlers for particular kinds of events. |
| 2101 | |
| 2102 | For example, QXmlDefaultHandler subclasses both |
| 2103 | QXmlContentHandler and QXmlErrorHandler, so by subclassing |
| 2104 | it we can use the same handler for both of the following |
| 2105 | reader functions: |
| 2106 | |
| 2107 | \snippet rsslisting/listing.cpp 0 |
| 2108 | |
| 2109 | Since the reader will inform the handler of parsing errors, it is |
| 2110 | necessary to reimplement QXmlErrorHandler::fatalError() if, for |
| 2111 | example, we want to stop parsing when such an error occurs: |
| 2112 | |
| 2113 | \snippet rsslisting/handler.cpp 0 |
| 2114 | |
| 2115 | The above function returns \c false, which tells the reader to stop |
| 2116 | parsing. To continue to use the same reader, |
| 2117 | it is necessary to create a new handler instance, and set up the |
| 2118 | reader to use it in the manner described above. |
| 2119 | |
| 2120 | It is useful to examine some of the functions inherited by |
| 2121 | QXmlDefaultHandler, and consider why they might be |
| 2122 | reimplemented in a custom handler. |
| 2123 | Custom handlers will typically reimplement |
| 2124 | QXmlContentHandler::startDocument() to prepare the handler for |
| 2125 | new content. Document elements and the text within them can be |
| 2126 | processed by reimplementing QXmlContentHandler::startElement(), |
| 2127 | QXmlContentHandler::endElement(), and |
| 2128 | QXmlContentHandler::characters(). |
| 2129 | You may want to reimplement QXmlContentHandler::endDocument() |
| 2130 | to perform some finalization or validation on the content once the |
| 2131 | document has been read completely. |
| 2132 | |
| 2133 | \sa QXmlDTDHandler, QXmlDeclHandler, QXmlContentHandler, QXmlEntityResolver, |
| 2134 | QXmlErrorHandler, QXmlLexicalHandler |
| 2135 | */ |
| 2136 | |
| 2137 | /*! |
| 2138 | \fn QXmlDefaultHandler::QXmlDefaultHandler() |
| 2139 | |
| 2140 | Constructs a handler for use with subclasses of QXmlReader. |
| 2141 | */ |
| 2142 | QXmlDefaultHandler::QXmlDefaultHandler() |
| 2143 | { |
| 2144 | // ### In Qt 5.0, this function was inlined and d was not initialized |
| 2145 | // The member cannot be used until Qt 6.0 |
| 2146 | Q_UNUSED(d); |
| 2147 | } |
| 2148 | |
| 2149 | /*! |
| 2150 | \fn QXmlDefaultHandler::~QXmlDefaultHandler() |
| 2151 | |
| 2152 | Destroys the handler. |
| 2153 | */ |
| 2154 | QXmlDefaultHandler::~QXmlDefaultHandler() |
| 2155 | { |
| 2156 | } |
| 2157 | |
| 2158 | /*! |
| 2159 | \reimp |
| 2160 | |
| 2161 | This reimplementation does nothing. |
| 2162 | */ |
| 2163 | void QXmlDefaultHandler::setDocumentLocator(QXmlLocator*) |
| 2164 | { |
| 2165 | } |
| 2166 | |
| 2167 | /*! |
| 2168 | \reimp |
| 2169 | |
| 2170 | This reimplementation does nothing. |
| 2171 | */ |
| 2172 | bool QXmlDefaultHandler::startDocument() |
| 2173 | { |
| 2174 | return true; |
| 2175 | } |
| 2176 | |
| 2177 | /*! |
| 2178 | \reimp |
| 2179 | |
| 2180 | This reimplementation does nothing. |
| 2181 | */ |
| 2182 | bool QXmlDefaultHandler::endDocument() |
| 2183 | { |
| 2184 | return true; |
| 2185 | } |
| 2186 | |
| 2187 | /*! |
| 2188 | \reimp |
| 2189 | |
| 2190 | This reimplementation does nothing. |
| 2191 | */ |
| 2192 | bool QXmlDefaultHandler::startPrefixMapping(const QString&, const QString&) |
| 2193 | { |
| 2194 | return true; |
| 2195 | } |
| 2196 | |
| 2197 | /*! |
| 2198 | \reimp |
| 2199 | |
| 2200 | This reimplementation does nothing. |
| 2201 | */ |
| 2202 | bool QXmlDefaultHandler::endPrefixMapping(const QString&) |
| 2203 | { |
| 2204 | return true; |
| 2205 | } |
| 2206 | |
| 2207 | /*! |
| 2208 | \reimp |
| 2209 | |
| 2210 | This reimplementation does nothing. |
| 2211 | */ |
| 2212 | bool QXmlDefaultHandler::startElement(const QString&, const QString&, |
| 2213 | const QString&, const QXmlAttributes&) |
| 2214 | { |
| 2215 | return true; |
| 2216 | } |
| 2217 | |
| 2218 | /*! |
| 2219 | \reimp |
| 2220 | |
| 2221 | This reimplementation does nothing. |
| 2222 | */ |
| 2223 | bool QXmlDefaultHandler::endElement(const QString&, const QString&, |
| 2224 | const QString&) |
| 2225 | { |
| 2226 | return true; |
| 2227 | } |
| 2228 | |
| 2229 | /*! |
| 2230 | \reimp |
| 2231 | |
| 2232 | This reimplementation does nothing. |
| 2233 | */ |
| 2234 | bool QXmlDefaultHandler::characters(const QString&) |
| 2235 | { |
| 2236 | return true; |
| 2237 | } |
| 2238 | |
| 2239 | /*! |
| 2240 | \reimp |
| 2241 | |
| 2242 | This reimplementation does nothing. |
| 2243 | */ |
| 2244 | bool QXmlDefaultHandler::ignorableWhitespace(const QString&) |
| 2245 | { |
| 2246 | return true; |
| 2247 | } |
| 2248 | |
| 2249 | /*! |
| 2250 | \reimp |
| 2251 | |
| 2252 | This reimplementation does nothing. |
| 2253 | */ |
| 2254 | bool QXmlDefaultHandler::processingInstruction(const QString&, |
| 2255 | const QString&) |
| 2256 | { |
| 2257 | return true; |
| 2258 | } |
| 2259 | |
| 2260 | /*! |
| 2261 | \reimp |
| 2262 | |
| 2263 | This reimplementation does nothing. |
| 2264 | */ |
| 2265 | bool QXmlDefaultHandler::skippedEntity(const QString&) |
| 2266 | { |
| 2267 | return true; |
| 2268 | } |
| 2269 | |
| 2270 | /*! |
| 2271 | \reimp |
| 2272 | |
| 2273 | This reimplementation does nothing. |
| 2274 | */ |
| 2275 | bool QXmlDefaultHandler::warning(const QXmlParseException&) |
| 2276 | { |
| 2277 | return true; |
| 2278 | } |
| 2279 | |
| 2280 | /*! |
| 2281 | \reimp |
| 2282 | |
| 2283 | This reimplementation does nothing. |
| 2284 | */ |
| 2285 | bool QXmlDefaultHandler::error(const QXmlParseException&) |
| 2286 | { |
| 2287 | return true; |
| 2288 | } |
| 2289 | |
| 2290 | /*! |
| 2291 | \reimp |
| 2292 | |
| 2293 | This reimplementation does nothing. |
| 2294 | */ |
| 2295 | bool QXmlDefaultHandler::fatalError(const QXmlParseException&) |
| 2296 | { |
| 2297 | return true; |
| 2298 | } |
| 2299 | |
| 2300 | /*! |
| 2301 | \reimp |
| 2302 | |
| 2303 | This reimplementation does nothing. |
| 2304 | */ |
| 2305 | bool QXmlDefaultHandler::notationDecl(const QString&, const QString&, |
| 2306 | const QString&) |
| 2307 | { |
| 2308 | return true; |
| 2309 | } |
| 2310 | |
| 2311 | /*! |
| 2312 | \reimp |
| 2313 | |
| 2314 | This reimplementation does nothing. |
| 2315 | */ |
| 2316 | bool QXmlDefaultHandler::unparsedEntityDecl(const QString&, const QString&, |
| 2317 | const QString&, const QString&) |
| 2318 | { |
| 2319 | return true; |
| 2320 | } |
| 2321 | |
| 2322 | /*! |
| 2323 | \reimp |
| 2324 | |
| 2325 | Sets \a ret to \nullptr, so that the reader uses the system identifier |
| 2326 | provided in the XML document. |
| 2327 | */ |
| 2328 | bool QXmlDefaultHandler::resolveEntity(const QString&, const QString&, |
| 2329 | QXmlInputSource*& ret) |
| 2330 | { |
| 2331 | ret = nullptr; |
| 2332 | return true; |
| 2333 | } |
| 2334 | |
| 2335 | /*! |
| 2336 | \reimp |
| 2337 | |
| 2338 | Returns the default error string. |
| 2339 | */ |
| 2340 | QString QXmlDefaultHandler::errorString() const |
| 2341 | { |
| 2342 | return QString::fromLatin1(XMLERR_ERRORBYCONSUMER); |
| 2343 | } |
| 2344 | |
| 2345 | /*! |
| 2346 | \reimp |
| 2347 | |
| 2348 | This reimplementation does nothing. |
| 2349 | */ |
| 2350 | bool QXmlDefaultHandler::startDTD(const QString&, const QString&, const QString&) |
| 2351 | { |
| 2352 | return true; |
| 2353 | } |
| 2354 | |
| 2355 | /*! |
| 2356 | \reimp |
| 2357 | |
| 2358 | This reimplementation does nothing. |
| 2359 | */ |
| 2360 | bool QXmlDefaultHandler::endDTD() |
| 2361 | { |
| 2362 | return true; |
| 2363 | } |
| 2364 | |
| 2365 | /*! |
| 2366 | \reimp |
| 2367 | |
| 2368 | This reimplementation does nothing. |
| 2369 | */ |
| 2370 | bool QXmlDefaultHandler::startEntity(const QString&) |
| 2371 | { |
| 2372 | return true; |
| 2373 | } |
| 2374 | |
| 2375 | /*! |
| 2376 | \reimp |
| 2377 | |
| 2378 | This reimplementation does nothing. |
| 2379 | */ |
| 2380 | bool QXmlDefaultHandler::endEntity(const QString&) |
| 2381 | { |
| 2382 | return true; |
| 2383 | } |
| 2384 | |
| 2385 | /*! |
| 2386 | \reimp |
| 2387 | |
| 2388 | This reimplementation does nothing. |
| 2389 | */ |
| 2390 | bool QXmlDefaultHandler::startCDATA() |
| 2391 | { |
| 2392 | return true; |
| 2393 | } |
| 2394 | |
| 2395 | /*! |
| 2396 | \reimp |
| 2397 | |
| 2398 | This reimplementation does nothing. |
| 2399 | */ |
| 2400 | bool QXmlDefaultHandler::endCDATA() |
| 2401 | { |
| 2402 | return true; |
| 2403 | } |
| 2404 | |
| 2405 | /*! |
| 2406 | \reimp |
| 2407 | |
| 2408 | This reimplementation does nothing. |
| 2409 | */ |
| 2410 | bool QXmlDefaultHandler::comment(const QString&) |
| 2411 | { |
| 2412 | return true; |
| 2413 | } |
| 2414 | |
| 2415 | /*! |
| 2416 | \reimp |
| 2417 | |
| 2418 | This reimplementation does nothing. |
| 2419 | */ |
| 2420 | bool QXmlDefaultHandler::attributeDecl(const QString&, const QString&, const QString&, const QString&, const QString&) |
| 2421 | { |
| 2422 | return true; |
| 2423 | } |
| 2424 | |
| 2425 | /*! |
| 2426 | \reimp |
| 2427 | |
| 2428 | This reimplementation does nothing. |
| 2429 | */ |
| 2430 | bool QXmlDefaultHandler::internalEntityDecl(const QString&, const QString&) |
| 2431 | { |
| 2432 | return true; |
| 2433 | } |
| 2434 | |
| 2435 | /*! |
| 2436 | \reimp |
| 2437 | |
| 2438 | This reimplementation does nothing. |
| 2439 | */ |
| 2440 | bool QXmlDefaultHandler::externalEntityDecl(const QString&, const QString&, const QString&) |
| 2441 | { |
| 2442 | return true; |
| 2443 | } |
| 2444 | |
| 2445 | /********************************************* |
| 2446 | * |
| 2447 | * QXmlSimpleReaderPrivate |
| 2448 | * |
| 2449 | *********************************************/ |
| 2450 | |
| 2451 | inline bool QXmlSimpleReaderPrivate::atEnd() |
| 2452 | { |
| 2453 | return (c.unicode()|0x0001) == 0xffff; |
| 2454 | } |
| 2455 | |
| 2456 | inline void QXmlSimpleReaderPrivate::stringClear() |
| 2457 | { |
| 2458 | stringValueLen = 0; stringArrayPos = 0; |
| 2459 | } |
| 2460 | inline void QXmlSimpleReaderPrivate::nameClear() |
| 2461 | { |
| 2462 | nameValueLen = 0; nameArrayPos = 0; |
| 2463 | } |
| 2464 | |
| 2465 | inline void QXmlSimpleReaderPrivate::refClear() |
| 2466 | { |
| 2467 | refValueLen = 0; refArrayPos = 0; |
| 2468 | } |
| 2469 | |
| 2470 | QXmlSimpleReaderPrivate::QXmlSimpleReaderPrivate(QXmlSimpleReader *reader) |
| 2471 | { |
| 2472 | q_ptr = reader; |
| 2473 | parseStack = nullptr; |
| 2474 | |
| 2475 | locator.reset(p: new QXmlSimpleReaderLocator(reader)); |
| 2476 | entityRes = nullptr; |
| 2477 | dtdHnd = nullptr; |
| 2478 | contentHnd = nullptr; |
| 2479 | errorHnd = nullptr; |
| 2480 | lexicalHnd = nullptr; |
| 2481 | declHnd = nullptr; |
| 2482 | |
| 2483 | // default feature settings |
| 2484 | useNamespaces = true; |
| 2485 | useNamespacePrefixes = false; |
| 2486 | reportWhitespaceCharData = true; |
| 2487 | reportEntities = false; |
| 2488 | } |
| 2489 | |
| 2490 | QXmlSimpleReaderPrivate::~QXmlSimpleReaderPrivate() |
| 2491 | { |
| 2492 | delete parseStack; |
| 2493 | } |
| 2494 | |
| 2495 | void QXmlSimpleReaderPrivate::initIncrementalParsing() |
| 2496 | { |
| 2497 | if (parseStack) |
| 2498 | parseStack->clear(); |
| 2499 | else |
| 2500 | parseStack = new QStack<ParseState>; |
| 2501 | } |
| 2502 | |
| 2503 | /********************************************* |
| 2504 | * |
| 2505 | * QXmlSimpleReader |
| 2506 | * |
| 2507 | *********************************************/ |
| 2508 | |
| 2509 | /*! |
| 2510 | \class QXmlReader |
| 2511 | \reentrant |
| 2512 | \brief The QXmlReader class provides an interface for XML readers (i.e. |
| 2513 | parsers). |
| 2514 | |
| 2515 | \inmodule QtCore5Compat |
| 2516 | \ingroup xml-tools |
| 2517 | |
| 2518 | This abstract class provides an interface for all of Qt's XML |
| 2519 | readers. Currently there is only one implementation of a reader |
| 2520 | included in Qt's XML module: QXmlSimpleReader. In future releases |
| 2521 | there might be more readers with different properties available |
| 2522 | (e.g. a validating parser). |
| 2523 | |
| 2524 | The design of the XML classes follows the |
| 2525 | \l {http://www.saxproject.org/}{SAX2 Java interface}, with |
| 2526 | the names adapted to fit Qt naming conventions. It should be very |
| 2527 | easy for anybody who has worked with SAX2 to get started with the |
| 2528 | Qt XML classes. |
| 2529 | |
| 2530 | All readers use the class QXmlInputSource to read the input |
| 2531 | document. Since you are normally interested in particular content |
| 2532 | in the XML document, the reader reports the content through |
| 2533 | special handler classes (QXmlDTDHandler, QXmlDeclHandler, |
| 2534 | QXmlContentHandler, QXmlEntityResolver, QXmlErrorHandler and |
| 2535 | QXmlLexicalHandler), which you must subclass, if you want to |
| 2536 | process the contents. |
| 2537 | |
| 2538 | Since the handler classes only describe interfaces you must |
| 2539 | implement all the functions. We provide the QXmlDefaultHandler |
| 2540 | class to make this easier: it implements a default behavior (do |
| 2541 | nothing) for all functions, so you can subclass it and just |
| 2542 | implement the functions you are interested in. |
| 2543 | |
| 2544 | Features and properties of the reader can be set with setFeature() |
| 2545 | and setProperty() respectively. You can set the reader to use your |
| 2546 | own subclasses with setEntityResolver(), setDTDHandler(), |
| 2547 | setContentHandler(), setErrorHandler(), setLexicalHandler() and |
| 2548 | setDeclHandler(). The parse itself is started with a call to |
| 2549 | parse(). |
| 2550 | |
| 2551 | Note that this class is now deprecated, please use QXmlStreamReader or |
| 2552 | QDomDocument for reading XML files. |
| 2553 | |
| 2554 | \sa QXmlSimpleReader |
| 2555 | */ |
| 2556 | |
| 2557 | /*! |
| 2558 | \fn QXmlReader::~QXmlReader() |
| 2559 | |
| 2560 | Destroys the reader. |
| 2561 | */ |
| 2562 | QXmlReader::~QXmlReader() |
| 2563 | = default; |
| 2564 | |
| 2565 | /*! |
| 2566 | \fn bool QXmlReader::feature(const QString& name, bool *ok) const |
| 2567 | |
| 2568 | If the reader has the feature called \a name, the feature's value |
| 2569 | is returned. If no such feature exists the return value is |
| 2570 | undefined. |
| 2571 | |
| 2572 | If \a ok is not \nullptr: \c{*}\a{ok} is set to true if the |
| 2573 | reader has the feature called \a name; otherwise \c{*}\a{ok} is |
| 2574 | set to false. |
| 2575 | |
| 2576 | \sa setFeature(), hasFeature() |
| 2577 | */ |
| 2578 | |
| 2579 | /*! |
| 2580 | \fn void QXmlReader::setFeature(const QString& name, bool value) |
| 2581 | |
| 2582 | Sets the feature called \a name to the given \a value. If the |
| 2583 | reader doesn't have the feature nothing happens. |
| 2584 | |
| 2585 | \sa feature(), hasFeature() |
| 2586 | */ |
| 2587 | |
| 2588 | /*! |
| 2589 | \fn bool QXmlReader::hasFeature(const QString& name) const |
| 2590 | |
| 2591 | Returns \c true if the reader has the feature called \a name; |
| 2592 | otherwise returns \c false. |
| 2593 | |
| 2594 | \sa feature(), setFeature() |
| 2595 | */ |
| 2596 | |
| 2597 | /*! |
| 2598 | \fn void* QXmlReader::property(const QString& name, bool *ok) const |
| 2599 | |
| 2600 | If the reader has the property \a name, this function returns the |
| 2601 | value of the property; otherwise the return value is undefined. |
| 2602 | |
| 2603 | If \a ok is not \nullptr: if the reader has the \a name property |
| 2604 | \c{*}\a{ok} is set to true; otherwise \c{*}\a{ok} is set to false. |
| 2605 | |
| 2606 | \sa setProperty(), hasProperty() |
| 2607 | */ |
| 2608 | |
| 2609 | /*! |
| 2610 | \fn void QXmlReader::setProperty(const QString& name, void* value) |
| 2611 | |
| 2612 | Sets the property \a name to \a value. If the reader doesn't have |
| 2613 | the property nothing happens. |
| 2614 | |
| 2615 | \sa property(), hasProperty() |
| 2616 | */ |
| 2617 | |
| 2618 | /*! |
| 2619 | \fn bool QXmlReader::hasProperty(const QString& name) const |
| 2620 | |
| 2621 | Returns \c true if the reader has the property \a name; otherwise |
| 2622 | returns \c false. |
| 2623 | |
| 2624 | \sa property(), setProperty() |
| 2625 | */ |
| 2626 | |
| 2627 | /*! |
| 2628 | \fn void QXmlReader::setEntityResolver(QXmlEntityResolver* handler) |
| 2629 | |
| 2630 | Sets the entity resolver to \a handler. |
| 2631 | |
| 2632 | \sa entityResolver() |
| 2633 | */ |
| 2634 | |
| 2635 | /*! |
| 2636 | \fn QXmlEntityResolver *QXmlReader::entityResolver() const |
| 2637 | |
| 2638 | Returns the entity resolver or \nullptr if none was set. |
| 2639 | |
| 2640 | \sa setEntityResolver() |
| 2641 | */ |
| 2642 | |
| 2643 | /*! |
| 2644 | \fn void QXmlReader::setDTDHandler(QXmlDTDHandler* handler) |
| 2645 | |
| 2646 | Sets the DTD handler to \a handler. |
| 2647 | |
| 2648 | \sa DTDHandler() |
| 2649 | */ |
| 2650 | |
| 2651 | /*! |
| 2652 | \fn QXmlDTDHandler *QXmlReader::DTDHandler() const |
| 2653 | |
| 2654 | Returns the DTD handler or \nullptr if none was set. |
| 2655 | |
| 2656 | \sa setDTDHandler() |
| 2657 | */ |
| 2658 | |
| 2659 | /*! |
| 2660 | \fn void QXmlReader::setContentHandler(QXmlContentHandler* handler) |
| 2661 | |
| 2662 | Sets the content handler to \a handler. |
| 2663 | |
| 2664 | \sa contentHandler() |
| 2665 | */ |
| 2666 | |
| 2667 | /*! |
| 2668 | \fn QXmlContentHandler *QXmlReader::contentHandler() const |
| 2669 | |
| 2670 | Returns the content handler or \nullptr if none was set. |
| 2671 | |
| 2672 | \sa setContentHandler() |
| 2673 | */ |
| 2674 | |
| 2675 | /*! |
| 2676 | \fn void QXmlReader::setErrorHandler(QXmlErrorHandler* handler) |
| 2677 | |
| 2678 | Sets the error handler to \a handler. Clears the error handler if |
| 2679 | \a handler is 0. |
| 2680 | |
| 2681 | \sa errorHandler() |
| 2682 | */ |
| 2683 | |
| 2684 | /*! |
| 2685 | \fn QXmlErrorHandler *QXmlReader::errorHandler() const |
| 2686 | |
| 2687 | Returns the error handler or \nullptr if none is set. |
| 2688 | |
| 2689 | \sa setErrorHandler() |
| 2690 | */ |
| 2691 | |
| 2692 | /*! |
| 2693 | \fn void QXmlReader::setLexicalHandler(QXmlLexicalHandler* handler) |
| 2694 | |
| 2695 | Sets the lexical handler to \a handler. |
| 2696 | |
| 2697 | \sa lexicalHandler() |
| 2698 | */ |
| 2699 | |
| 2700 | /*! |
| 2701 | \fn QXmlLexicalHandler *QXmlReader::lexicalHandler() const |
| 2702 | |
| 2703 | Returns the lexical handler or \nullptr if none was set. |
| 2704 | |
| 2705 | \sa setLexicalHandler() |
| 2706 | */ |
| 2707 | |
| 2708 | /*! |
| 2709 | \fn void QXmlReader::setDeclHandler(QXmlDeclHandler* handler) |
| 2710 | |
| 2711 | Sets the declaration handler to \a handler. |
| 2712 | |
| 2713 | \sa declHandler() |
| 2714 | */ |
| 2715 | |
| 2716 | /*! |
| 2717 | \fn QXmlDeclHandler *QXmlReader::declHandler() const |
| 2718 | |
| 2719 | Returns the declaration handler or \nullptr if none was set. |
| 2720 | |
| 2721 | \sa setDeclHandler() |
| 2722 | */ |
| 2723 | |
| 2724 | /*! |
| 2725 | \fn bool QXmlReader::parse(const QXmlInputSource &input) |
| 2726 | |
| 2727 | \deprecated |
| 2728 | |
| 2729 | Parses the given \a input. |
| 2730 | */ |
| 2731 | |
| 2732 | /*! |
| 2733 | \fn bool QXmlReader::parse(const QXmlInputSource *input) |
| 2734 | |
| 2735 | Reads an XML document from \a input and parses it. Returns \c true if |
| 2736 | the parsing was successful; otherwise returns \c false. |
| 2737 | */ |
| 2738 | |
| 2739 | /*! |
| 2740 | \class QXmlSimpleReader |
| 2741 | \nonreentrant |
| 2742 | \brief The QXmlSimpleReader class provides an implementation of a |
| 2743 | simple XML parser. |
| 2744 | |
| 2745 | \inmodule QtCore5Compat |
| 2746 | \ingroup xml-tools |
| 2747 | |
| 2748 | This XML reader is suitable for a wide range of applications. It |
| 2749 | is able to parse well-formed XML and can report the namespaces of |
| 2750 | elements to a content handler; however, it does not parse any |
| 2751 | external entities. For historical reasons, Attribute Value |
| 2752 | Normalization and End-of-Line Handling as described in the XML 1.0 |
| 2753 | specification is not performed. |
| 2754 | |
| 2755 | The easiest pattern of use for this class is to create a reader |
| 2756 | instance, define an input source, specify the handlers to be used |
| 2757 | by the reader, and parse the data. |
| 2758 | |
| 2759 | For example, we could use a QFile to supply the input. Here, we |
| 2760 | create a reader, and define an input source to be used by the |
| 2761 | reader: |
| 2762 | |
| 2763 | \snippet simpleparse/main.cpp 0 |
| 2764 | |
| 2765 | A handler lets us perform actions when the reader encounters |
| 2766 | certain types of content, or if errors in the input are found. The |
| 2767 | reader must be told which handler to use for each type of |
| 2768 | event. For many common applications, we can create a custom |
| 2769 | handler by subclassing QXmlDefaultHandler, and use this to handle |
| 2770 | both error and content events: |
| 2771 | |
| 2772 | \snippet simpleparse/main.cpp 1 |
| 2773 | |
| 2774 | If you don't set at least the content and error handlers, the |
| 2775 | parser will fall back on its default behavior---and will do |
| 2776 | nothing. |
| 2777 | |
| 2778 | The most convenient way to handle the input is to read it in a |
| 2779 | single pass using the parse() function with an argument that |
| 2780 | specifies the input source: |
| 2781 | |
| 2782 | \snippet simpleparse/main.cpp 2 |
| 2783 | |
| 2784 | If you can't parse the entire input in one go (for example, it is |
| 2785 | huge, or is being delivered over a network connection), data can |
| 2786 | be fed to the parser in pieces. This is achieved by telling |
| 2787 | parse() to work incrementally, and making subsequent calls to the |
| 2788 | parseContinue() function, until all the data has been processed. |
| 2789 | |
| 2790 | A common way to perform incremental parsing is to connect the \c |
| 2791 | readyRead() signal of a \l{QNetworkReply} {network reply} a slot, |
| 2792 | and handle the incoming data there. See QNetworkAccessManager. |
| 2793 | |
| 2794 | Aspects of the parsing behavior can be adapted using setFeature() |
| 2795 | and setProperty(). |
| 2796 | |
| 2797 | \snippet code/src_xml_sax_qxml.cpp 0 |
| 2798 | |
| 2799 | QXmlSimpleReader is not reentrant. If you want to use the class |
| 2800 | in threaded code, lock the code using QXmlSimpleReader with a |
| 2801 | locking mechanism, such as a QMutex. |
| 2802 | |
| 2803 | Note that this class is now deprecated, please use QXmlStreamReader or |
| 2804 | QDomDocument for reading XML files. |
| 2805 | */ |
| 2806 | |
| 2807 | static inline bool is_S(QChar ch) |
| 2808 | { |
| 2809 | ushort uc = ch.unicode(); |
| 2810 | return (uc == ' ' || uc == '\t' || uc == '\n' || uc == '\r'); |
| 2811 | } |
| 2812 | |
| 2813 | enum NameChar { NameBeginning, NameNotBeginning, NotName }; |
| 2814 | |
| 2815 | static const char Begi = (char)NameBeginning; |
| 2816 | static const char NtBg = (char)NameNotBeginning; |
| 2817 | static const char NotN = (char)NotName; |
| 2818 | |
| 2819 | static const char nameCharTable[128] = |
| 2820 | { |
| 2821 | // 0x00 |
| 2822 | NotN, NotN, NotN, NotN, NotN, NotN, NotN, NotN, |
| 2823 | NotN, NotN, NotN, NotN, NotN, NotN, NotN, NotN, |
| 2824 | // 0x10 |
| 2825 | NotN, NotN, NotN, NotN, NotN, NotN, NotN, NotN, |
| 2826 | NotN, NotN, NotN, NotN, NotN, NotN, NotN, NotN, |
| 2827 | // 0x20 (0x2D is '-', 0x2E is '.') |
| 2828 | NotN, NotN, NotN, NotN, NotN, NotN, NotN, NotN, |
| 2829 | NotN, NotN, NotN, NotN, NotN, NtBg, NtBg, NotN, |
| 2830 | // 0x30 (0x30..0x39 are '0'..'9', 0x3A is ':') |
| 2831 | NtBg, NtBg, NtBg, NtBg, NtBg, NtBg, NtBg, NtBg, |
| 2832 | NtBg, NtBg, Begi, NotN, NotN, NotN, NotN, NotN, |
| 2833 | // 0x40 (0x41..0x5A are 'A'..'Z') |
| 2834 | NotN, Begi, Begi, Begi, Begi, Begi, Begi, Begi, |
| 2835 | Begi, Begi, Begi, Begi, Begi, Begi, Begi, Begi, |
| 2836 | // 0x50 (0x5F is '_') |
| 2837 | Begi, Begi, Begi, Begi, Begi, Begi, Begi, Begi, |
| 2838 | Begi, Begi, Begi, NotN, NotN, NotN, NotN, Begi, |
| 2839 | // 0x60 (0x61..0x7A are 'a'..'z') |
| 2840 | NotN, Begi, Begi, Begi, Begi, Begi, Begi, Begi, |
| 2841 | Begi, Begi, Begi, Begi, Begi, Begi, Begi, Begi, |
| 2842 | // 0x70 |
| 2843 | Begi, Begi, Begi, Begi, Begi, Begi, Begi, Begi, |
| 2844 | Begi, Begi, Begi, NotN, NotN, NotN, NotN, NotN |
| 2845 | }; |
| 2846 | |
| 2847 | static inline NameChar fastDetermineNameChar(QChar ch) |
| 2848 | { |
| 2849 | ushort uc = ch.unicode(); |
| 2850 | if (!(uc & ~0x7f)) // uc < 128 |
| 2851 | return (NameChar)nameCharTable[uc]; |
| 2852 | |
| 2853 | QChar::Category cat = ch.category(); |
| 2854 | // ### some these categories might be slightly wrong |
| 2855 | if ((cat >= QChar::Letter_Uppercase && cat <= QChar::Letter_Other) |
| 2856 | || cat == QChar::Number_Letter) |
| 2857 | return NameBeginning; |
| 2858 | if ((cat >= QChar::Number_DecimalDigit && cat <= QChar::Number_Other) |
| 2859 | || (cat >= QChar::Mark_NonSpacing && cat <= QChar::Mark_Enclosing)) |
| 2860 | return NameNotBeginning; |
| 2861 | return NotName; |
| 2862 | } |
| 2863 | |
| 2864 | static NameChar determineNameChar(QChar ch) |
| 2865 | { |
| 2866 | return fastDetermineNameChar(ch); |
| 2867 | } |
| 2868 | |
| 2869 | /*! |
| 2870 | Constructs a simple XML reader. |
| 2871 | |
| 2872 | */ |
| 2873 | QXmlSimpleReader::QXmlSimpleReader() |
| 2874 | : d_ptr(new QXmlSimpleReaderPrivate(this)) |
| 2875 | { |
| 2876 | } |
| 2877 | |
| 2878 | /*! |
| 2879 | Destroys the simple XML reader. |
| 2880 | */ |
| 2881 | QXmlSimpleReader::~QXmlSimpleReader() |
| 2882 | { |
| 2883 | } |
| 2884 | |
| 2885 | /*! |
| 2886 | \reimp |
| 2887 | */ |
| 2888 | bool QXmlSimpleReader::feature(const QString& name, bool *ok) const |
| 2889 | { |
| 2890 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 2891 | |
| 2892 | if (ok) |
| 2893 | *ok = true; |
| 2894 | if (name == QLatin1String("http://xml.org/sax/features/namespaces" )) { |
| 2895 | return d->useNamespaces; |
| 2896 | } else if (name == QLatin1String("http://xml.org/sax/features/namespace-prefixes" )) { |
| 2897 | return d->useNamespacePrefixes; |
| 2898 | } else if (name == QLatin1String("http://trolltech.com/xml/features/report-whitespace-only-CharData" ) // For compat with Qt 4 |
| 2899 | || name == QLatin1String("http://qt-project.org/xml/features/report-whitespace-only-CharData" )) { |
| 2900 | return d->reportWhitespaceCharData; |
| 2901 | } else if (name == QLatin1String("http://trolltech.com/xml/features/report-start-end-entity" ) // For compat with Qt 4 |
| 2902 | || name == QLatin1String("http://qt-project.org/xml/features/report-start-end-entity" )) { |
| 2903 | return d->reportEntities; |
| 2904 | } else { |
| 2905 | qWarning(msg: "Unknown feature %s" , name.toLatin1().data()); |
| 2906 | if (ok) |
| 2907 | *ok = false; |
| 2908 | } |
| 2909 | return false; |
| 2910 | } |
| 2911 | |
| 2912 | /*! |
| 2913 | Turns on the feature \a name if \a enable is true; otherwise turns it off. |
| 2914 | |
| 2915 | The \a name parameter must be one of the following strings: |
| 2916 | \table |
| 2917 | \header \li Feature \li Default \li Notes |
| 2918 | \row \li \e http://xml.org/sax/features/namespaces |
| 2919 | \li true |
| 2920 | \li If enabled, namespaces are reported to the content handler. |
| 2921 | \row \li \e http://xml.org/sax/features/namespace-prefixes |
| 2922 | \li false |
| 2923 | \li If enabled, the original prefixed names |
| 2924 | and attributes used for namespace declarations are |
| 2925 | reported. |
| 2926 | \row \li \e http://qt-project.org/xml/features/report-whitespace-only-CharData |
| 2927 | \li true |
| 2928 | \li If enabled, CharData that consist of |
| 2929 | only whitespace characters are reported |
| 2930 | using QXmlContentHandler::characters(). If disabled, whitespace is silently |
| 2931 | discarded. |
| 2932 | \row \li \e http://qt-project.org/xml/features/report-start-end-entity |
| 2933 | \li false |
| 2934 | \li If enabled, the parser reports |
| 2935 | QXmlContentHandler::startEntity() and |
| 2936 | QXmlContentHandler::endEntity() events, so character data |
| 2937 | might be reported in chunks. |
| 2938 | If disabled, the parser does not report these events, but |
| 2939 | silently substitutes the entities, and reports the character |
| 2940 | data in one chunk. |
| 2941 | \endtable |
| 2942 | |
| 2943 | \sa feature(), hasFeature() |
| 2944 | */ |
| 2945 | void QXmlSimpleReader::setFeature(const QString& name, bool enable) |
| 2946 | { |
| 2947 | Q_D(QXmlSimpleReader); |
| 2948 | if (name == QLatin1String("http://xml.org/sax/features/namespaces" )) { |
| 2949 | d->useNamespaces = enable; |
| 2950 | } else if (name == QLatin1String("http://xml.org/sax/features/namespace-prefixes" )) { |
| 2951 | d->useNamespacePrefixes = enable; |
| 2952 | } else if (name == QLatin1String("http://trolltech.com/xml/features/report-whitespace-only-CharData" ) // For compat with Qt 4 |
| 2953 | || name == QLatin1String("http://qt-project.org/xml/features/report-whitespace-only-CharData" )) { |
| 2954 | d->reportWhitespaceCharData = enable; |
| 2955 | } else if (name == QLatin1String("http://trolltech.com/xml/features/report-start-end-entity" ) // For compat with Qt 4 |
| 2956 | || name == QLatin1String("http://qt-project.org/xml/features/report-start-end-entity" )) { |
| 2957 | d->reportEntities = enable; |
| 2958 | } else { |
| 2959 | qWarning(msg: "Unknown feature %s" , name.toLatin1().data()); |
| 2960 | } |
| 2961 | } |
| 2962 | |
| 2963 | /*! \reimp |
| 2964 | */ |
| 2965 | bool QXmlSimpleReader::hasFeature(const QString& name) const |
| 2966 | { |
| 2967 | if (name == QLatin1String("http://xml.org/sax/features/namespaces" ) |
| 2968 | || name == QLatin1String("http://xml.org/sax/features/namespace-prefixes" ) |
| 2969 | || name == QLatin1String("http://trolltech.com/xml/features/report-whitespace-only-CharData" ) // For compat with Qt 4 |
| 2970 | || name == QLatin1String("http://qt-project.org/xml/features/report-whitespace-only-CharData" ) |
| 2971 | || name == QLatin1String("http://trolltech.com/xml/features/report-start-end-entity" ) // For compat with Qt 4 |
| 2972 | || name == QLatin1String("http://qt-project.org/xml/features/report-start-end-entity" )) { |
| 2973 | return true; |
| 2974 | } else { |
| 2975 | return false; |
| 2976 | } |
| 2977 | } |
| 2978 | |
| 2979 | /*! \reimp |
| 2980 | */ |
| 2981 | void* QXmlSimpleReader::property(const QString&, bool *ok) const |
| 2982 | { |
| 2983 | if (ok) |
| 2984 | *ok = false; |
| 2985 | return nullptr; |
| 2986 | } |
| 2987 | |
| 2988 | /*! \reimp |
| 2989 | */ |
| 2990 | void QXmlSimpleReader::setProperty(const QString&, void*) |
| 2991 | { |
| 2992 | } |
| 2993 | |
| 2994 | /*! |
| 2995 | \reimp |
| 2996 | */ |
| 2997 | bool QXmlSimpleReader::hasProperty(const QString&) const |
| 2998 | { |
| 2999 | return false; |
| 3000 | } |
| 3001 | |
| 3002 | /*! |
| 3003 | \reimp |
| 3004 | */ |
| 3005 | void QXmlSimpleReader::setEntityResolver(QXmlEntityResolver* handler) |
| 3006 | { |
| 3007 | Q_D(QXmlSimpleReader); |
| 3008 | d->entityRes = handler; |
| 3009 | } |
| 3010 | |
| 3011 | /*! |
| 3012 | \reimp |
| 3013 | */ |
| 3014 | QXmlEntityResolver* QXmlSimpleReader::entityResolver() const |
| 3015 | { |
| 3016 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 3017 | return d->entityRes; |
| 3018 | } |
| 3019 | |
| 3020 | /*! |
| 3021 | \reimp |
| 3022 | */ |
| 3023 | void QXmlSimpleReader::setDTDHandler(QXmlDTDHandler* handler) |
| 3024 | { |
| 3025 | Q_D(QXmlSimpleReader); |
| 3026 | d->dtdHnd = handler; |
| 3027 | } |
| 3028 | |
| 3029 | /*! |
| 3030 | \reimp |
| 3031 | */ |
| 3032 | QXmlDTDHandler* QXmlSimpleReader::DTDHandler() const |
| 3033 | { |
| 3034 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 3035 | return d->dtdHnd; |
| 3036 | } |
| 3037 | |
| 3038 | /*! |
| 3039 | \reimp |
| 3040 | */ |
| 3041 | void QXmlSimpleReader::setContentHandler(QXmlContentHandler* handler) |
| 3042 | { |
| 3043 | Q_D(QXmlSimpleReader); |
| 3044 | d->contentHnd = handler; |
| 3045 | } |
| 3046 | |
| 3047 | /*! |
| 3048 | \reimp |
| 3049 | */ |
| 3050 | QXmlContentHandler* QXmlSimpleReader::contentHandler() const |
| 3051 | { |
| 3052 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 3053 | return d->contentHnd; |
| 3054 | } |
| 3055 | |
| 3056 | /*! |
| 3057 | \reimp |
| 3058 | */ |
| 3059 | void QXmlSimpleReader::setErrorHandler(QXmlErrorHandler* handler) |
| 3060 | { |
| 3061 | Q_D(QXmlSimpleReader); |
| 3062 | d->errorHnd = handler; |
| 3063 | } |
| 3064 | |
| 3065 | /*! |
| 3066 | \reimp |
| 3067 | */ |
| 3068 | QXmlErrorHandler* QXmlSimpleReader::errorHandler() const |
| 3069 | { |
| 3070 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 3071 | return d->errorHnd; |
| 3072 | } |
| 3073 | |
| 3074 | /*! |
| 3075 | \reimp |
| 3076 | */ |
| 3077 | void QXmlSimpleReader::setLexicalHandler(QXmlLexicalHandler* handler) |
| 3078 | { |
| 3079 | Q_D(QXmlSimpleReader); |
| 3080 | d->lexicalHnd = handler; |
| 3081 | } |
| 3082 | |
| 3083 | /*! |
| 3084 | \reimp |
| 3085 | */ |
| 3086 | QXmlLexicalHandler* QXmlSimpleReader::lexicalHandler() const |
| 3087 | { |
| 3088 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 3089 | return d->lexicalHnd; |
| 3090 | } |
| 3091 | |
| 3092 | /*! |
| 3093 | \reimp |
| 3094 | */ |
| 3095 | void QXmlSimpleReader::setDeclHandler(QXmlDeclHandler* handler) |
| 3096 | { |
| 3097 | Q_D(QXmlSimpleReader); |
| 3098 | d->declHnd = handler; |
| 3099 | } |
| 3100 | |
| 3101 | /*! |
| 3102 | \reimp |
| 3103 | */ |
| 3104 | QXmlDeclHandler* QXmlSimpleReader::declHandler() const |
| 3105 | { |
| 3106 | const QXmlSimpleReaderPrivate *d = d_func(); |
| 3107 | return d->declHnd; |
| 3108 | } |
| 3109 | |
| 3110 | /*! |
| 3111 | \reimp |
| 3112 | */ |
| 3113 | bool QXmlSimpleReader::parse(const QXmlInputSource& input) |
| 3114 | { |
| 3115 | return parse(input: &input, incremental: false); |
| 3116 | } |
| 3117 | |
| 3118 | /*! |
| 3119 | Reads an XML document from \a input and parses it in one pass (non-incrementally). |
| 3120 | Returns \c true if the parsing was successful; otherwise returns \c false. |
| 3121 | */ |
| 3122 | bool QXmlSimpleReader::parse(const QXmlInputSource* input) |
| 3123 | { |
| 3124 | return parse(input, incremental: false); |
| 3125 | } |
| 3126 | |
| 3127 | /*! |
| 3128 | Reads an XML document from \a input and parses it. Returns \c true |
| 3129 | if the parsing is completed successfully; otherwise returns \c false, |
| 3130 | indicating that an error occurred. |
| 3131 | |
| 3132 | If \a incremental is false, this function will return false if the XML |
| 3133 | file is not read completely. The parsing cannot be continued in this |
| 3134 | case. |
| 3135 | |
| 3136 | If \a incremental is true, the parser does not return false if |
| 3137 | it reaches the end of the \a input before reaching the end |
| 3138 | of the XML file. Instead, it stores the state of the parser so that |
| 3139 | parsing can be continued later when more data is available. |
| 3140 | In such a case, you can use the function parseContinue() to |
| 3141 | continue with parsing. This class stores a pointer to the input |
| 3142 | source \a input and the parseContinue() function tries to read from |
| 3143 | that input source. Therefore, you should not delete the input |
| 3144 | source \a input until you no longer need to call parseContinue(). |
| 3145 | |
| 3146 | If this function is called with \a incremental set to true |
| 3147 | while an incremental parse is in progress, a new parsing |
| 3148 | session will be started, and the previous session will be lost. |
| 3149 | |
| 3150 | \sa parseContinue(), QTcpSocket |
| 3151 | */ |
| 3152 | bool QXmlSimpleReader::parse(const QXmlInputSource *input, bool incremental) |
| 3153 | { |
| 3154 | Q_D(QXmlSimpleReader); |
| 3155 | |
| 3156 | d->literalEntitySizes.clear(); |
| 3157 | d->referencesToOtherEntities.clear(); |
| 3158 | d->expandedSizes.clear(); |
| 3159 | |
| 3160 | if (incremental) { |
| 3161 | d->initIncrementalParsing(); |
| 3162 | } else { |
| 3163 | delete d->parseStack; |
| 3164 | d->parseStack = nullptr; |
| 3165 | } |
| 3166 | d->init(i: input); |
| 3167 | |
| 3168 | // call the handler |
| 3169 | if (d->contentHnd) { |
| 3170 | d->contentHnd->setDocumentLocator(d->locator.get()); |
| 3171 | if (!d->contentHnd->startDocument()) { |
| 3172 | d->reportParseError(error: d->contentHnd->errorString()); |
| 3173 | clear(c&: d->tags); |
| 3174 | return false; |
| 3175 | } |
| 3176 | } |
| 3177 | d->skipped_entity_in_content = false; |
| 3178 | return d->parseBeginOrContinue(state: 0, incremental); |
| 3179 | } |
| 3180 | |
| 3181 | /*! |
| 3182 | Continues incremental parsing, taking input from the |
| 3183 | QXmlInputSource that was specified with the most recent |
| 3184 | call to parse(). To use this function, you \e must have called |
| 3185 | parse() with the incremental argument set to true. |
| 3186 | |
| 3187 | Returns \c false if a parsing error occurs; otherwise returns \c true, |
| 3188 | even if the end of the XML file has not been reached. You can |
| 3189 | continue parsing at a later stage by calling this function again |
| 3190 | when there is more data available to parse. |
| 3191 | |
| 3192 | Calling this function when there is no data available in the input |
| 3193 | source indicates to the reader that the end of the XML file has |
| 3194 | been reached. If the input supplied up to this point was |
| 3195 | not well-formed then a parsing error occurs, and false is returned. |
| 3196 | If the input supplied was well-formed, true is returned. |
| 3197 | It is important to end the input in this way because it allows you |
| 3198 | to reuse the reader to parse other XML files. |
| 3199 | |
| 3200 | Calling this function after the end of file has been reached, but |
| 3201 | without available data will cause false to be returned whether the |
| 3202 | previous input was well-formed or not. |
| 3203 | |
| 3204 | \sa parse(), QXmlInputSource::data(), QXmlInputSource::next() |
| 3205 | */ |
| 3206 | bool QXmlSimpleReader::parseContinue() |
| 3207 | { |
| 3208 | Q_D(QXmlSimpleReader); |
| 3209 | if (d->parseStack == nullptr || d->parseStack->isEmpty()) |
| 3210 | return false; |
| 3211 | d->initData(); |
| 3212 | int state = d->parseStack->pop().state; |
| 3213 | return d->parseBeginOrContinue(state, incremental: true); |
| 3214 | } |
| 3215 | |
| 3216 | /* |
| 3217 | Common part of parse() and parseContinue() |
| 3218 | */ |
| 3219 | bool QXmlSimpleReaderPrivate::parseBeginOrContinue(int state, bool incremental) |
| 3220 | { |
| 3221 | bool atEndOrig = atEnd(); |
| 3222 | |
| 3223 | if (state==0) { |
| 3224 | if (!parseProlog()) { |
| 3225 | if (incremental && error.isNull()) { |
| 3226 | pushParseState(function: nullptr, state: 0); |
| 3227 | return true; |
| 3228 | } else { |
| 3229 | clear(c&: tags); |
| 3230 | return false; |
| 3231 | } |
| 3232 | } |
| 3233 | state = 1; |
| 3234 | } |
| 3235 | if (state==1) { |
| 3236 | if (!parseElement()) { |
| 3237 | if (incremental && error.isNull()) { |
| 3238 | pushParseState(function: nullptr, state: 1); |
| 3239 | return true; |
| 3240 | } else { |
| 3241 | clear(c&: tags); |
| 3242 | return false; |
| 3243 | } |
| 3244 | } |
| 3245 | state = 2; |
| 3246 | } |
| 3247 | // parse Misc* |
| 3248 | while (!atEnd()) { |
| 3249 | if (!parseMisc()) { |
| 3250 | if (incremental && error.isNull()) { |
| 3251 | pushParseState(function: nullptr, state: 2); |
| 3252 | return true; |
| 3253 | } else { |
| 3254 | clear(c&: tags); |
| 3255 | return false; |
| 3256 | } |
| 3257 | } |
| 3258 | } |
| 3259 | if (!atEndOrig && incremental) { |
| 3260 | // we parsed something at all, so be prepared to come back later |
| 3261 | pushParseState(function: nullptr, state: 2); |
| 3262 | return true; |
| 3263 | } |
| 3264 | // is stack empty? |
| 3265 | if (!tags.empty() && !error.isNull()) { |
| 3266 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDEOF)); |
| 3267 | clear(c&: tags); |
| 3268 | return false; |
| 3269 | } |
| 3270 | // call the handler |
| 3271 | if (contentHnd) { |
| 3272 | delete parseStack; |
| 3273 | parseStack = nullptr; |
| 3274 | if (!contentHnd->endDocument()) { |
| 3275 | reportParseError(error: contentHnd->errorString()); |
| 3276 | return false; |
| 3277 | } |
| 3278 | } |
| 3279 | return true; |
| 3280 | } |
| 3281 | |
| 3282 | // |
| 3283 | // The following private parse functions have another semantics for the return |
| 3284 | // value: They return true iff parsing has finished successfully (i.e. the end |
| 3285 | // of the XML file must be reached!). If one of these functions return false, |
| 3286 | // there is only an error when d->error.isNULL() is also false. |
| 3287 | // |
| 3288 | |
| 3289 | /* |
| 3290 | For the incremental parsing, it is very important that the parse...() |
| 3291 | functions have a certain structure. Since it might be hard to understand how |
| 3292 | they work, here is a description of the layout of these functions: |
| 3293 | |
| 3294 | bool QXmlSimpleReader::parse...() |
| 3295 | { |
| 3296 | (1) const signed char Init = 0; |
| 3297 | ... |
| 3298 | |
| 3299 | (2) const signed char Inp... = 0; |
| 3300 | ... |
| 3301 | |
| 3302 | (3) static const signed char table[3][2] = { |
| 3303 | ... |
| 3304 | }; |
| 3305 | signed char state; |
| 3306 | signed char input; |
| 3307 | |
| 3308 | (4) if (d->parseStack == nullptr || d->parseStack->isEmpty()) { |
| 3309 | (4a) ... |
| 3310 | } else { |
| 3311 | (4b) ... |
| 3312 | } |
| 3313 | |
| 3314 | for (; ;) { |
| 3315 | (5) switch (state) { |
| 3316 | ... |
| 3317 | } |
| 3318 | |
| 3319 | (6) |
| 3320 | (6a) if (atEnd()) { |
| 3321 | unexpectedEof(&QXmlSimpleReader::parseNmtoken, state); |
| 3322 | return false; |
| 3323 | } |
| 3324 | (6b) if (determineNameChar(c) != NotName) { |
| 3325 | ... |
| 3326 | } |
| 3327 | (7) state = table[state][input]; |
| 3328 | |
| 3329 | (8) switch (state) { |
| 3330 | ... |
| 3331 | } |
| 3332 | } |
| 3333 | } |
| 3334 | |
| 3335 | Explanation: |
| 3336 | ad 1: constants for the states (used in the transition table) |
| 3337 | ad 2: constants for the input (used in the transition table) |
| 3338 | ad 3: the transition table for the state machine |
| 3339 | ad 4: test if we are in a parseContinue() step |
| 3340 | a) if no, do initializations |
| 3341 | b) if yes, restore the state and call parse functions recursively |
| 3342 | ad 5: Do some actions according to the state; from the logical execution |
| 3343 | order, this code belongs after 8 (see there for an explanation) |
| 3344 | ad 6: Check the character that is at the actual "cursor" position: |
| 3345 | a) If we reached the EOF, report either error or push the state (in the |
| 3346 | case of incremental parsing). |
| 3347 | b) Otherwise, set the input character constant for the transition |
| 3348 | table. |
| 3349 | ad 7: Get the new state according to the input that was read. |
| 3350 | ad 8: Do some actions according to the state. The last line in every case |
| 3351 | statement reads new data (i.e. it move the cursor). This can also be |
| 3352 | done by calling another parse...() function. If you need processing for |
| 3353 | this state after that, you have to put it into the switch statement 5. |
| 3354 | This ensures that you have a well defined re-entry point, when you ran |
| 3355 | out of data. |
| 3356 | */ |
| 3357 | |
| 3358 | /* |
| 3359 | Parses the prolog [22]. |
| 3360 | */ |
| 3361 | |
| 3362 | bool QXmlSimpleReaderPrivate::parseProlog() |
| 3363 | { |
| 3364 | const signed char Init = 0; |
| 3365 | const signed char EatWS = 1; // eat white spaces |
| 3366 | const signed char Lt = 2; // '<' read |
| 3367 | const signed char Em = 3; // '!' read |
| 3368 | const signed char DocType = 4; // read doctype |
| 3369 | const signed char = 5; // read comment |
| 3370 | const signed char = 6; // same as Comment, but already reported |
| 3371 | const signed char PInstr = 7; // read PI |
| 3372 | const signed char PInstrR = 8; // same as PInstr, but already reported |
| 3373 | const signed char Done = 9; |
| 3374 | |
| 3375 | const signed char InpWs = 0; |
| 3376 | const signed char InpLt = 1; // < |
| 3377 | const signed char InpQm = 2; // ? |
| 3378 | const signed char InpEm = 3; // ! |
| 3379 | const signed char InpD = 4; // D |
| 3380 | const signed char InpDash = 5; // - |
| 3381 | const signed char InpUnknown = 6; |
| 3382 | |
| 3383 | static const signed char table[9][7] = { |
| 3384 | /* InpWs InpLt InpQm InpEm InpD InpDash InpUnknown */ |
| 3385 | { EatWS, Lt, -1, -1, -1, -1, -1 }, // Init |
| 3386 | { -1, Lt, -1, -1, -1, -1, -1 }, // EatWS |
| 3387 | { -1, -1, PInstr,Em, Done, -1, Done }, // Lt |
| 3388 | { -1, -1, -1, -1, DocType, Comment, -1 }, // Em |
| 3389 | { EatWS, Lt, -1, -1, -1, -1, -1 }, // DocType |
| 3390 | { EatWS, Lt, -1, -1, -1, -1, -1 }, // Comment |
| 3391 | { EatWS, Lt, -1, -1, -1, -1, -1 }, // CommentR |
| 3392 | { EatWS, Lt, -1, -1, -1, -1, -1 }, // PInstr |
| 3393 | { EatWS, Lt, -1, -1, -1, -1, -1 } // PInstrR |
| 3394 | }; |
| 3395 | signed char state; |
| 3396 | signed char input; |
| 3397 | |
| 3398 | if (parseStack == nullptr|| parseStack->isEmpty()) { |
| 3399 | xmldecl_possible = true; |
| 3400 | doctype_read = false; |
| 3401 | state = Init; |
| 3402 | } else { |
| 3403 | state = parseStack->pop().state; |
| 3404 | #if defined(QT_QXML_DEBUG) |
| 3405 | qDebug("QXmlSimpleReader: parseProlog (cont) in state %d" , state); |
| 3406 | #endif |
| 3407 | if (!parseStack->isEmpty()) { |
| 3408 | ParseFunction function = parseStack->top().function; |
| 3409 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 3410 | parseStack->pop(); |
| 3411 | #if defined(QT_QXML_DEBUG) |
| 3412 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 3413 | #endif |
| 3414 | } |
| 3415 | if (!(this->*function)()) { |
| 3416 | parseFailed(where: &QXmlSimpleReaderPrivate::parseProlog, state); |
| 3417 | return false; |
| 3418 | } |
| 3419 | } |
| 3420 | } |
| 3421 | |
| 3422 | for (;;) { |
| 3423 | switch (state) { |
| 3424 | case DocType: |
| 3425 | if (doctype_read) { |
| 3426 | reportParseError(error: QLatin1String(XMLERR_MORETHANONEDOCTYPE)); |
| 3427 | return false; |
| 3428 | } else { |
| 3429 | doctype_read = false; |
| 3430 | } |
| 3431 | break; |
| 3432 | case Comment: |
| 3433 | if (lexicalHnd) { |
| 3434 | if (!lexicalHnd->comment(ch: string())) { |
| 3435 | reportParseError(error: lexicalHnd->errorString()); |
| 3436 | return false; |
| 3437 | } |
| 3438 | } |
| 3439 | state = CommentR; |
| 3440 | break; |
| 3441 | case PInstr: |
| 3442 | // call the handler |
| 3443 | if (contentHnd) { |
| 3444 | if (xmldecl_possible && !xmlVersion.isEmpty()) { |
| 3445 | QString value(QLatin1String("version='" )); |
| 3446 | value += xmlVersion; |
| 3447 | value += QLatin1Char('\''); |
| 3448 | if (!encoding.isEmpty()) { |
| 3449 | value += QLatin1String(" encoding='" ); |
| 3450 | value += encoding; |
| 3451 | value += QLatin1Char('\''); |
| 3452 | } |
| 3453 | if (standalone == QXmlSimpleReaderPrivate::Yes) { |
| 3454 | value += QLatin1String(" standalone='yes'" ); |
| 3455 | } else if (standalone == QXmlSimpleReaderPrivate::No) { |
| 3456 | value += QLatin1String(" standalone='no'" ); |
| 3457 | } |
| 3458 | if (!contentHnd->processingInstruction(target: QLatin1String("xml" ), data: value)) { |
| 3459 | reportParseError(error: contentHnd->errorString()); |
| 3460 | return false; |
| 3461 | } |
| 3462 | } else { |
| 3463 | if (!contentHnd->processingInstruction(target: name(), data: string())) { |
| 3464 | reportParseError(error: contentHnd->errorString()); |
| 3465 | return false; |
| 3466 | } |
| 3467 | } |
| 3468 | } |
| 3469 | // XML declaration only on first position possible |
| 3470 | xmldecl_possible = false; |
| 3471 | state = PInstrR; |
| 3472 | break; |
| 3473 | case Done: |
| 3474 | return true; |
| 3475 | case -1: |
| 3476 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGELEMENT)); |
| 3477 | return false; |
| 3478 | } |
| 3479 | |
| 3480 | if (atEnd()) { |
| 3481 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseProlog, state); |
| 3482 | return false; |
| 3483 | } |
| 3484 | if (is_S(ch: c)) { |
| 3485 | input = InpWs; |
| 3486 | } else if (c == QLatin1Char('<')) { |
| 3487 | input = InpLt; |
| 3488 | } else if (c == QLatin1Char('?')) { |
| 3489 | input = InpQm; |
| 3490 | } else if (c == QLatin1Char('!')) { |
| 3491 | input = InpEm; |
| 3492 | } else if (c == QLatin1Char('D')) { |
| 3493 | input = InpD; |
| 3494 | } else if (c == QLatin1Char('-')) { |
| 3495 | input = InpDash; |
| 3496 | } else { |
| 3497 | input = InpUnknown; |
| 3498 | } |
| 3499 | state = table[state][input]; |
| 3500 | |
| 3501 | switch (state) { |
| 3502 | case EatWS: |
| 3503 | // XML declaration only on first position possible |
| 3504 | xmldecl_possible = false; |
| 3505 | if (!eat_ws()) { |
| 3506 | parseFailed(where: &QXmlSimpleReaderPrivate::parseProlog, state); |
| 3507 | return false; |
| 3508 | } |
| 3509 | break; |
| 3510 | case Lt: |
| 3511 | next(); |
| 3512 | break; |
| 3513 | case Em: |
| 3514 | // XML declaration only on first position possible |
| 3515 | xmldecl_possible = false; |
| 3516 | next(); |
| 3517 | break; |
| 3518 | case DocType: |
| 3519 | if (!parseDoctype()) { |
| 3520 | parseFailed(where: &QXmlSimpleReaderPrivate::parseProlog, state); |
| 3521 | return false; |
| 3522 | } |
| 3523 | break; |
| 3524 | case Comment: |
| 3525 | case CommentR: |
| 3526 | if (!parseComment()) { |
| 3527 | parseFailed(where: &QXmlSimpleReaderPrivate::parseProlog, state); |
| 3528 | return false; |
| 3529 | } |
| 3530 | break; |
| 3531 | case PInstr: |
| 3532 | case PInstrR: |
| 3533 | parsePI_xmldecl = xmldecl_possible; |
| 3534 | if (!parsePI()) { |
| 3535 | parseFailed(where: &QXmlSimpleReaderPrivate::parseProlog, state); |
| 3536 | return false; |
| 3537 | } |
| 3538 | break; |
| 3539 | } |
| 3540 | } |
| 3541 | return false; |
| 3542 | } |
| 3543 | |
| 3544 | /* |
| 3545 | Parse an element [39]. |
| 3546 | |
| 3547 | Precondition: the opening '<' is already read. |
| 3548 | */ |
| 3549 | bool QXmlSimpleReaderPrivate::parseElement() |
| 3550 | { |
| 3551 | const int Init = 0; |
| 3552 | const int ReadName = 1; |
| 3553 | const int Ws1 = 2; |
| 3554 | const int STagEnd = 3; |
| 3555 | const int STagEnd2 = 4; |
| 3556 | const int ETagBegin = 5; |
| 3557 | const int ETagBegin2 = 6; |
| 3558 | const int Ws2 = 7; |
| 3559 | const int EmptyTag = 8; |
| 3560 | const int Attrib = 9; |
| 3561 | const int AttribPro = 10; // like Attrib, but processAttribute was already called |
| 3562 | const int Ws3 = 11; |
| 3563 | const int Done = 12; |
| 3564 | |
| 3565 | const int InpWs = 0; // whitespace |
| 3566 | const int InpNameBe = 1; // NameBeginning |
| 3567 | const int InpGt = 2; // > |
| 3568 | const int InpSlash = 3; // / |
| 3569 | const int InpUnknown = 4; |
| 3570 | |
| 3571 | static const int table[12][5] = { |
| 3572 | /* InpWs InpNameBe InpGt InpSlash InpUnknown */ |
| 3573 | { -1, ReadName, -1, -1, -1 }, // Init |
| 3574 | { Ws1, Attrib, STagEnd, EmptyTag, -1 }, // ReadName |
| 3575 | { -1, Attrib, STagEnd, EmptyTag, -1 }, // Ws1 |
| 3576 | { STagEnd2, STagEnd2, STagEnd2, STagEnd2, STagEnd2 }, // STagEnd |
| 3577 | { -1, -1, -1, ETagBegin, -1 }, // STagEnd2 |
| 3578 | { -1, ETagBegin2, -1, -1, -1 }, // ETagBegin |
| 3579 | { Ws2, -1, Done, -1, -1 }, // ETagBegin2 |
| 3580 | { -1, -1, Done, -1, -1 }, // Ws2 |
| 3581 | { -1, -1, Done, -1, -1 }, // EmptyTag |
| 3582 | { Ws3, Attrib, STagEnd, EmptyTag, -1 }, // Attrib |
| 3583 | { Ws3, Attrib, STagEnd, EmptyTag, -1 }, // AttribPro |
| 3584 | { -1, Attrib, STagEnd, EmptyTag, -1 } // Ws3 |
| 3585 | }; |
| 3586 | int state; |
| 3587 | int input; |
| 3588 | |
| 3589 | if (parseStack == nullptr|| parseStack->isEmpty()) { |
| 3590 | state = Init; |
| 3591 | } else { |
| 3592 | state = parseStack->pop().state; |
| 3593 | #if defined(QT_QXML_DEBUG) |
| 3594 | qDebug("QXmlSimpleReader: parseElement (cont) in state %d" , state); |
| 3595 | #endif |
| 3596 | if (!parseStack->isEmpty()) { |
| 3597 | ParseFunction function = parseStack->top().function; |
| 3598 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 3599 | parseStack->pop(); |
| 3600 | #if defined(QT_QXML_DEBUG) |
| 3601 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 3602 | #endif |
| 3603 | } |
| 3604 | if (!(this->*function)()) { |
| 3605 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3606 | return false; |
| 3607 | } |
| 3608 | } |
| 3609 | } |
| 3610 | |
| 3611 | for (;;) { |
| 3612 | switch (state) { |
| 3613 | case ReadName: |
| 3614 | // store it on the stack |
| 3615 | tags.push(x: name()); |
| 3616 | // empty the attributes |
| 3617 | attList.clear(); |
| 3618 | if (useNamespaces) |
| 3619 | namespaceSupport.pushContext(); |
| 3620 | break; |
| 3621 | case ETagBegin2: |
| 3622 | if (!processElementETagBegin2()) |
| 3623 | return false; |
| 3624 | break; |
| 3625 | case Attrib: |
| 3626 | if (!processElementAttribute()) |
| 3627 | return false; |
| 3628 | state = AttribPro; |
| 3629 | break; |
| 3630 | case Done: |
| 3631 | return true; |
| 3632 | case -1: |
| 3633 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGELEMENT)); |
| 3634 | return false; |
| 3635 | } |
| 3636 | |
| 3637 | if (atEnd()) { |
| 3638 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3639 | return false; |
| 3640 | } |
| 3641 | if (fastDetermineNameChar(ch: c) == NameBeginning) { |
| 3642 | input = InpNameBe; |
| 3643 | } else if (c == QLatin1Char('>')) { |
| 3644 | input = InpGt; |
| 3645 | } else if (is_S(ch: c)) { |
| 3646 | input = InpWs; |
| 3647 | } else if (c == QLatin1Char('/')) { |
| 3648 | input = InpSlash; |
| 3649 | } else { |
| 3650 | input = InpUnknown; |
| 3651 | } |
| 3652 | state = table[state][input]; |
| 3653 | |
| 3654 | switch (state) { |
| 3655 | case ReadName: |
| 3656 | parseName_useRef = false; |
| 3657 | if (!parseName()) { |
| 3658 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3659 | return false; |
| 3660 | } |
| 3661 | break; |
| 3662 | case Ws1: |
| 3663 | case Ws2: |
| 3664 | case Ws3: |
| 3665 | if (!eat_ws()) { |
| 3666 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3667 | return false; |
| 3668 | } |
| 3669 | break; |
| 3670 | case STagEnd: |
| 3671 | // call the handler |
| 3672 | if (contentHnd) { |
| 3673 | if (useNamespaces) { |
| 3674 | QString uri, lname; |
| 3675 | namespaceSupport.processName(qname: tags.top(), isAttribute: false, nsuri&: uri, localname&: lname); |
| 3676 | if (!contentHnd->startElement(namespaceURI: uri, localName: lname, qName: tags.top(), atts: attList)) { |
| 3677 | reportParseError(error: contentHnd->errorString()); |
| 3678 | return false; |
| 3679 | } |
| 3680 | } else { |
| 3681 | if (!contentHnd->startElement(namespaceURI: QString(), localName: QString(), qName: tags.top(), atts: attList)) { |
| 3682 | reportParseError(error: contentHnd->errorString()); |
| 3683 | return false; |
| 3684 | } |
| 3685 | } |
| 3686 | } |
| 3687 | next(); |
| 3688 | break; |
| 3689 | case STagEnd2: |
| 3690 | if (!parseContent()) { |
| 3691 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3692 | return false; |
| 3693 | } |
| 3694 | break; |
| 3695 | case ETagBegin: |
| 3696 | next(); |
| 3697 | break; |
| 3698 | case ETagBegin2: |
| 3699 | // get the name of the tag |
| 3700 | parseName_useRef = false; |
| 3701 | if (!parseName()) { |
| 3702 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3703 | return false; |
| 3704 | } |
| 3705 | break; |
| 3706 | case EmptyTag: |
| 3707 | if (tags.empty()) { |
| 3708 | reportParseError(error: QLatin1String(XMLERR_TAGMISMATCH)); |
| 3709 | return false; |
| 3710 | } |
| 3711 | if (!processElementEmptyTag()) |
| 3712 | return false; |
| 3713 | next(); |
| 3714 | break; |
| 3715 | case Attrib: |
| 3716 | case AttribPro: |
| 3717 | // get name and value of attribute |
| 3718 | if (!parseAttribute()) { |
| 3719 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElement, state); |
| 3720 | return false; |
| 3721 | } |
| 3722 | break; |
| 3723 | case Done: |
| 3724 | next(); |
| 3725 | break; |
| 3726 | } |
| 3727 | } |
| 3728 | return false; |
| 3729 | } |
| 3730 | |
| 3731 | /* |
| 3732 | Helper to break down the size of the code in the case statement. |
| 3733 | Return false on error, otherwise true. |
| 3734 | */ |
| 3735 | bool QXmlSimpleReaderPrivate::processElementEmptyTag() |
| 3736 | { |
| 3737 | QString uri, lname; |
| 3738 | // pop the stack and call the handler |
| 3739 | if (contentHnd) { |
| 3740 | if (useNamespaces) { |
| 3741 | // report startElement first... |
| 3742 | namespaceSupport.processName(qname: tags.top(), isAttribute: false, nsuri&: uri, localname&: lname); |
| 3743 | if (!contentHnd->startElement(namespaceURI: uri, localName: lname, qName: tags.top(), atts: attList)) { |
| 3744 | reportParseError(error: contentHnd->errorString()); |
| 3745 | return false; |
| 3746 | } |
| 3747 | // ... followed by endElement... |
| 3748 | const bool endElementReturnedFalse = !contentHnd->endElement(namespaceURI: uri, localName: lname, qName: tags.top()); |
| 3749 | tags.pop(); |
| 3750 | if (endElementReturnedFalse) { |
| 3751 | reportParseError(error: contentHnd->errorString()); |
| 3752 | return false; |
| 3753 | } |
| 3754 | // ... followed by endPrefixMapping |
| 3755 | QStringList prefixesBefore, prefixesAfter; |
| 3756 | if (contentHnd) { |
| 3757 | prefixesBefore = namespaceSupport.prefixes(); |
| 3758 | } |
| 3759 | namespaceSupport.popContext(); |
| 3760 | // call the handler for prefix mapping |
| 3761 | prefixesAfter = namespaceSupport.prefixes(); |
| 3762 | for (QStringList::Iterator it = prefixesBefore.begin(); it != prefixesBefore.end(); ++it) { |
| 3763 | if (!prefixesAfter.contains(str: *it)) { |
| 3764 | if (!contentHnd->endPrefixMapping(prefix: *it)) { |
| 3765 | reportParseError(error: contentHnd->errorString()); |
| 3766 | return false; |
| 3767 | } |
| 3768 | } |
| 3769 | } |
| 3770 | } else { |
| 3771 | // report startElement first... |
| 3772 | if (!contentHnd->startElement(namespaceURI: QString(), localName: QString(), qName: tags.top(), atts: attList)) { |
| 3773 | reportParseError(error: contentHnd->errorString()); |
| 3774 | return false; |
| 3775 | } |
| 3776 | // ... followed by endElement |
| 3777 | const bool endElementReturnedFalse = !contentHnd->endElement(namespaceURI: QString(), localName: QString(), qName: tags.top()); |
| 3778 | tags.pop(); |
| 3779 | if (endElementReturnedFalse) { |
| 3780 | reportParseError(error: contentHnd->errorString()); |
| 3781 | return false; |
| 3782 | } |
| 3783 | } |
| 3784 | } else { |
| 3785 | tags.pop(); |
| 3786 | namespaceSupport.popContext(); |
| 3787 | } |
| 3788 | return true; |
| 3789 | } |
| 3790 | /* |
| 3791 | Helper to break down the size of the code in the case statement. |
| 3792 | Return false on error, otherwise true. |
| 3793 | */ |
| 3794 | bool QXmlSimpleReaderPrivate::processElementETagBegin2() |
| 3795 | { |
| 3796 | const QString &name = QXmlSimpleReaderPrivate::name(); |
| 3797 | |
| 3798 | // pop the stack and compare it with the name |
| 3799 | const bool nameIsTagsTop = tags.top() == name; |
| 3800 | tags.pop(); |
| 3801 | if (!nameIsTagsTop) { |
| 3802 | reportParseError(error: QLatin1String(XMLERR_TAGMISMATCH)); |
| 3803 | return false; |
| 3804 | } |
| 3805 | // call the handler |
| 3806 | if (contentHnd) { |
| 3807 | QString uri, lname; |
| 3808 | |
| 3809 | if (useNamespaces) |
| 3810 | namespaceSupport.processName(qname: name, isAttribute: false, nsuri&: uri, localname&: lname); |
| 3811 | if (!contentHnd->endElement(namespaceURI: uri, localName: lname, qName: name)) { |
| 3812 | reportParseError(error: contentHnd->errorString()); |
| 3813 | return false; |
| 3814 | } |
| 3815 | } |
| 3816 | if (useNamespaces) { |
| 3817 | NamespaceMap prefixesBefore, prefixesAfter; |
| 3818 | if (contentHnd) |
| 3819 | prefixesBefore = namespaceSupport.d->ns; |
| 3820 | |
| 3821 | namespaceSupport.popContext(); |
| 3822 | // call the handler for prefix mapping |
| 3823 | if (contentHnd) { |
| 3824 | prefixesAfter = namespaceSupport.d->ns; |
| 3825 | if (prefixesBefore.size() != prefixesAfter.size()) { |
| 3826 | for (NamespaceMap::const_iterator it = prefixesBefore.constBegin(); it != prefixesBefore.constEnd(); ++it) { |
| 3827 | if (!it.key().isEmpty() && !prefixesAfter.contains(key: it.key())) { |
| 3828 | if (!contentHnd->endPrefixMapping(prefix: it.key())) { |
| 3829 | reportParseError(error: contentHnd->errorString()); |
| 3830 | return false; |
| 3831 | } |
| 3832 | } |
| 3833 | } |
| 3834 | } |
| 3835 | } |
| 3836 | } |
| 3837 | return true; |
| 3838 | } |
| 3839 | /* |
| 3840 | Helper to break down the size of the code in the case statement. |
| 3841 | Return false on error, otherwise true. |
| 3842 | */ |
| 3843 | bool QXmlSimpleReaderPrivate::processElementAttribute() |
| 3844 | { |
| 3845 | QString uri, lname, prefix; |
| 3846 | const QString &name = QXmlSimpleReaderPrivate::name(); |
| 3847 | const QString &string = QXmlSimpleReaderPrivate::string(); |
| 3848 | |
| 3849 | // add the attribute to the list |
| 3850 | if (useNamespaces) { |
| 3851 | // is it a namespace declaration? |
| 3852 | namespaceSupport.splitName(qname: name, prefix, localname&: lname); |
| 3853 | if (prefix == QLatin1String("xmlns" )) { |
| 3854 | // namespace declaration |
| 3855 | namespaceSupport.setPrefix(pre: lname, uri: string); |
| 3856 | if (useNamespacePrefixes) { |
| 3857 | // according to http://www.w3.org/2000/xmlns/, the "prefix" |
| 3858 | // xmlns maps to the namespace name |
| 3859 | // http://www.w3.org/2000/xmlns/ |
| 3860 | attList.append(qName: name, uri: QLatin1String("http://www.w3.org/2000/xmlns/" ), localPart: lname, value: string); |
| 3861 | } |
| 3862 | // call the handler for prefix mapping |
| 3863 | if (contentHnd) { |
| 3864 | if (!contentHnd->startPrefixMapping(prefix: lname, uri: string)) { |
| 3865 | reportParseError(error: contentHnd->errorString()); |
| 3866 | return false; |
| 3867 | } |
| 3868 | } |
| 3869 | } else { |
| 3870 | // no namespace delcaration |
| 3871 | namespaceSupport.processName(qname: name, isAttribute: true, nsuri&: uri, localname&: lname); |
| 3872 | attList.append(qName: name, uri, localPart: lname, value: string); |
| 3873 | } |
| 3874 | } else { |
| 3875 | // no namespace support |
| 3876 | attList.append(qName: name, uri, localPart: lname, value: string); |
| 3877 | } |
| 3878 | return true; |
| 3879 | } |
| 3880 | |
| 3881 | /* |
| 3882 | Parse a content [43]. |
| 3883 | |
| 3884 | A content is only used between tags. If a end tag is found the < is already |
| 3885 | read and the head stand on the '/' of the end tag '</name>'. |
| 3886 | */ |
| 3887 | bool QXmlSimpleReaderPrivate::parseContent() |
| 3888 | { |
| 3889 | const signed char Init = 0; |
| 3890 | const signed char ChD = 1; // CharData |
| 3891 | const signed char ChD1 = 2; // CharData help state |
| 3892 | const signed char ChD2 = 3; // CharData help state |
| 3893 | const signed char Ref = 4; // Reference |
| 3894 | const signed char Lt = 5; // '<' read |
| 3895 | const signed char PInstr = 6; // PI |
| 3896 | const signed char PInstrR = 7; // same as PInstr, but already reported |
| 3897 | const signed char Elem = 8; // Element |
| 3898 | const signed char Em = 9; // '!' read |
| 3899 | const signed char Com = 10; // Comment |
| 3900 | const signed char ComR = 11; // same as Com, but already reported |
| 3901 | const signed char CDS = 12; // CDSect |
| 3902 | const signed char CDS1 = 13; // read a CDSect |
| 3903 | const signed char CDS2 = 14; // read a CDSect (help state) |
| 3904 | const signed char CDS3 = 15; // read a CDSect (help state) |
| 3905 | const signed char Done = 16; // finished reading content |
| 3906 | |
| 3907 | const signed char InpLt = 0; // < |
| 3908 | const signed char InpGt = 1; // > |
| 3909 | const signed char InpSlash = 2; // / |
| 3910 | const signed char InpQMark = 3; // ? |
| 3911 | const signed char InpEMark = 4; // ! |
| 3912 | const signed char InpAmp = 5; // & |
| 3913 | const signed char InpDash = 6; // - |
| 3914 | const signed char InpOpenB = 7; // [ |
| 3915 | const signed char InpCloseB = 8; //] |
| 3916 | const signed char InpUnknown = 9; |
| 3917 | |
| 3918 | static const signed char mapCLT2FSMChar[] = { |
| 3919 | InpUnknown, // white space |
| 3920 | InpUnknown, // % |
| 3921 | InpAmp, // & |
| 3922 | InpGt, // > |
| 3923 | InpLt, // < |
| 3924 | InpSlash, // / |
| 3925 | InpQMark, // ? |
| 3926 | InpEMark, // ! |
| 3927 | InpDash, // - |
| 3928 | InpCloseB, //] |
| 3929 | InpOpenB, // [ |
| 3930 | InpUnknown, // = |
| 3931 | InpUnknown, // " |
| 3932 | InpUnknown, // ' |
| 3933 | InpUnknown // unknown |
| 3934 | }; |
| 3935 | |
| 3936 | static const signed char table[16][10] = { |
| 3937 | /* InpLt InpGt InpSlash InpQMark InpEMark InpAmp InpDash InpOpenB InpCloseB InpUnknown */ |
| 3938 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD1, ChD }, // Init |
| 3939 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD1, ChD }, // ChD |
| 3940 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD2, ChD }, // ChD1 |
| 3941 | { Lt, -1, ChD, ChD, ChD, Ref, ChD, ChD, ChD2, ChD }, // ChD2 |
| 3942 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD, ChD }, // Ref (same as Init) |
| 3943 | { -1, -1, Done, PInstr, Em, -1, -1, -1, -1, Elem }, // Lt |
| 3944 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD, ChD }, // PInstr (same as Init) |
| 3945 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD, ChD }, // PInstrR |
| 3946 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD, ChD }, // Elem (same as Init) |
| 3947 | { -1, -1, -1, -1, -1, -1, Com, CDS, -1, -1 }, // Em |
| 3948 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD, ChD }, // Com (same as Init) |
| 3949 | { Lt, ChD, ChD, ChD, ChD, Ref, ChD, ChD, ChD, ChD }, // ComR |
| 3950 | { CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS2, CDS1 }, // CDS |
| 3951 | { CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS2, CDS1 }, // CDS1 |
| 3952 | { CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS3, CDS1 }, // CDS2 |
| 3953 | { CDS1, Init, CDS1, CDS1, CDS1, CDS1, CDS1, CDS1, CDS3, CDS1 } // CDS3 |
| 3954 | }; |
| 3955 | signed char state; |
| 3956 | signed char input; |
| 3957 | |
| 3958 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 3959 | contentCharDataRead = false; |
| 3960 | state = Init; |
| 3961 | } else { |
| 3962 | state = parseStack->pop().state; |
| 3963 | #if defined(QT_QXML_DEBUG) |
| 3964 | qDebug("QXmlSimpleReader: parseContent (cont) in state %d" , state); |
| 3965 | #endif |
| 3966 | if (!parseStack->isEmpty()) { |
| 3967 | ParseFunction function = parseStack->top().function; |
| 3968 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 3969 | parseStack->pop(); |
| 3970 | #if defined(QT_QXML_DEBUG) |
| 3971 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 3972 | #endif |
| 3973 | } |
| 3974 | if (!(this->*function)()) { |
| 3975 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 3976 | return false; |
| 3977 | } |
| 3978 | } |
| 3979 | } |
| 3980 | |
| 3981 | for (;;) { |
| 3982 | switch (state) { |
| 3983 | case Ref: |
| 3984 | if (!contentCharDataRead) |
| 3985 | contentCharDataRead = parseReference_charDataRead; |
| 3986 | break; |
| 3987 | case PInstr: |
| 3988 | if (contentHnd) { |
| 3989 | if (!contentHnd->processingInstruction(target: name(),data: string())) { |
| 3990 | reportParseError(error: contentHnd->errorString()); |
| 3991 | return false; |
| 3992 | } |
| 3993 | } |
| 3994 | state = PInstrR; |
| 3995 | break; |
| 3996 | case Com: |
| 3997 | if (lexicalHnd) { |
| 3998 | if (!lexicalHnd->comment(ch: string())) { |
| 3999 | reportParseError(error: lexicalHnd->errorString()); |
| 4000 | return false; |
| 4001 | } |
| 4002 | } |
| 4003 | state = ComR; |
| 4004 | break; |
| 4005 | case CDS: |
| 4006 | stringClear(); |
| 4007 | break; |
| 4008 | case CDS2: |
| 4009 | if (!atEnd() && c != QLatin1Char(']')) |
| 4010 | stringAddC(QLatin1Char(']')); |
| 4011 | break; |
| 4012 | case CDS3: |
| 4013 | // test if this skipping was legal |
| 4014 | if (!atEnd()) { |
| 4015 | if (c == QLatin1Char('>')) { |
| 4016 | // the end of the CDSect |
| 4017 | if (lexicalHnd) { |
| 4018 | if (!lexicalHnd->startCDATA()) { |
| 4019 | reportParseError(error: lexicalHnd->errorString()); |
| 4020 | return false; |
| 4021 | } |
| 4022 | } |
| 4023 | if (contentHnd) { |
| 4024 | if (!contentHnd->characters(ch: string())) { |
| 4025 | reportParseError(error: contentHnd->errorString()); |
| 4026 | return false; |
| 4027 | } |
| 4028 | } |
| 4029 | if (lexicalHnd) { |
| 4030 | if (!lexicalHnd->endCDATA()) { |
| 4031 | reportParseError(error: lexicalHnd->errorString()); |
| 4032 | return false; |
| 4033 | } |
| 4034 | } |
| 4035 | } else if (c == QLatin1Char(']')) { |
| 4036 | // three or more ']' |
| 4037 | stringAddC(QLatin1Char(']')); |
| 4038 | } else { |
| 4039 | // after ']]' comes another character |
| 4040 | stringAddC(QLatin1Char(']')); |
| 4041 | stringAddC(QLatin1Char(']')); |
| 4042 | } |
| 4043 | } |
| 4044 | break; |
| 4045 | case Done: |
| 4046 | // call the handler for CharData |
| 4047 | if (contentHnd) { |
| 4048 | if (contentCharDataRead) { |
| 4049 | if (reportWhitespaceCharData || !string().simplified().isEmpty()) { |
| 4050 | if (!contentHnd->characters(ch: string())) { |
| 4051 | reportParseError(error: contentHnd->errorString()); |
| 4052 | return false; |
| 4053 | } |
| 4054 | } |
| 4055 | } |
| 4056 | } |
| 4057 | // Done |
| 4058 | return true; |
| 4059 | case -1: |
| 4060 | // Error |
| 4061 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGCONTENT)); |
| 4062 | return false; |
| 4063 | } |
| 4064 | |
| 4065 | // get input (use lookup-table instead of nested ifs for performance |
| 4066 | // reasons) |
| 4067 | if (atEnd()) { |
| 4068 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4069 | return false; |
| 4070 | } |
| 4071 | if (c.row()) { |
| 4072 | input = InpUnknown; |
| 4073 | } else { |
| 4074 | input = mapCLT2FSMChar[charLookupTable[c.cell()]]; |
| 4075 | } |
| 4076 | state = table[state][input]; |
| 4077 | |
| 4078 | switch (state) { |
| 4079 | case Init: |
| 4080 | // skip the ending '>' of a CDATASection |
| 4081 | next(); |
| 4082 | break; |
| 4083 | case ChD: |
| 4084 | // on first call: clear string |
| 4085 | if (!contentCharDataRead) { |
| 4086 | contentCharDataRead = true; |
| 4087 | stringClear(); |
| 4088 | } |
| 4089 | stringAddC(); |
| 4090 | if (reportEntities) { |
| 4091 | if (!reportEndEntities()) |
| 4092 | return false; |
| 4093 | } |
| 4094 | next(); |
| 4095 | break; |
| 4096 | case ChD1: |
| 4097 | // on first call: clear string |
| 4098 | if (!contentCharDataRead) { |
| 4099 | contentCharDataRead = true; |
| 4100 | stringClear(); |
| 4101 | } |
| 4102 | stringAddC(); |
| 4103 | if (reportEntities) { |
| 4104 | if (!reportEndEntities()) |
| 4105 | return false; |
| 4106 | } |
| 4107 | next(); |
| 4108 | break; |
| 4109 | case ChD2: |
| 4110 | stringAddC(); |
| 4111 | if (reportEntities) { |
| 4112 | if (!reportEndEntities()) |
| 4113 | return false; |
| 4114 | } |
| 4115 | next(); |
| 4116 | break; |
| 4117 | case Ref: |
| 4118 | if (!contentCharDataRead) { |
| 4119 | // reference may be CharData; so clear string to be safe |
| 4120 | stringClear(); |
| 4121 | parseReference_context = InContent; |
| 4122 | if (!parseReference()) { |
| 4123 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4124 | return false; |
| 4125 | } |
| 4126 | } else { |
| 4127 | if (reportEntities) { |
| 4128 | // report character data in chunks |
| 4129 | if (contentHnd) { |
| 4130 | if (reportWhitespaceCharData || !string().simplified().isEmpty()) { |
| 4131 | if (!contentHnd->characters(ch: string())) { |
| 4132 | reportParseError(error: contentHnd->errorString()); |
| 4133 | return false; |
| 4134 | } |
| 4135 | } |
| 4136 | } |
| 4137 | stringClear(); |
| 4138 | } |
| 4139 | parseReference_context = InContent; |
| 4140 | if (!parseReference()) { |
| 4141 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4142 | return false; |
| 4143 | } |
| 4144 | } |
| 4145 | break; |
| 4146 | case Lt: |
| 4147 | // call the handler for CharData |
| 4148 | if (contentHnd) { |
| 4149 | if (contentCharDataRead) { |
| 4150 | if (reportWhitespaceCharData || !string().simplified().isEmpty()) { |
| 4151 | if (!contentHnd->characters(ch: string())) { |
| 4152 | reportParseError(error: contentHnd->errorString()); |
| 4153 | return false; |
| 4154 | } |
| 4155 | } |
| 4156 | } |
| 4157 | } |
| 4158 | contentCharDataRead = false; |
| 4159 | next(); |
| 4160 | break; |
| 4161 | case PInstr: |
| 4162 | case PInstrR: |
| 4163 | parsePI_xmldecl = false; |
| 4164 | if (!parsePI()) { |
| 4165 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4166 | return false; |
| 4167 | } |
| 4168 | break; |
| 4169 | case Elem: |
| 4170 | if (!parseElement()) { |
| 4171 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4172 | return false; |
| 4173 | } |
| 4174 | break; |
| 4175 | case Em: |
| 4176 | next(); |
| 4177 | break; |
| 4178 | case Com: |
| 4179 | case ComR: |
| 4180 | if (!parseComment()) { |
| 4181 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4182 | return false; |
| 4183 | } |
| 4184 | break; |
| 4185 | case CDS: |
| 4186 | parseString_s = QLatin1String("[CDATA[" ); |
| 4187 | if (!parseString()) { |
| 4188 | parseFailed(where: &QXmlSimpleReaderPrivate::parseContent, state); |
| 4189 | return false; |
| 4190 | } |
| 4191 | break; |
| 4192 | case CDS1: |
| 4193 | stringAddC(); |
| 4194 | next(); |
| 4195 | break; |
| 4196 | case CDS2: |
| 4197 | // skip ']' |
| 4198 | next(); |
| 4199 | break; |
| 4200 | case CDS3: |
| 4201 | // skip ']'... |
| 4202 | next(); |
| 4203 | break; |
| 4204 | } |
| 4205 | } |
| 4206 | return false; |
| 4207 | } |
| 4208 | |
| 4209 | bool QXmlSimpleReaderPrivate::reportEndEntities() |
| 4210 | { |
| 4211 | int count = (int)xmlRefStack.size(); |
| 4212 | while (count != 0 && xmlRefStack.top().isEmpty()) { |
| 4213 | if (contentHnd) { |
| 4214 | if (reportWhitespaceCharData || !string().simplified().isEmpty()) { |
| 4215 | if (!contentHnd->characters(ch: string())) { |
| 4216 | reportParseError(error: contentHnd->errorString()); |
| 4217 | return false; |
| 4218 | } |
| 4219 | } |
| 4220 | } |
| 4221 | stringClear(); |
| 4222 | if (lexicalHnd) { |
| 4223 | if (!lexicalHnd->endEntity(name: xmlRefStack.top().name)) { |
| 4224 | reportParseError(error: lexicalHnd->errorString()); |
| 4225 | return false; |
| 4226 | } |
| 4227 | } |
| 4228 | xmlRefStack.pop_back(); |
| 4229 | count--; |
| 4230 | } |
| 4231 | return true; |
| 4232 | } |
| 4233 | |
| 4234 | /* |
| 4235 | Parse Misc [27]. |
| 4236 | */ |
| 4237 | bool QXmlSimpleReaderPrivate::parseMisc() |
| 4238 | { |
| 4239 | const signed char Init = 0; |
| 4240 | const signed char Lt = 1; // '<' was read |
| 4241 | const signed char = 2; // read comment |
| 4242 | const signed char eatWS = 3; // eat whitespaces |
| 4243 | const signed char PInstr = 4; // read PI |
| 4244 | const signed char = 5; // read comment |
| 4245 | |
| 4246 | const signed char InpWs = 0; // S |
| 4247 | const signed char InpLt = 1; // < |
| 4248 | const signed char InpQm = 2; // ? |
| 4249 | const signed char InpEm = 3; // ! |
| 4250 | const signed char InpUnknown = 4; |
| 4251 | |
| 4252 | static const signed char table[3][5] = { |
| 4253 | /* InpWs InpLt InpQm InpEm InpUnknown */ |
| 4254 | { eatWS, Lt, -1, -1, -1 }, // Init |
| 4255 | { -1, -1, PInstr,Comment, -1 }, // Lt |
| 4256 | { -1, -1, -1, -1, Comment2 } // Comment |
| 4257 | }; |
| 4258 | signed char state; |
| 4259 | signed char input; |
| 4260 | |
| 4261 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 4262 | state = Init; |
| 4263 | } else { |
| 4264 | state = parseStack->pop().state; |
| 4265 | #if defined(QT_QXML_DEBUG) |
| 4266 | qDebug("QXmlSimpleReader: parseMisc (cont) in state %d" , state); |
| 4267 | #endif |
| 4268 | if (!parseStack->isEmpty()) { |
| 4269 | ParseFunction function = parseStack->top().function; |
| 4270 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 4271 | parseStack->pop(); |
| 4272 | #if defined(QT_QXML_DEBUG) |
| 4273 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 4274 | #endif |
| 4275 | } |
| 4276 | if (!(this->*function)()) { |
| 4277 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMisc, state); |
| 4278 | return false; |
| 4279 | } |
| 4280 | } |
| 4281 | } |
| 4282 | |
| 4283 | for (;;) { |
| 4284 | switch (state) { |
| 4285 | case eatWS: |
| 4286 | return true; |
| 4287 | case PInstr: |
| 4288 | if (contentHnd) { |
| 4289 | if (!contentHnd->processingInstruction(target: name(),data: string())) { |
| 4290 | reportParseError(error: contentHnd->errorString()); |
| 4291 | return false; |
| 4292 | } |
| 4293 | } |
| 4294 | return true; |
| 4295 | case Comment2: |
| 4296 | if (lexicalHnd) { |
| 4297 | if (!lexicalHnd->comment(ch: string())) { |
| 4298 | reportParseError(error: lexicalHnd->errorString()); |
| 4299 | return false; |
| 4300 | } |
| 4301 | } |
| 4302 | return true; |
| 4303 | case -1: |
| 4304 | // Error |
| 4305 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 4306 | return false; |
| 4307 | } |
| 4308 | |
| 4309 | if (atEnd()) { |
| 4310 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseMisc, state); |
| 4311 | return false; |
| 4312 | } |
| 4313 | if (is_S(ch: c)) { |
| 4314 | input = InpWs; |
| 4315 | } else if (c == QLatin1Char('<')) { |
| 4316 | input = InpLt; |
| 4317 | } else if (c == QLatin1Char('?')) { |
| 4318 | input = InpQm; |
| 4319 | } else if (c == QLatin1Char('!')) { |
| 4320 | input = InpEm; |
| 4321 | } else { |
| 4322 | input = InpUnknown; |
| 4323 | } |
| 4324 | state = table[state][input]; |
| 4325 | |
| 4326 | switch (state) { |
| 4327 | case eatWS: |
| 4328 | if (!eat_ws()) { |
| 4329 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMisc, state); |
| 4330 | return false; |
| 4331 | } |
| 4332 | break; |
| 4333 | case Lt: |
| 4334 | next(); |
| 4335 | break; |
| 4336 | case PInstr: |
| 4337 | parsePI_xmldecl = false; |
| 4338 | if (!parsePI()) { |
| 4339 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMisc, state); |
| 4340 | return false; |
| 4341 | } |
| 4342 | break; |
| 4343 | case Comment: |
| 4344 | next(); |
| 4345 | break; |
| 4346 | case Comment2: |
| 4347 | if (!parseComment()) { |
| 4348 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMisc, state); |
| 4349 | return false; |
| 4350 | } |
| 4351 | break; |
| 4352 | } |
| 4353 | } |
| 4354 | return false; |
| 4355 | } |
| 4356 | |
| 4357 | /* |
| 4358 | Parse a processing instruction [16]. |
| 4359 | |
| 4360 | If xmldec is true, it tries to parse a PI or a XML declaration [23]. |
| 4361 | |
| 4362 | Precondition: the beginning '<' of the PI is already read and the head stand |
| 4363 | on the '?' of '<?'. |
| 4364 | |
| 4365 | If this funktion was successful, the head-position is on the first |
| 4366 | character after the PI. |
| 4367 | */ |
| 4368 | bool QXmlSimpleReaderPrivate::parsePI() |
| 4369 | { |
| 4370 | const signed char Init = 0; |
| 4371 | const signed char QmI = 1; // ? was read |
| 4372 | const signed char Name = 2; // read Name |
| 4373 | const signed char XMLDecl = 3; // read XMLDecl |
| 4374 | const signed char Ws1 = 4; // eat ws after "xml" of XMLDecl |
| 4375 | const signed char PInstr = 5; // read PI |
| 4376 | const signed char Ws2 = 6; // eat ws after Name of PI |
| 4377 | const signed char Version = 7; // read versionInfo |
| 4378 | const signed char Ws3 = 8; // eat ws after versionInfo |
| 4379 | const signed char EorSD = 9; // read EDecl or SDDecl |
| 4380 | const signed char Ws4 = 10; // eat ws after EDecl or SDDecl |
| 4381 | const signed char SD = 11; // read SDDecl |
| 4382 | const signed char Ws5 = 12; // eat ws after SDDecl |
| 4383 | const signed char ADone = 13; // almost done |
| 4384 | const signed char Char = 14; // Char was read |
| 4385 | const signed char Qm = 15; // Qm was read |
| 4386 | const signed char Done = 16; // finished reading content |
| 4387 | |
| 4388 | const signed char InpWs = 0; // whitespace |
| 4389 | const signed char InpNameBe = 1; // NameBeginning |
| 4390 | const signed char InpGt = 2; // > |
| 4391 | const signed char InpQm = 3; // ? |
| 4392 | const signed char InpUnknown = 4; |
| 4393 | |
| 4394 | static const signed char table[16][5] = { |
| 4395 | /* InpWs, InpNameBe InpGt InpQm InpUnknown */ |
| 4396 | { -1, -1, -1, QmI, -1 }, // Init |
| 4397 | { -1, Name, -1, -1, -1 }, // QmI |
| 4398 | { -1, -1, -1, -1, -1 }, // Name (this state is left not through input) |
| 4399 | { Ws1, -1, -1, -1, -1 }, // XMLDecl |
| 4400 | { -1, Version, -1, -1, -1 }, // Ws1 |
| 4401 | { Ws2, -1, -1, Qm, -1 }, // PInstr |
| 4402 | { Char, Char, Char, Qm, Char }, // Ws2 |
| 4403 | { Ws3, -1, -1, ADone, -1 }, // Version |
| 4404 | { -1, EorSD, -1, ADone, -1 }, // Ws3 |
| 4405 | { Ws4, -1, -1, ADone, -1 }, // EorSD |
| 4406 | { -1, SD, -1, ADone, -1 }, // Ws4 |
| 4407 | { Ws5, -1, -1, ADone, -1 }, // SD |
| 4408 | { -1, -1, -1, ADone, -1 }, // Ws5 |
| 4409 | { -1, -1, Done, -1, -1 }, // ADone |
| 4410 | { Char, Char, Char, Qm, Char }, // Char |
| 4411 | { Char, Char, Done, Qm, Char }, // Qm |
| 4412 | }; |
| 4413 | signed char state; |
| 4414 | signed char input; |
| 4415 | |
| 4416 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 4417 | state = Init; |
| 4418 | } else { |
| 4419 | state = parseStack->pop().state; |
| 4420 | #if defined(QT_QXML_DEBUG) |
| 4421 | qDebug("QXmlSimpleReader: parsePI (cont) in state %d" , state); |
| 4422 | #endif |
| 4423 | if (!parseStack->isEmpty()) { |
| 4424 | ParseFunction function = parseStack->top().function; |
| 4425 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 4426 | parseStack->pop(); |
| 4427 | #if defined(QT_QXML_DEBUG) |
| 4428 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 4429 | #endif |
| 4430 | } |
| 4431 | if (!(this->*function)()) { |
| 4432 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4433 | return false; |
| 4434 | } |
| 4435 | } |
| 4436 | } |
| 4437 | |
| 4438 | for (;;) { |
| 4439 | switch (state) { |
| 4440 | case Name: |
| 4441 | // test what name was read and determine the next state |
| 4442 | // (not very beautiful, I admit) |
| 4443 | if (name().toLower() == QLatin1String("xml" )) { |
| 4444 | if (parsePI_xmldecl && name() == QLatin1String("xml" )) { |
| 4445 | state = XMLDecl; |
| 4446 | } else { |
| 4447 | reportParseError(error: QLatin1String(XMLERR_INVALIDNAMEFORPI)); |
| 4448 | return false; |
| 4449 | } |
| 4450 | } else { |
| 4451 | state = PInstr; |
| 4452 | stringClear(); |
| 4453 | } |
| 4454 | break; |
| 4455 | case Version: |
| 4456 | // get version (syntax like an attribute) |
| 4457 | if (name() != QLatin1String("version" )) { |
| 4458 | reportParseError(error: QLatin1String(XMLERR_VERSIONEXPECTED)); |
| 4459 | return false; |
| 4460 | } |
| 4461 | xmlVersion = string(); |
| 4462 | break; |
| 4463 | case EorSD: |
| 4464 | // get the EDecl or SDDecl (syntax like an attribute) |
| 4465 | if (name() == QLatin1String("standalone" )) { |
| 4466 | if (string()== QLatin1String("yes" )) { |
| 4467 | standalone = QXmlSimpleReaderPrivate::Yes; |
| 4468 | } else if (string() == QLatin1String("no" )) { |
| 4469 | standalone = QXmlSimpleReaderPrivate::No; |
| 4470 | } else { |
| 4471 | reportParseError(error: QLatin1String(XMLERR_WRONGVALUEFORSDECL)); |
| 4472 | return false; |
| 4473 | } |
| 4474 | } else if (name() == QLatin1String("encoding" )) { |
| 4475 | encoding = string(); |
| 4476 | } else { |
| 4477 | reportParseError(error: QLatin1String(XMLERR_EDECLORSDDECLEXPECTED)); |
| 4478 | return false; |
| 4479 | } |
| 4480 | break; |
| 4481 | case SD: |
| 4482 | if (name() != QLatin1String("standalone" )) { |
| 4483 | reportParseError(error: QLatin1String(XMLERR_SDDECLEXPECTED)); |
| 4484 | return false; |
| 4485 | } |
| 4486 | if (string() == QLatin1String("yes" )) { |
| 4487 | standalone = QXmlSimpleReaderPrivate::Yes; |
| 4488 | } else if (string() == QLatin1String("no" )) { |
| 4489 | standalone = QXmlSimpleReaderPrivate::No; |
| 4490 | } else { |
| 4491 | reportParseError(error: QLatin1String(XMLERR_WRONGVALUEFORSDECL)); |
| 4492 | return false; |
| 4493 | } |
| 4494 | break; |
| 4495 | case Qm: |
| 4496 | // test if the skipping was legal |
| 4497 | if (!atEnd() && c != QLatin1Char('>')) |
| 4498 | stringAddC(QLatin1Char('?')); |
| 4499 | break; |
| 4500 | case Done: |
| 4501 | return true; |
| 4502 | case -1: |
| 4503 | // Error |
| 4504 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 4505 | return false; |
| 4506 | } |
| 4507 | |
| 4508 | if (atEnd()) { |
| 4509 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4510 | return false; |
| 4511 | } |
| 4512 | if (is_S(ch: c)) { |
| 4513 | input = InpWs; |
| 4514 | } else if (determineNameChar(ch: c) == NameBeginning) { |
| 4515 | input = InpNameBe; |
| 4516 | } else if (c == QLatin1Char('>')) { |
| 4517 | input = InpGt; |
| 4518 | } else if (c == QLatin1Char('?')) { |
| 4519 | input = InpQm; |
| 4520 | } else { |
| 4521 | input = InpUnknown; |
| 4522 | } |
| 4523 | state = table[state][input]; |
| 4524 | |
| 4525 | switch (state) { |
| 4526 | case QmI: |
| 4527 | next(); |
| 4528 | break; |
| 4529 | case Name: |
| 4530 | parseName_useRef = false; |
| 4531 | if (!parseName()) { |
| 4532 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4533 | return false; |
| 4534 | } |
| 4535 | break; |
| 4536 | case Ws1: |
| 4537 | case Ws2: |
| 4538 | case Ws3: |
| 4539 | case Ws4: |
| 4540 | case Ws5: |
| 4541 | if (!eat_ws()) { |
| 4542 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4543 | return false; |
| 4544 | } |
| 4545 | break; |
| 4546 | case Version: |
| 4547 | if (!parseAttribute()) { |
| 4548 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4549 | return false; |
| 4550 | } |
| 4551 | break; |
| 4552 | case EorSD: |
| 4553 | if (!parseAttribute()) { |
| 4554 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4555 | return false; |
| 4556 | } |
| 4557 | break; |
| 4558 | case SD: |
| 4559 | // get the SDDecl (syntax like an attribute) |
| 4560 | if (standalone != QXmlSimpleReaderPrivate::Unknown) { |
| 4561 | // already parsed the standalone declaration |
| 4562 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 4563 | return false; |
| 4564 | } |
| 4565 | if (!parseAttribute()) { |
| 4566 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePI, state); |
| 4567 | return false; |
| 4568 | } |
| 4569 | break; |
| 4570 | case ADone: |
| 4571 | next(); |
| 4572 | break; |
| 4573 | case Char: |
| 4574 | stringAddC(); |
| 4575 | next(); |
| 4576 | break; |
| 4577 | case Qm: |
| 4578 | // skip the '?' |
| 4579 | next(); |
| 4580 | break; |
| 4581 | case Done: |
| 4582 | next(); |
| 4583 | break; |
| 4584 | } |
| 4585 | } |
| 4586 | return false; |
| 4587 | } |
| 4588 | |
| 4589 | /* |
| 4590 | Parse a document type definition (doctypedecl [28]). |
| 4591 | |
| 4592 | Precondition: the beginning '<!' of the doctype is already read the head |
| 4593 | stands on the 'D' of '<!DOCTYPE'. |
| 4594 | |
| 4595 | If this function was successful, the head-position is on the first |
| 4596 | character after the document type definition. |
| 4597 | */ |
| 4598 | bool QXmlSimpleReaderPrivate::parseDoctype() |
| 4599 | { |
| 4600 | const signed char Init = 0; |
| 4601 | const signed char Doctype = 1; // read the doctype |
| 4602 | const signed char Ws1 = 2; // eat_ws |
| 4603 | const signed char Doctype2 = 3; // read the doctype, part 2 |
| 4604 | const signed char Ws2 = 4; // eat_ws |
| 4605 | const signed char Sys = 5; // read SYSTEM or PUBLIC |
| 4606 | const signed char Ws3 = 6; // eat_ws |
| 4607 | const signed char MP = 7; // markupdecl or PEReference |
| 4608 | const signed char MPR = 8; // same as MP, but already reported |
| 4609 | const signed char PER = 9; // PERReference |
| 4610 | const signed char Mup = 10; // markupdecl |
| 4611 | const signed char Ws4 = 11; // eat_ws |
| 4612 | const signed char MPE = 12; // end of markupdecl or PEReference |
| 4613 | const signed char Done = 13; |
| 4614 | |
| 4615 | const signed char InpWs = 0; |
| 4616 | const signed char InpD = 1; // 'D' |
| 4617 | const signed char InpS = 2; // 'S' or 'P' |
| 4618 | const signed char InpOB = 3; // [ |
| 4619 | const signed char InpCB = 4; //] |
| 4620 | const signed char InpPer = 5; // % |
| 4621 | const signed char InpGt = 6; // > |
| 4622 | const signed char InpUnknown = 7; |
| 4623 | |
| 4624 | static const signed char table[13][8] = { |
| 4625 | /* InpWs, InpD InpS InpOB InpCB InpPer InpGt InpUnknown */ |
| 4626 | { -1, Doctype, -1, -1, -1, -1, -1, -1 }, // Init |
| 4627 | { Ws1, -1, -1, -1, -1, -1, -1, -1 }, // Doctype |
| 4628 | { -1, Doctype2, Doctype2, -1, -1, -1, -1, Doctype2 }, // Ws1 |
| 4629 | { Ws2, -1, Sys, MP, -1, -1, Done, -1 }, // Doctype2 |
| 4630 | { -1, -1, Sys, MP, -1, -1, Done, -1 }, // Ws2 |
| 4631 | { Ws3, -1, -1, MP, -1, -1, Done, -1 }, // Sys |
| 4632 | { -1, -1, -1, MP, -1, -1, Done, -1 }, // Ws3 |
| 4633 | { -1, -1, -1, -1, MPE, PER, -1, Mup }, // MP |
| 4634 | { -1, -1, -1, -1, MPE, PER, -1, Mup }, // MPR |
| 4635 | { Ws4, -1, -1, -1, MPE, PER, -1, Mup }, // PER |
| 4636 | { Ws4, -1, -1, -1, MPE, PER, -1, Mup }, // Mup |
| 4637 | { -1, -1, -1, -1, MPE, PER, -1, Mup }, // Ws4 |
| 4638 | { -1, -1, -1, -1, -1, -1, Done, -1 } // MPE |
| 4639 | }; |
| 4640 | signed char state; |
| 4641 | signed char input; |
| 4642 | |
| 4643 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 4644 | startDTDwasReported = false; |
| 4645 | systemId.clear(); |
| 4646 | publicId.clear(); |
| 4647 | state = Init; |
| 4648 | } else { |
| 4649 | state = parseStack->pop().state; |
| 4650 | #if defined(QT_QXML_DEBUG) |
| 4651 | qDebug("QXmlSimpleReader: parseDoctype (cont) in state %d" , state); |
| 4652 | #endif |
| 4653 | if (!parseStack->isEmpty()) { |
| 4654 | ParseFunction function = parseStack->top().function; |
| 4655 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 4656 | parseStack->pop(); |
| 4657 | #if defined(QT_QXML_DEBUG) |
| 4658 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 4659 | #endif |
| 4660 | } |
| 4661 | if (!(this->*function)()) { |
| 4662 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4663 | return false; |
| 4664 | } |
| 4665 | } |
| 4666 | } |
| 4667 | |
| 4668 | for (;;) { |
| 4669 | switch (state) { |
| 4670 | case Doctype2: |
| 4671 | doctype = name(); |
| 4672 | break; |
| 4673 | case MP: |
| 4674 | if (!startDTDwasReported && lexicalHnd ) { |
| 4675 | startDTDwasReported = true; |
| 4676 | if (!lexicalHnd->startDTD(name: doctype, publicId, systemId)) { |
| 4677 | reportParseError(error: lexicalHnd->errorString()); |
| 4678 | return false; |
| 4679 | } |
| 4680 | } |
| 4681 | state = MPR; |
| 4682 | break; |
| 4683 | case Done: |
| 4684 | return true; |
| 4685 | case -1: |
| 4686 | // Error |
| 4687 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGDOCTYPE)); |
| 4688 | return false; |
| 4689 | } |
| 4690 | |
| 4691 | if (atEnd()) { |
| 4692 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4693 | return false; |
| 4694 | } |
| 4695 | if (is_S(ch: c)) { |
| 4696 | input = InpWs; |
| 4697 | } else if (c == QLatin1Char('D')) { |
| 4698 | input = InpD; |
| 4699 | } else if (c == QLatin1Char('S')) { |
| 4700 | input = InpS; |
| 4701 | } else if (c == QLatin1Char('P')) { |
| 4702 | input = InpS; |
| 4703 | } else if (c == QLatin1Char('[')) { |
| 4704 | input = InpOB; |
| 4705 | } else if (c == QLatin1Char(']')) { |
| 4706 | input = InpCB; |
| 4707 | } else if (c == QLatin1Char('%')) { |
| 4708 | input = InpPer; |
| 4709 | } else if (c == QLatin1Char('>')) { |
| 4710 | input = InpGt; |
| 4711 | } else { |
| 4712 | input = InpUnknown; |
| 4713 | } |
| 4714 | state = table[state][input]; |
| 4715 | |
| 4716 | switch (state) { |
| 4717 | case Doctype: |
| 4718 | parseString_s = QLatin1String("DOCTYPE" ); |
| 4719 | if (!parseString()) { |
| 4720 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4721 | return false; |
| 4722 | } |
| 4723 | break; |
| 4724 | case Ws1: |
| 4725 | case Ws2: |
| 4726 | case Ws3: |
| 4727 | case Ws4: |
| 4728 | if (!eat_ws()) { |
| 4729 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4730 | return false; |
| 4731 | } |
| 4732 | break; |
| 4733 | case Doctype2: |
| 4734 | parseName_useRef = false; |
| 4735 | if (!parseName()) { |
| 4736 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4737 | return false; |
| 4738 | } |
| 4739 | break; |
| 4740 | case Sys: |
| 4741 | parseExternalID_allowPublicID = false; |
| 4742 | if (!parseExternalID()) { |
| 4743 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4744 | return false; |
| 4745 | } |
| 4746 | thisPublicId = publicId; |
| 4747 | thisSystemId = systemId; |
| 4748 | break; |
| 4749 | case MP: |
| 4750 | case MPR: |
| 4751 | if (!next_eat_ws()) { |
| 4752 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4753 | return false; |
| 4754 | } |
| 4755 | break; |
| 4756 | case PER: |
| 4757 | parsePEReference_context = InDTD; |
| 4758 | if (!parsePEReference()) { |
| 4759 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4760 | return false; |
| 4761 | } |
| 4762 | break; |
| 4763 | case Mup: |
| 4764 | if (dtdRecursionLimit > 0 && parameterEntities.size() > dtdRecursionLimit) { |
| 4765 | reportParseError(error: QString::fromLatin1( |
| 4766 | ba: "DTD parsing exceeded recursion limit of %1." ).arg(a: dtdRecursionLimit)); |
| 4767 | return false; |
| 4768 | } |
| 4769 | if (!parseMarkupdecl()) { |
| 4770 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4771 | return false; |
| 4772 | } |
| 4773 | break; |
| 4774 | case MPE: |
| 4775 | if (!next_eat_ws()) { |
| 4776 | parseFailed(where: &QXmlSimpleReaderPrivate::parseDoctype, state); |
| 4777 | return false; |
| 4778 | } |
| 4779 | break; |
| 4780 | case Done: |
| 4781 | if (lexicalHnd) { |
| 4782 | if (!startDTDwasReported) { |
| 4783 | startDTDwasReported = true; |
| 4784 | if (!lexicalHnd->startDTD(name: doctype, publicId, systemId)) { |
| 4785 | reportParseError(error: lexicalHnd->errorString()); |
| 4786 | return false; |
| 4787 | } |
| 4788 | } |
| 4789 | if (!lexicalHnd->endDTD()) { |
| 4790 | reportParseError(error: lexicalHnd->errorString()); |
| 4791 | return false; |
| 4792 | } |
| 4793 | } |
| 4794 | next(); |
| 4795 | break; |
| 4796 | } |
| 4797 | } |
| 4798 | return false; |
| 4799 | } |
| 4800 | |
| 4801 | /* |
| 4802 | Parse a ExternalID [75]. |
| 4803 | |
| 4804 | If allowPublicID is true parse ExternalID [75] or PublicID [83]. |
| 4805 | */ |
| 4806 | bool QXmlSimpleReaderPrivate::parseExternalID() |
| 4807 | { |
| 4808 | const signed char Init = 0; |
| 4809 | const signed char Sys = 1; // parse 'SYSTEM' |
| 4810 | const signed char SysWS = 2; // parse the whitespace after 'SYSTEM' |
| 4811 | const signed char SysSQ = 3; // parse SystemLiteral with ' |
| 4812 | const signed char SysSQ2 = 4; // parse SystemLiteral with ' |
| 4813 | const signed char SysDQ = 5; // parse SystemLiteral with " |
| 4814 | const signed char SysDQ2 = 6; // parse SystemLiteral with " |
| 4815 | const signed char Pub = 7; // parse 'PUBLIC' |
| 4816 | const signed char PubWS = 8; // parse the whitespace after 'PUBLIC' |
| 4817 | const signed char PubSQ = 9; // parse PubidLiteral with ' |
| 4818 | const signed char PubSQ2 = 10; // parse PubidLiteral with ' |
| 4819 | const signed char PubDQ = 11; // parse PubidLiteral with " |
| 4820 | const signed char PubDQ2 = 12; // parse PubidLiteral with " |
| 4821 | const signed char PubE = 13; // finished parsing the PubidLiteral |
| 4822 | const signed char PubWS2 = 14; // parse the whitespace after the PubidLiteral |
| 4823 | const signed char PDone = 15; // done if allowPublicID is true |
| 4824 | const signed char Done = 16; |
| 4825 | |
| 4826 | const signed char InpSQ = 0; // ' |
| 4827 | const signed char InpDQ = 1; // " |
| 4828 | const signed char InpS = 2; // S |
| 4829 | const signed char InpP = 3; // P |
| 4830 | const signed char InpWs = 4; // white space |
| 4831 | const signed char InpUnknown = 5; |
| 4832 | |
| 4833 | static const signed char table[15][6] = { |
| 4834 | /* InpSQ InpDQ InpS InpP InpWs InpUnknown */ |
| 4835 | { -1, -1, Sys, Pub, -1, -1 }, // Init |
| 4836 | { -1, -1, -1, -1, SysWS, -1 }, // Sys |
| 4837 | { SysSQ, SysDQ, -1, -1, -1, -1 }, // SysWS |
| 4838 | { Done, SysSQ2, SysSQ2, SysSQ2, SysSQ2, SysSQ2 }, // SysSQ |
| 4839 | { Done, SysSQ2, SysSQ2, SysSQ2, SysSQ2, SysSQ2 }, // SysSQ2 |
| 4840 | { SysDQ2, Done, SysDQ2, SysDQ2, SysDQ2, SysDQ2 }, // SysDQ |
| 4841 | { SysDQ2, Done, SysDQ2, SysDQ2, SysDQ2, SysDQ2 }, // SysDQ2 |
| 4842 | { -1, -1, -1, -1, PubWS, -1 }, // Pub |
| 4843 | { PubSQ, PubDQ, -1, -1, -1, -1 }, // PubWS |
| 4844 | { PubE, -1, PubSQ2, PubSQ2, PubSQ2, PubSQ2 }, // PubSQ |
| 4845 | { PubE, -1, PubSQ2, PubSQ2, PubSQ2, PubSQ2 }, // PubSQ2 |
| 4846 | { -1, PubE, PubDQ2, PubDQ2, PubDQ2, PubDQ2 }, // PubDQ |
| 4847 | { -1, PubE, PubDQ2, PubDQ2, PubDQ2, PubDQ2 }, // PubDQ2 |
| 4848 | { PDone, PDone, PDone, PDone, PubWS2, PDone }, // PubE |
| 4849 | { SysSQ, SysDQ, PDone, PDone, PDone, PDone } // PubWS2 |
| 4850 | }; |
| 4851 | signed char state; |
| 4852 | signed char input; |
| 4853 | |
| 4854 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 4855 | systemId.clear(); |
| 4856 | publicId.clear(); |
| 4857 | state = Init; |
| 4858 | } else { |
| 4859 | state = parseStack->pop().state; |
| 4860 | #if defined(QT_QXML_DEBUG) |
| 4861 | qDebug("QXmlSimpleReader: parseExternalID (cont) in state %d" , state); |
| 4862 | #endif |
| 4863 | if (!parseStack->isEmpty()) { |
| 4864 | ParseFunction function = parseStack->top().function; |
| 4865 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 4866 | parseStack->pop(); |
| 4867 | #if defined(QT_QXML_DEBUG) |
| 4868 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 4869 | #endif |
| 4870 | } |
| 4871 | if (!(this->*function)()) { |
| 4872 | parseFailed(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4873 | return false; |
| 4874 | } |
| 4875 | } |
| 4876 | } |
| 4877 | |
| 4878 | for (;;) { |
| 4879 | switch (state) { |
| 4880 | case PDone: |
| 4881 | if (parseExternalID_allowPublicID) { |
| 4882 | publicId = string(); |
| 4883 | return true; |
| 4884 | } else { |
| 4885 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 4886 | return false; |
| 4887 | } |
| 4888 | case Done: |
| 4889 | return true; |
| 4890 | case -1: |
| 4891 | // Error |
| 4892 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 4893 | return false; |
| 4894 | } |
| 4895 | |
| 4896 | if (atEnd()) { |
| 4897 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4898 | return false; |
| 4899 | } |
| 4900 | if (is_S(ch: c)) { |
| 4901 | input = InpWs; |
| 4902 | } else if (c == QLatin1Char('\'')) { |
| 4903 | input = InpSQ; |
| 4904 | } else if (c == QLatin1Char('"')) { |
| 4905 | input = InpDQ; |
| 4906 | } else if (c == QLatin1Char('S')) { |
| 4907 | input = InpS; |
| 4908 | } else if (c == QLatin1Char('P')) { |
| 4909 | input = InpP; |
| 4910 | } else { |
| 4911 | input = InpUnknown; |
| 4912 | } |
| 4913 | state = table[state][input]; |
| 4914 | |
| 4915 | switch (state) { |
| 4916 | case Sys: |
| 4917 | parseString_s = QLatin1String("SYSTEM" ); |
| 4918 | if (!parseString()) { |
| 4919 | parseFailed(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4920 | return false; |
| 4921 | } |
| 4922 | break; |
| 4923 | case SysWS: |
| 4924 | if (!eat_ws()) { |
| 4925 | parseFailed(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4926 | return false; |
| 4927 | } |
| 4928 | break; |
| 4929 | case SysSQ: |
| 4930 | case SysDQ: |
| 4931 | stringClear(); |
| 4932 | next(); |
| 4933 | break; |
| 4934 | case SysSQ2: |
| 4935 | case SysDQ2: |
| 4936 | stringAddC(); |
| 4937 | next(); |
| 4938 | break; |
| 4939 | case Pub: |
| 4940 | parseString_s = QLatin1String("PUBLIC" ); |
| 4941 | if (!parseString()) { |
| 4942 | parseFailed(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4943 | return false; |
| 4944 | } |
| 4945 | break; |
| 4946 | case PubWS: |
| 4947 | if (!eat_ws()) { |
| 4948 | parseFailed(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4949 | return false; |
| 4950 | } |
| 4951 | break; |
| 4952 | case PubSQ: |
| 4953 | case PubDQ: |
| 4954 | stringClear(); |
| 4955 | next(); |
| 4956 | break; |
| 4957 | case PubSQ2: |
| 4958 | case PubDQ2: |
| 4959 | stringAddC(); |
| 4960 | next(); |
| 4961 | break; |
| 4962 | case PubE: |
| 4963 | next(); |
| 4964 | break; |
| 4965 | case PubWS2: |
| 4966 | publicId = string(); |
| 4967 | if (!eat_ws()) { |
| 4968 | parseFailed(where: &QXmlSimpleReaderPrivate::parseExternalID, state); |
| 4969 | return false; |
| 4970 | } |
| 4971 | break; |
| 4972 | case Done: |
| 4973 | systemId = string(); |
| 4974 | next(); |
| 4975 | break; |
| 4976 | } |
| 4977 | } |
| 4978 | return false; |
| 4979 | } |
| 4980 | |
| 4981 | /* |
| 4982 | Parse a markupdecl [29]. |
| 4983 | */ |
| 4984 | bool QXmlSimpleReaderPrivate::parseMarkupdecl() |
| 4985 | { |
| 4986 | const signed char Init = 0; |
| 4987 | const signed char Lt = 1; // < was read |
| 4988 | const signed char Em = 2; // ! was read |
| 4989 | const signed char CE = 3; // E was read |
| 4990 | const signed char Qm = 4; // ? was read |
| 4991 | const signed char Dash = 5; // - was read |
| 4992 | const signed char CA = 6; // A was read |
| 4993 | const signed char CEL = 7; // EL was read |
| 4994 | const signed char CEN = 8; // EN was read |
| 4995 | const signed char CN = 9; // N was read |
| 4996 | const signed char Done = 10; |
| 4997 | |
| 4998 | const signed char InpLt = 0; // < |
| 4999 | const signed char InpQm = 1; // ? |
| 5000 | const signed char InpEm = 2; // ! |
| 5001 | const signed char InpDash = 3; // - |
| 5002 | const signed char InpA = 4; // A |
| 5003 | const signed char InpE = 5; // E |
| 5004 | const signed char InpL = 6; // L |
| 5005 | const signed char InpN = 7; // N |
| 5006 | const signed char InpUnknown = 8; |
| 5007 | |
| 5008 | static const signed char table[4][9] = { |
| 5009 | /* InpLt InpQm InpEm InpDash InpA InpE InpL InpN InpUnknown */ |
| 5010 | { Lt, -1, -1, -1, -1, -1, -1, -1, -1 }, // Init |
| 5011 | { -1, Qm, Em, -1, -1, -1, -1, -1, -1 }, // Lt |
| 5012 | { -1, -1, -1, Dash, CA, CE, -1, CN, -1 }, // Em |
| 5013 | { -1, -1, -1, -1, -1, -1, CEL, CEN, -1 } // CE |
| 5014 | }; |
| 5015 | signed char state; |
| 5016 | signed char input; |
| 5017 | |
| 5018 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 5019 | state = Init; |
| 5020 | } else { |
| 5021 | state = parseStack->pop().state; |
| 5022 | #if defined(QT_QXML_DEBUG) |
| 5023 | qDebug("QXmlSimpleReader: parseMarkupdecl (cont) in state %d" , state); |
| 5024 | #endif |
| 5025 | if (!parseStack->isEmpty()) { |
| 5026 | ParseFunction function = parseStack->top().function; |
| 5027 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 5028 | parseStack->pop(); |
| 5029 | #if defined(QT_QXML_DEBUG) |
| 5030 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 5031 | #endif |
| 5032 | } |
| 5033 | if (!(this->*function)()) { |
| 5034 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5035 | return false; |
| 5036 | } |
| 5037 | } |
| 5038 | } |
| 5039 | |
| 5040 | for (;;) { |
| 5041 | switch (state) { |
| 5042 | case Qm: |
| 5043 | if (contentHnd) { |
| 5044 | if (!contentHnd->processingInstruction(target: name(),data: string())) { |
| 5045 | reportParseError(error: contentHnd->errorString()); |
| 5046 | return false; |
| 5047 | } |
| 5048 | } |
| 5049 | return true; |
| 5050 | case Dash: |
| 5051 | if (lexicalHnd) { |
| 5052 | if (!lexicalHnd->comment(ch: string())) { |
| 5053 | reportParseError(error: lexicalHnd->errorString()); |
| 5054 | return false; |
| 5055 | } |
| 5056 | } |
| 5057 | return true; |
| 5058 | case CA: |
| 5059 | return true; |
| 5060 | case CEL: |
| 5061 | return true; |
| 5062 | case CEN: |
| 5063 | return true; |
| 5064 | case CN: |
| 5065 | return true; |
| 5066 | case Done: |
| 5067 | return true; |
| 5068 | case -1: |
| 5069 | // Error |
| 5070 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 5071 | return false; |
| 5072 | } |
| 5073 | |
| 5074 | if (atEnd()) { |
| 5075 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5076 | return false; |
| 5077 | } |
| 5078 | if (c == QLatin1Char('<')) { |
| 5079 | input = InpLt; |
| 5080 | } else if (c == QLatin1Char('?')) { |
| 5081 | input = InpQm; |
| 5082 | } else if (c == QLatin1Char('!')) { |
| 5083 | input = InpEm; |
| 5084 | } else if (c == QLatin1Char('-')) { |
| 5085 | input = InpDash; |
| 5086 | } else if (c == QLatin1Char('A')) { |
| 5087 | input = InpA; |
| 5088 | } else if (c == QLatin1Char('E')) { |
| 5089 | input = InpE; |
| 5090 | } else if (c == QLatin1Char('L')) { |
| 5091 | input = InpL; |
| 5092 | } else if (c == QLatin1Char('N')) { |
| 5093 | input = InpN; |
| 5094 | } else { |
| 5095 | input = InpUnknown; |
| 5096 | } |
| 5097 | state = table[state][input]; |
| 5098 | |
| 5099 | switch (state) { |
| 5100 | case Lt: |
| 5101 | next(); |
| 5102 | break; |
| 5103 | case Em: |
| 5104 | next(); |
| 5105 | break; |
| 5106 | case CE: |
| 5107 | next(); |
| 5108 | break; |
| 5109 | case Qm: |
| 5110 | parsePI_xmldecl = false; |
| 5111 | if (!parsePI()) { |
| 5112 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5113 | return false; |
| 5114 | } |
| 5115 | break; |
| 5116 | case Dash: |
| 5117 | if (!parseComment()) { |
| 5118 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5119 | return false; |
| 5120 | } |
| 5121 | break; |
| 5122 | case CA: |
| 5123 | if (!parseAttlistDecl()) { |
| 5124 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5125 | return false; |
| 5126 | } |
| 5127 | break; |
| 5128 | case CEL: |
| 5129 | if (!parseElementDecl()) { |
| 5130 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5131 | return false; |
| 5132 | } |
| 5133 | break; |
| 5134 | case CEN: |
| 5135 | if (!parseEntityDecl()) { |
| 5136 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5137 | return false; |
| 5138 | } |
| 5139 | break; |
| 5140 | case CN: |
| 5141 | if (!parseNotationDecl()) { |
| 5142 | parseFailed(where: &QXmlSimpleReaderPrivate::parseMarkupdecl, state); |
| 5143 | return false; |
| 5144 | } |
| 5145 | break; |
| 5146 | } |
| 5147 | } |
| 5148 | return false; |
| 5149 | } |
| 5150 | |
| 5151 | /* |
| 5152 | Parse a PEReference [69] |
| 5153 | */ |
| 5154 | bool QXmlSimpleReaderPrivate::parsePEReference() |
| 5155 | { |
| 5156 | const signed char Init = 0; |
| 5157 | const signed char Next = 1; |
| 5158 | const signed char Name = 2; |
| 5159 | const signed char NameR = 3; // same as Name, but already reported |
| 5160 | const signed char Done = 4; |
| 5161 | |
| 5162 | const signed char InpSemi = 0; // ; |
| 5163 | const signed char InpPer = 1; // % |
| 5164 | const signed char InpUnknown = 2; |
| 5165 | |
| 5166 | static const signed char table[4][3] = { |
| 5167 | /* InpSemi InpPer InpUnknown */ |
| 5168 | { -1, Next, -1 }, // Init |
| 5169 | { -1, -1, Name }, // Next |
| 5170 | { Done, -1, -1 }, // Name |
| 5171 | { Done, -1, -1 } // NameR |
| 5172 | }; |
| 5173 | signed char state; |
| 5174 | signed char input; |
| 5175 | |
| 5176 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 5177 | state = Init; |
| 5178 | } else { |
| 5179 | state = parseStack->pop().state; |
| 5180 | #if defined(QT_QXML_DEBUG) |
| 5181 | qDebug("QXmlSimpleReader: parsePEReference (cont) in state %d" , state); |
| 5182 | #endif |
| 5183 | if (!parseStack->isEmpty()) { |
| 5184 | ParseFunction function = parseStack->top().function; |
| 5185 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 5186 | parseStack->pop(); |
| 5187 | #if defined(QT_QXML_DEBUG) |
| 5188 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 5189 | #endif |
| 5190 | } |
| 5191 | if (!(this->*function)()) { |
| 5192 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePEReference, state); |
| 5193 | return false; |
| 5194 | } |
| 5195 | } |
| 5196 | } |
| 5197 | |
| 5198 | for (;;) { |
| 5199 | switch (state) { |
| 5200 | case Name: |
| 5201 | { |
| 5202 | bool skipIt = true; |
| 5203 | QString xmlRefString; |
| 5204 | |
| 5205 | QMap<QString,QString>::Iterator it; |
| 5206 | it = parameterEntities.find(key: ref()); |
| 5207 | if (it != parameterEntities.end()) { |
| 5208 | skipIt = false; |
| 5209 | xmlRefString = *it; |
| 5210 | } else if (entityRes) { |
| 5211 | QMap<QString,QXmlSimpleReaderPrivate::ExternParameterEntity>::Iterator it2; |
| 5212 | it2 = externParameterEntities.find(key: ref()); |
| 5213 | QXmlInputSource *ret = nullptr; |
| 5214 | if (it2 != externParameterEntities.end()) { |
| 5215 | if (!entityRes->resolveEntity(publicId: (*it2).publicId, systemId: (*it2).systemId, ret)) { |
| 5216 | delete ret; |
| 5217 | reportParseError(error: entityRes->errorString()); |
| 5218 | return false; |
| 5219 | } |
| 5220 | if (ret) { |
| 5221 | QString buffer = ret->data(); |
| 5222 | while (!buffer.isEmpty()) { |
| 5223 | xmlRefString += buffer; |
| 5224 | ret->fetchData(); |
| 5225 | buffer = ret->data(); |
| 5226 | } |
| 5227 | delete ret; |
| 5228 | if (!stripTextDecl(str&: xmlRefString)) { |
| 5229 | reportParseError(error: QLatin1String(XMLERR_ERRORINTEXTDECL)); |
| 5230 | return false; |
| 5231 | } |
| 5232 | skipIt = false; |
| 5233 | } |
| 5234 | } |
| 5235 | } |
| 5236 | |
| 5237 | if (skipIt) { |
| 5238 | if (contentHnd) { |
| 5239 | if (!contentHnd->skippedEntity(name: QLatin1Char('%') + ref())) { |
| 5240 | reportParseError(error: contentHnd->errorString()); |
| 5241 | return false; |
| 5242 | } |
| 5243 | } |
| 5244 | } else { |
| 5245 | if (parsePEReference_context == InEntityValue) { |
| 5246 | // Included in literal |
| 5247 | if (!insertXmlRef(xmlRefString, ref(), true)) |
| 5248 | return false; |
| 5249 | } else if (parsePEReference_context == InDTD) { |
| 5250 | // Included as PE |
| 5251 | if (!insertXmlRef(QLatin1Char(' ') + xmlRefString + QLatin1Char(' '), ref(), false)) |
| 5252 | return false; |
| 5253 | } |
| 5254 | } |
| 5255 | } |
| 5256 | state = NameR; |
| 5257 | break; |
| 5258 | case Done: |
| 5259 | return true; |
| 5260 | case -1: |
| 5261 | // Error |
| 5262 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 5263 | return false; |
| 5264 | } |
| 5265 | |
| 5266 | if (atEnd()) { |
| 5267 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parsePEReference, state); |
| 5268 | return false; |
| 5269 | } |
| 5270 | if (c == QLatin1Char(';')) { |
| 5271 | input = InpSemi; |
| 5272 | } else if (c == QLatin1Char('%')) { |
| 5273 | input = InpPer; |
| 5274 | } else { |
| 5275 | input = InpUnknown; |
| 5276 | } |
| 5277 | state = table[state][input]; |
| 5278 | |
| 5279 | switch (state) { |
| 5280 | case Next: |
| 5281 | next(); |
| 5282 | break; |
| 5283 | case Name: |
| 5284 | case NameR: |
| 5285 | parseName_useRef = true; |
| 5286 | if (!parseName()) { |
| 5287 | parseFailed(where: &QXmlSimpleReaderPrivate::parsePEReference, state); |
| 5288 | return false; |
| 5289 | } |
| 5290 | break; |
| 5291 | case Done: |
| 5292 | next(); |
| 5293 | break; |
| 5294 | } |
| 5295 | } |
| 5296 | return false; |
| 5297 | } |
| 5298 | |
| 5299 | /* |
| 5300 | Parse a AttlistDecl [52]. |
| 5301 | |
| 5302 | Precondition: the beginning '<!' is already read and the head |
| 5303 | stands on the 'A' of '<!ATTLIST' |
| 5304 | */ |
| 5305 | bool QXmlSimpleReaderPrivate::parseAttlistDecl() |
| 5306 | { |
| 5307 | const signed char Init = 0; |
| 5308 | const signed char Attlist = 1; // parse the string "ATTLIST" |
| 5309 | const signed char Ws = 2; // whitespace read |
| 5310 | const signed char Name = 3; // parse name |
| 5311 | const signed char Ws1 = 4; // whitespace read |
| 5312 | const signed char Attdef = 5; // parse the AttDef |
| 5313 | const signed char Ws2 = 6; // whitespace read |
| 5314 | const signed char Atttype = 7; // parse the AttType |
| 5315 | const signed char Ws3 = 8; // whitespace read |
| 5316 | const signed char DDecH = 9; // DefaultDecl with # |
| 5317 | const signed char DefReq = 10; // parse the string "REQUIRED" |
| 5318 | const signed char DefImp = 11; // parse the string "IMPLIED" |
| 5319 | const signed char DefFix = 12; // parse the string "FIXED" |
| 5320 | const signed char Attval = 13; // parse the AttValue |
| 5321 | const signed char Ws4 = 14; // whitespace read |
| 5322 | const signed char Done = 15; |
| 5323 | |
| 5324 | const signed char InpWs = 0; // white space |
| 5325 | const signed char InpGt = 1; // > |
| 5326 | const signed char InpHash = 2; // # |
| 5327 | const signed char InpA = 3; // A |
| 5328 | const signed char InpI = 4; // I |
| 5329 | const signed char InpF = 5; // F |
| 5330 | const signed char InpR = 6; // R |
| 5331 | const signed char InpUnknown = 7; |
| 5332 | |
| 5333 | static const signed char table[15][8] = { |
| 5334 | /* InpWs InpGt InpHash InpA InpI InpF InpR InpUnknown */ |
| 5335 | { -1, -1, -1, Attlist, -1, -1, -1, -1 }, // Init |
| 5336 | { Ws, -1, -1, -1, -1, -1, -1, -1 }, // Attlist |
| 5337 | { -1, -1, -1, Name, Name, Name, Name, Name }, // Ws |
| 5338 | { Ws1, Done, Attdef, Attdef, Attdef, Attdef, Attdef, Attdef }, // Name |
| 5339 | { -1, Done, Attdef, Attdef, Attdef, Attdef, Attdef, Attdef }, // Ws1 |
| 5340 | { Ws2, -1, -1, -1, -1, -1, -1, -1 }, // Attdef |
| 5341 | { -1, Atttype, Atttype, Atttype, Atttype, Atttype, Atttype, Atttype }, // Ws2 |
| 5342 | { Ws3, -1, -1, -1, -1, -1, -1, -1 }, // Attype |
| 5343 | { -1, Attval, DDecH, Attval, Attval, Attval, Attval, Attval }, // Ws3 |
| 5344 | { -1, -1, -1, -1, DefImp, DefFix, DefReq, -1 }, // DDecH |
| 5345 | { Ws4, Ws4, -1, -1, -1, -1, -1, -1 }, // DefReq |
| 5346 | { Ws4, Ws4, -1, -1, -1, -1, -1, -1 }, // DefImp |
| 5347 | { Ws3, -1, -1, -1, -1, -1, -1, -1 }, // DefFix |
| 5348 | { Ws4, Ws4, -1, -1, -1, -1, -1, -1 }, // Attval |
| 5349 | { -1, Done, Attdef, Attdef, Attdef, Attdef, Attdef, Attdef } // Ws4 |
| 5350 | }; |
| 5351 | signed char state; |
| 5352 | signed char input; |
| 5353 | |
| 5354 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 5355 | state = Init; |
| 5356 | } else { |
| 5357 | state = parseStack->pop().state; |
| 5358 | #if defined(QT_QXML_DEBUG) |
| 5359 | qDebug("QXmlSimpleReader: parseAttlistDecl (cont) in state %d" , state); |
| 5360 | #endif |
| 5361 | if (!parseStack->isEmpty()) { |
| 5362 | ParseFunction function = parseStack->top().function; |
| 5363 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 5364 | parseStack->pop(); |
| 5365 | #if defined(QT_QXML_DEBUG) |
| 5366 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 5367 | #endif |
| 5368 | } |
| 5369 | if (!(this->*function)()) { |
| 5370 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5371 | return false; |
| 5372 | } |
| 5373 | } |
| 5374 | } |
| 5375 | |
| 5376 | for (;;) { |
| 5377 | switch (state) { |
| 5378 | case Name: |
| 5379 | attDeclEName = name(); |
| 5380 | break; |
| 5381 | case Attdef: |
| 5382 | attDeclAName = name(); |
| 5383 | break; |
| 5384 | case Done: |
| 5385 | return true; |
| 5386 | case -1: |
| 5387 | // Error |
| 5388 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 5389 | return false; |
| 5390 | } |
| 5391 | |
| 5392 | if (atEnd()) { |
| 5393 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5394 | return false; |
| 5395 | } |
| 5396 | if (is_S(ch: c)) { |
| 5397 | input = InpWs; |
| 5398 | } else if (c == QLatin1Char('>')) { |
| 5399 | input = InpGt; |
| 5400 | } else if (c == QLatin1Char('#')) { |
| 5401 | input = InpHash; |
| 5402 | } else if (c == QLatin1Char('A')) { |
| 5403 | input = InpA; |
| 5404 | } else if (c == QLatin1Char('I')) { |
| 5405 | input = InpI; |
| 5406 | } else if (c == QLatin1Char('F')) { |
| 5407 | input = InpF; |
| 5408 | } else if (c == QLatin1Char('R')) { |
| 5409 | input = InpR; |
| 5410 | } else { |
| 5411 | input = InpUnknown; |
| 5412 | } |
| 5413 | state = table[state][input]; |
| 5414 | |
| 5415 | switch (state) { |
| 5416 | case Attlist: |
| 5417 | parseString_s = QLatin1String("ATTLIST" ); |
| 5418 | if (!parseString()) { |
| 5419 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5420 | return false; |
| 5421 | } |
| 5422 | break; |
| 5423 | case Ws: |
| 5424 | case Ws1: |
| 5425 | case Ws2: |
| 5426 | case Ws3: |
| 5427 | if (!eat_ws()) { |
| 5428 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5429 | return false; |
| 5430 | } |
| 5431 | break; |
| 5432 | case Name: |
| 5433 | parseName_useRef = false; |
| 5434 | if (!parseName()) { |
| 5435 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5436 | return false; |
| 5437 | } |
| 5438 | break; |
| 5439 | case Attdef: |
| 5440 | parseName_useRef = false; |
| 5441 | if (!parseName()) { |
| 5442 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5443 | return false; |
| 5444 | } |
| 5445 | break; |
| 5446 | case Atttype: |
| 5447 | if (!parseAttType()) { |
| 5448 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5449 | return false; |
| 5450 | } |
| 5451 | break; |
| 5452 | case DDecH: |
| 5453 | next(); |
| 5454 | break; |
| 5455 | case DefReq: |
| 5456 | parseString_s = QLatin1String("REQUIRED" ); |
| 5457 | if (!parseString()) { |
| 5458 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5459 | return false; |
| 5460 | } |
| 5461 | break; |
| 5462 | case DefImp: |
| 5463 | parseString_s = QLatin1String("IMPLIED" ); |
| 5464 | if (!parseString()) { |
| 5465 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5466 | return false; |
| 5467 | } |
| 5468 | break; |
| 5469 | case DefFix: |
| 5470 | parseString_s = QLatin1String("FIXED" ); |
| 5471 | if (!parseString()) { |
| 5472 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5473 | return false; |
| 5474 | } |
| 5475 | break; |
| 5476 | case Attval: |
| 5477 | if (!parseAttValue()) { |
| 5478 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5479 | return false; |
| 5480 | } |
| 5481 | break; |
| 5482 | case Ws4: |
| 5483 | if (declHnd) { |
| 5484 | // ### not all values are computed yet... |
| 5485 | if (!declHnd->attributeDecl(eName: attDeclEName, aName: attDeclAName, type: QLatin1String("" ), valueDefault: QLatin1String("" ), value: QLatin1String("" ))) { |
| 5486 | reportParseError(error: declHnd->errorString()); |
| 5487 | return false; |
| 5488 | } |
| 5489 | } |
| 5490 | if (!eat_ws()) { |
| 5491 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttlistDecl, state); |
| 5492 | return false; |
| 5493 | } |
| 5494 | break; |
| 5495 | case Done: |
| 5496 | next(); |
| 5497 | break; |
| 5498 | } |
| 5499 | } |
| 5500 | return false; |
| 5501 | } |
| 5502 | |
| 5503 | /* |
| 5504 | Parse a AttType [54] |
| 5505 | */ |
| 5506 | bool QXmlSimpleReaderPrivate::parseAttType() |
| 5507 | { |
| 5508 | const signed char Init = 0; |
| 5509 | const signed char ST = 1; // StringType |
| 5510 | const signed char TTI = 2; // TokenizedType starting with 'I' |
| 5511 | const signed char TTI2 = 3; // TokenizedType helpstate |
| 5512 | const signed char TTI3 = 4; // TokenizedType helpstate |
| 5513 | const signed char TTE = 5; // TokenizedType starting with 'E' |
| 5514 | const signed char TTEY = 6; // TokenizedType starting with 'ENTITY' |
| 5515 | const signed char TTEI = 7; // TokenizedType starting with 'ENTITI' |
| 5516 | const signed char N = 8; // N read (TokenizedType or Notation) |
| 5517 | const signed char TTNM = 9; // TokenizedType starting with 'NM' |
| 5518 | const signed char TTNM2 = 10; // TokenizedType helpstate |
| 5519 | const signed char NO = 11; // Notation |
| 5520 | const signed char NO2 = 12; // Notation helpstate |
| 5521 | const signed char NO3 = 13; // Notation helpstate |
| 5522 | const signed char NOName = 14; // Notation, read name |
| 5523 | const signed char NO4 = 15; // Notation helpstate |
| 5524 | const signed char EN = 16; // Enumeration |
| 5525 | const signed char ENNmt = 17; // Enumeration, read Nmtoken |
| 5526 | const signed char EN2 = 18; // Enumeration helpstate |
| 5527 | const signed char ADone = 19; // almost done (make next and accept) |
| 5528 | const signed char Done = 20; |
| 5529 | |
| 5530 | const signed char InpWs = 0; // whitespace |
| 5531 | const signed char InpOp = 1; // ( |
| 5532 | const signed char InpCp = 2; //) |
| 5533 | const signed char InpPipe = 3; // | |
| 5534 | const signed char InpC = 4; // C |
| 5535 | const signed char InpE = 5; // E |
| 5536 | const signed char InpI = 6; // I |
| 5537 | const signed char InpM = 7; // M |
| 5538 | const signed char InpN = 8; // N |
| 5539 | const signed char InpO = 9; // O |
| 5540 | const signed char InpR = 10; // R |
| 5541 | const signed char InpS = 11; // S |
| 5542 | const signed char InpY = 12; // Y |
| 5543 | const signed char InpUnknown = 13; |
| 5544 | |
| 5545 | static const signed char table[19][14] = { |
| 5546 | /* InpWs InpOp InpCp InpPipe InpC InpE InpI InpM InpN InpO InpR InpS InpY InpUnknown */ |
| 5547 | { -1, EN, -1, -1, ST, TTE, TTI, -1, N, -1, -1, -1, -1, -1 }, // Init |
| 5548 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done }, // ST |
| 5549 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, TTI2, Done, Done, Done }, // TTI |
| 5550 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, TTI3, Done, Done }, // TTI2 |
| 5551 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done }, // TTI3 |
| 5552 | { -1, -1, -1, -1, -1, -1, TTEI, -1, -1, -1, -1, -1, TTEY, -1 }, // TTE |
| 5553 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done }, // TTEY |
| 5554 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done }, // TTEI |
| 5555 | { -1, -1, -1, -1, -1, -1, -1, TTNM, -1, NO, -1, -1, -1, -1 }, // N |
| 5556 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, TTNM2, Done, Done }, // TTNM |
| 5557 | { Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done, Done }, // TTNM2 |
| 5558 | { NO2, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // NO |
| 5559 | { -1, NO3, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // NO2 |
| 5560 | { NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName, NOName }, // NO3 |
| 5561 | { NO4, -1, ADone, NO3, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // NOName |
| 5562 | { -1, -1, ADone, NO3, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // NO4 |
| 5563 | { -1, -1, ENNmt, -1, ENNmt, ENNmt, ENNmt, ENNmt, ENNmt, ENNmt, ENNmt, ENNmt, ENNmt, ENNmt }, // EN |
| 5564 | { EN2, -1, ADone, EN, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // ENNmt |
| 5565 | { -1, -1, ADone, EN, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 } // EN2 |
| 5566 | }; |
| 5567 | signed char state; |
| 5568 | signed char input; |
| 5569 | |
| 5570 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 5571 | state = Init; |
| 5572 | } else { |
| 5573 | state = parseStack->pop().state; |
| 5574 | #if defined(QT_QXML_DEBUG) |
| 5575 | qDebug("QXmlSimpleReader: parseAttType (cont) in state %d" , state); |
| 5576 | #endif |
| 5577 | if (!parseStack->isEmpty()) { |
| 5578 | ParseFunction function = parseStack->top().function; |
| 5579 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 5580 | parseStack->pop(); |
| 5581 | #if defined(QT_QXML_DEBUG) |
| 5582 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 5583 | #endif |
| 5584 | } |
| 5585 | if (!(this->*function)()) { |
| 5586 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5587 | return false; |
| 5588 | } |
| 5589 | } |
| 5590 | } |
| 5591 | |
| 5592 | for (;;) { |
| 5593 | switch (state) { |
| 5594 | case ADone: |
| 5595 | return true; |
| 5596 | case Done: |
| 5597 | return true; |
| 5598 | case -1: |
| 5599 | // Error |
| 5600 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 5601 | return false; |
| 5602 | } |
| 5603 | |
| 5604 | if (atEnd()) { |
| 5605 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5606 | return false; |
| 5607 | } |
| 5608 | if (is_S(ch: c)) { |
| 5609 | input = InpWs; |
| 5610 | } else if (c == QLatin1Char('(')) { |
| 5611 | input = InpOp; |
| 5612 | } else if (c == QLatin1Char(')')) { |
| 5613 | input = InpCp; |
| 5614 | } else if (c == QLatin1Char('|')) { |
| 5615 | input = InpPipe; |
| 5616 | } else if (c == QLatin1Char('C')) { |
| 5617 | input = InpC; |
| 5618 | } else if (c == QLatin1Char('E')) { |
| 5619 | input = InpE; |
| 5620 | } else if (c == QLatin1Char('I')) { |
| 5621 | input = InpI; |
| 5622 | } else if (c == QLatin1Char('M')) { |
| 5623 | input = InpM; |
| 5624 | } else if (c == QLatin1Char('N')) { |
| 5625 | input = InpN; |
| 5626 | } else if (c == QLatin1Char('O')) { |
| 5627 | input = InpO; |
| 5628 | } else if (c == QLatin1Char('R')) { |
| 5629 | input = InpR; |
| 5630 | } else if (c == QLatin1Char('S')) { |
| 5631 | input = InpS; |
| 5632 | } else if (c == QLatin1Char('Y')) { |
| 5633 | input = InpY; |
| 5634 | } else { |
| 5635 | input = InpUnknown; |
| 5636 | } |
| 5637 | state = table[state][input]; |
| 5638 | |
| 5639 | switch (state) { |
| 5640 | case ST: |
| 5641 | parseString_s = QLatin1String("CDATA" ); |
| 5642 | if (!parseString()) { |
| 5643 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5644 | return false; |
| 5645 | } |
| 5646 | break; |
| 5647 | case TTI: |
| 5648 | parseString_s = QLatin1String("ID" ); |
| 5649 | if (!parseString()) { |
| 5650 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5651 | return false; |
| 5652 | } |
| 5653 | break; |
| 5654 | case TTI2: |
| 5655 | parseString_s = QLatin1String("REF" ); |
| 5656 | if (!parseString()) { |
| 5657 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5658 | return false; |
| 5659 | } |
| 5660 | break; |
| 5661 | case TTI3: |
| 5662 | next(); // S |
| 5663 | break; |
| 5664 | case TTE: |
| 5665 | parseString_s = QLatin1String("ENTIT" ); |
| 5666 | if (!parseString()) { |
| 5667 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5668 | return false; |
| 5669 | } |
| 5670 | break; |
| 5671 | case TTEY: |
| 5672 | next(); // Y |
| 5673 | break; |
| 5674 | case TTEI: |
| 5675 | parseString_s = QLatin1String("IES" ); |
| 5676 | if (!parseString()) { |
| 5677 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5678 | return false; |
| 5679 | } |
| 5680 | break; |
| 5681 | case N: |
| 5682 | next(); // N |
| 5683 | break; |
| 5684 | case TTNM: |
| 5685 | parseString_s = QLatin1String("MTOKEN" ); |
| 5686 | if (!parseString()) { |
| 5687 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5688 | return false; |
| 5689 | } |
| 5690 | break; |
| 5691 | case TTNM2: |
| 5692 | next(); // S |
| 5693 | break; |
| 5694 | case NO: |
| 5695 | parseString_s = QLatin1String("OTATION" ); |
| 5696 | if (!parseString()) { |
| 5697 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5698 | return false; |
| 5699 | } |
| 5700 | break; |
| 5701 | case NO2: |
| 5702 | if (!eat_ws()) { |
| 5703 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5704 | return false; |
| 5705 | } |
| 5706 | break; |
| 5707 | case NO3: |
| 5708 | if (!next_eat_ws()) { |
| 5709 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5710 | return false; |
| 5711 | } |
| 5712 | break; |
| 5713 | case NOName: |
| 5714 | parseName_useRef = false; |
| 5715 | if (!parseName()) { |
| 5716 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5717 | return false; |
| 5718 | } |
| 5719 | break; |
| 5720 | case NO4: |
| 5721 | if (!eat_ws()) { |
| 5722 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5723 | return false; |
| 5724 | } |
| 5725 | break; |
| 5726 | case EN: |
| 5727 | if (!next_eat_ws()) { |
| 5728 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5729 | return false; |
| 5730 | } |
| 5731 | break; |
| 5732 | case ENNmt: |
| 5733 | if (!parseNmtoken()) { |
| 5734 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5735 | return false; |
| 5736 | } |
| 5737 | break; |
| 5738 | case EN2: |
| 5739 | if (!eat_ws()) { |
| 5740 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttType, state); |
| 5741 | return false; |
| 5742 | } |
| 5743 | break; |
| 5744 | case ADone: |
| 5745 | next(); |
| 5746 | break; |
| 5747 | } |
| 5748 | } |
| 5749 | return false; |
| 5750 | } |
| 5751 | |
| 5752 | /* |
| 5753 | Parse a AttValue [10] |
| 5754 | |
| 5755 | Precondition: the head stands on the beginning " or ' |
| 5756 | |
| 5757 | If this function was successful, the head stands on the first |
| 5758 | character after the closing " or ' and the value of the attribute |
| 5759 | is in string(). |
| 5760 | */ |
| 5761 | bool QXmlSimpleReaderPrivate::parseAttValue() |
| 5762 | { |
| 5763 | const signed char Init = 0; |
| 5764 | const signed char Dq = 1; // double quotes were read |
| 5765 | const signed char DqRef = 2; // read references in double quotes |
| 5766 | const signed char DqC = 3; // signed character read in double quotes |
| 5767 | const signed char Sq = 4; // single quotes were read |
| 5768 | const signed char SqRef = 5; // read references in single quotes |
| 5769 | const signed char SqC = 6; // signed character read in single quotes |
| 5770 | const signed char Done = 7; |
| 5771 | |
| 5772 | const signed char InpDq = 0; // " |
| 5773 | const signed char InpSq = 1; // ' |
| 5774 | const signed char InpAmp = 2; // & |
| 5775 | const signed char InpLt = 3; // < |
| 5776 | const signed char InpUnknown = 4; |
| 5777 | |
| 5778 | static const signed char table[7][5] = { |
| 5779 | /* InpDq InpSq InpAmp InpLt InpUnknown */ |
| 5780 | { Dq, Sq, -1, -1, -1 }, // Init |
| 5781 | { Done, DqC, DqRef, -1, DqC }, // Dq |
| 5782 | { Done, DqC, DqRef, -1, DqC }, // DqRef |
| 5783 | { Done, DqC, DqRef, -1, DqC }, // DqC |
| 5784 | { SqC, Done, SqRef, -1, SqC }, // Sq |
| 5785 | { SqC, Done, SqRef, -1, SqC }, // SqRef |
| 5786 | { SqC, Done, SqRef, -1, SqC } // SqRef |
| 5787 | }; |
| 5788 | signed char state; |
| 5789 | signed char input; |
| 5790 | |
| 5791 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 5792 | state = Init; |
| 5793 | } else { |
| 5794 | state = parseStack->pop().state; |
| 5795 | #if defined(QT_QXML_DEBUG) |
| 5796 | qDebug("QXmlSimpleReader: parseAttValue (cont) in state %d" , state); |
| 5797 | #endif |
| 5798 | if (!parseStack->isEmpty()) { |
| 5799 | ParseFunction function = parseStack->top().function; |
| 5800 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 5801 | parseStack->pop(); |
| 5802 | #if defined(QT_QXML_DEBUG) |
| 5803 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 5804 | #endif |
| 5805 | } |
| 5806 | if (!(this->*function)()) { |
| 5807 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttValue, state); |
| 5808 | return false; |
| 5809 | } |
| 5810 | } |
| 5811 | } |
| 5812 | |
| 5813 | for (;;) { |
| 5814 | switch (state) { |
| 5815 | case Done: |
| 5816 | return true; |
| 5817 | case -1: |
| 5818 | // Error |
| 5819 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 5820 | return false; |
| 5821 | } |
| 5822 | |
| 5823 | if (atEnd()) { |
| 5824 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseAttValue, state); |
| 5825 | return false; |
| 5826 | } |
| 5827 | if (c == QLatin1Char('"')) { |
| 5828 | input = InpDq; |
| 5829 | } else if (c == QLatin1Char('\'')) { |
| 5830 | input = InpSq; |
| 5831 | } else if (c == QLatin1Char('&')) { |
| 5832 | input = InpAmp; |
| 5833 | } else if (c == QLatin1Char('<')) { |
| 5834 | input = InpLt; |
| 5835 | } else { |
| 5836 | input = InpUnknown; |
| 5837 | } |
| 5838 | state = table[state][input]; |
| 5839 | |
| 5840 | switch (state) { |
| 5841 | case Dq: |
| 5842 | case Sq: |
| 5843 | stringClear(); |
| 5844 | next(); |
| 5845 | break; |
| 5846 | case DqRef: |
| 5847 | case SqRef: |
| 5848 | parseReference_context = InAttributeValue; |
| 5849 | if (!parseReference()) { |
| 5850 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttValue, state); |
| 5851 | return false; |
| 5852 | } |
| 5853 | break; |
| 5854 | case DqC: |
| 5855 | case SqC: |
| 5856 | stringAddC(); |
| 5857 | next(); |
| 5858 | break; |
| 5859 | case Done: |
| 5860 | next(); |
| 5861 | break; |
| 5862 | } |
| 5863 | } |
| 5864 | return false; |
| 5865 | } |
| 5866 | |
| 5867 | /* |
| 5868 | Parse a elementdecl [45]. |
| 5869 | |
| 5870 | Precondition: the beginning '<!E' is already read and the head |
| 5871 | stands on the 'L' of '<!ELEMENT' |
| 5872 | */ |
| 5873 | bool QXmlSimpleReaderPrivate::parseElementDecl() |
| 5874 | { |
| 5875 | const signed char Init = 0; |
| 5876 | const signed char Elem = 1; // parse the beginning string |
| 5877 | const signed char Ws1 = 2; // whitespace required |
| 5878 | const signed char Nam = 3; // parse Name |
| 5879 | const signed char Ws2 = 4; // whitespace required |
| 5880 | const signed char Empty = 5; // read EMPTY |
| 5881 | const signed char Any = 6; // read ANY |
| 5882 | const signed char Cont = 7; // read contentspec (except ANY or EMPTY) |
| 5883 | const signed char Mix = 8; // read Mixed |
| 5884 | const signed char Mix2 = 9; // |
| 5885 | const signed char Mix3 = 10; // |
| 5886 | const signed char MixN1 = 11; // |
| 5887 | const signed char MixN2 = 12; // |
| 5888 | const signed char MixN3 = 13; // |
| 5889 | const signed char MixN4 = 14; // |
| 5890 | const signed char Cp = 15; // parse cp |
| 5891 | const signed char Cp2 = 16; // |
| 5892 | const signed char WsD = 17; // eat whitespace before Done |
| 5893 | const signed char Done = 18; |
| 5894 | |
| 5895 | const signed char InpWs = 0; |
| 5896 | const signed char InpGt = 1; // > |
| 5897 | const signed char InpPipe = 2; // | |
| 5898 | const signed char InpOp = 3; // ( |
| 5899 | const signed char InpCp = 4; //) |
| 5900 | const signed char InpHash = 5; // # |
| 5901 | const signed char InpQm = 6; // ? |
| 5902 | const signed char InpAst = 7; // * |
| 5903 | const signed char InpPlus = 8; // + |
| 5904 | const signed char InpA = 9; // A |
| 5905 | const signed char InpE = 10; // E |
| 5906 | const signed char InpL = 11; // L |
| 5907 | const signed char InpUnknown = 12; |
| 5908 | |
| 5909 | static const signed char table[18][13] = { |
| 5910 | /* InpWs InpGt InpPipe InpOp InpCp InpHash InpQm InpAst InpPlus InpA InpE InpL InpUnknown */ |
| 5911 | { -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, Elem, -1 }, // Init |
| 5912 | { Ws1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // Elem |
| 5913 | { -1, -1, -1, -1, -1, -1, -1, -1, -1, Nam, Nam, Nam, Nam }, // Ws1 |
| 5914 | { Ws2, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // Nam |
| 5915 | { -1, -1, -1, Cont, -1, -1, -1, -1, -1, Any, Empty, -1, -1 }, // Ws2 |
| 5916 | { WsD, Done, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // Empty |
| 5917 | { WsD, Done, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // Any |
| 5918 | { -1, -1, -1, Cp, Cp, Mix, -1, -1, -1, Cp, Cp, Cp, Cp }, // Cont |
| 5919 | { Mix2, -1, MixN1, -1, Mix3, -1, -1, -1, -1, -1, -1, -1, -1 }, // Mix |
| 5920 | { -1, -1, MixN1, -1, Mix3, -1, -1, -1, -1, -1, -1, -1, -1 }, // Mix2 |
| 5921 | { WsD, Done, -1, -1, -1, -1, -1, WsD, -1, -1, -1, -1, -1 }, // Mix3 |
| 5922 | { -1, -1, -1, -1, -1, -1, -1, -1, -1, MixN2, MixN2, MixN2, MixN2 }, // MixN1 |
| 5923 | { MixN3, -1, MixN1, -1, MixN4, -1, -1, -1, -1, -1, -1, -1, -1 }, // MixN2 |
| 5924 | { -1, -1, MixN1, -1, MixN4, -1, -1, -1, -1, -1, -1, -1, -1 }, // MixN3 |
| 5925 | { -1, -1, -1, -1, -1, -1, -1, WsD, -1, -1, -1, -1, -1 }, // MixN4 |
| 5926 | { WsD, Done, -1, -1, -1, -1, Cp2, Cp2, Cp2, -1, -1, -1, -1 }, // Cp |
| 5927 | { WsD, Done, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 }, // Cp2 |
| 5928 | { -1, Done, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1 } // WsD |
| 5929 | }; |
| 5930 | signed char state; |
| 5931 | signed char input; |
| 5932 | |
| 5933 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 5934 | state = Init; |
| 5935 | } else { |
| 5936 | state = parseStack->pop().state; |
| 5937 | #if defined(QT_QXML_DEBUG) |
| 5938 | qDebug("QXmlSimpleReader: parseElementDecl (cont) in state %d" , state); |
| 5939 | #endif |
| 5940 | if (!parseStack->isEmpty()) { |
| 5941 | ParseFunction function = parseStack->top().function; |
| 5942 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 5943 | parseStack->pop(); |
| 5944 | #if defined(QT_QXML_DEBUG) |
| 5945 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 5946 | #endif |
| 5947 | } |
| 5948 | if (!(this->*function)()) { |
| 5949 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 5950 | return false; |
| 5951 | } |
| 5952 | } |
| 5953 | } |
| 5954 | |
| 5955 | for (;;) { |
| 5956 | switch (state) { |
| 5957 | case Done: |
| 5958 | return true; |
| 5959 | case -1: |
| 5960 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 5961 | return false; |
| 5962 | } |
| 5963 | |
| 5964 | if (atEnd()) { |
| 5965 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 5966 | return false; |
| 5967 | } |
| 5968 | if (is_S(ch: c)) { |
| 5969 | input = InpWs; |
| 5970 | } else if (c == QLatin1Char('>')) { |
| 5971 | input = InpGt; |
| 5972 | } else if (c == QLatin1Char('|')) { |
| 5973 | input = InpPipe; |
| 5974 | } else if (c == QLatin1Char('(')) { |
| 5975 | input = InpOp; |
| 5976 | } else if (c == QLatin1Char(')')) { |
| 5977 | input = InpCp; |
| 5978 | } else if (c == QLatin1Char('#')) { |
| 5979 | input = InpHash; |
| 5980 | } else if (c == QLatin1Char('?')) { |
| 5981 | input = InpQm; |
| 5982 | } else if (c == QLatin1Char('*')) { |
| 5983 | input = InpAst; |
| 5984 | } else if (c == QLatin1Char('+')) { |
| 5985 | input = InpPlus; |
| 5986 | } else if (c == QLatin1Char('A')) { |
| 5987 | input = InpA; |
| 5988 | } else if (c == QLatin1Char('E')) { |
| 5989 | input = InpE; |
| 5990 | } else if (c == QLatin1Char('L')) { |
| 5991 | input = InpL; |
| 5992 | } else { |
| 5993 | input = InpUnknown; |
| 5994 | } |
| 5995 | state = table[state][input]; |
| 5996 | |
| 5997 | switch (state) { |
| 5998 | case Elem: |
| 5999 | parseString_s = QLatin1String("LEMENT" ); |
| 6000 | if (!parseString()) { |
| 6001 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6002 | return false; |
| 6003 | } |
| 6004 | break; |
| 6005 | case Ws1: |
| 6006 | if (!eat_ws()) { |
| 6007 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6008 | return false; |
| 6009 | } |
| 6010 | break; |
| 6011 | case Nam: |
| 6012 | parseName_useRef = false; |
| 6013 | if (!parseName()) { |
| 6014 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6015 | return false; |
| 6016 | } |
| 6017 | break; |
| 6018 | case Ws2: |
| 6019 | if (!eat_ws()) { |
| 6020 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6021 | return false; |
| 6022 | } |
| 6023 | break; |
| 6024 | case Empty: |
| 6025 | parseString_s = QLatin1String("EMPTY" ); |
| 6026 | if (!parseString()) { |
| 6027 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6028 | return false; |
| 6029 | } |
| 6030 | break; |
| 6031 | case Any: |
| 6032 | parseString_s = QLatin1String("ANY" ); |
| 6033 | if (!parseString()) { |
| 6034 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6035 | return false; |
| 6036 | } |
| 6037 | break; |
| 6038 | case Cont: |
| 6039 | if (!next_eat_ws()) { |
| 6040 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6041 | return false; |
| 6042 | } |
| 6043 | break; |
| 6044 | case Mix: |
| 6045 | parseString_s = QLatin1String("#PCDATA" ); |
| 6046 | if (!parseString()) { |
| 6047 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6048 | return false; |
| 6049 | } |
| 6050 | break; |
| 6051 | case Mix2: |
| 6052 | if (!eat_ws()) { |
| 6053 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6054 | return false; |
| 6055 | } |
| 6056 | break; |
| 6057 | case Mix3: |
| 6058 | next(); |
| 6059 | break; |
| 6060 | case MixN1: |
| 6061 | if (!next_eat_ws()) { |
| 6062 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6063 | return false; |
| 6064 | } |
| 6065 | break; |
| 6066 | case MixN2: |
| 6067 | parseName_useRef = false; |
| 6068 | if (!parseName()) { |
| 6069 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6070 | return false; |
| 6071 | } |
| 6072 | break; |
| 6073 | case MixN3: |
| 6074 | if (!eat_ws()) { |
| 6075 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6076 | return false; |
| 6077 | } |
| 6078 | break; |
| 6079 | case MixN4: |
| 6080 | next(); |
| 6081 | break; |
| 6082 | case Cp: |
| 6083 | if (!parseChoiceSeq()) { |
| 6084 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6085 | return false; |
| 6086 | } |
| 6087 | break; |
| 6088 | case Cp2: |
| 6089 | next(); |
| 6090 | break; |
| 6091 | case WsD: |
| 6092 | if (!next_eat_ws()) { |
| 6093 | parseFailed(where: &QXmlSimpleReaderPrivate::parseElementDecl, state); |
| 6094 | return false; |
| 6095 | } |
| 6096 | break; |
| 6097 | case Done: |
| 6098 | next(); |
| 6099 | break; |
| 6100 | } |
| 6101 | } |
| 6102 | return false; |
| 6103 | } |
| 6104 | |
| 6105 | /* |
| 6106 | Parse a NotationDecl [82]. |
| 6107 | |
| 6108 | Precondition: the beginning '<!' is already read and the head |
| 6109 | stands on the 'N' of '<!NOTATION' |
| 6110 | */ |
| 6111 | bool QXmlSimpleReaderPrivate::parseNotationDecl() |
| 6112 | { |
| 6113 | const signed char Init = 0; |
| 6114 | const signed char Not = 1; // read NOTATION |
| 6115 | const signed char Ws1 = 2; // eat whitespaces |
| 6116 | const signed char Nam = 3; // read Name |
| 6117 | const signed char Ws2 = 4; // eat whitespaces |
| 6118 | const signed char ExtID = 5; // parse ExternalID |
| 6119 | const signed char ExtIDR = 6; // same as ExtID, but already reported |
| 6120 | const signed char Ws3 = 7; // eat whitespaces |
| 6121 | const signed char Done = 8; |
| 6122 | |
| 6123 | const signed char InpWs = 0; |
| 6124 | const signed char InpGt = 1; // > |
| 6125 | const signed char InpN = 2; // N |
| 6126 | const signed char InpUnknown = 3; |
| 6127 | |
| 6128 | static const signed char table[8][4] = { |
| 6129 | /* InpWs InpGt InpN InpUnknown */ |
| 6130 | { -1, -1, Not, -1 }, // Init |
| 6131 | { Ws1, -1, -1, -1 }, // Not |
| 6132 | { -1, -1, Nam, Nam }, // Ws1 |
| 6133 | { Ws2, Done, -1, -1 }, // Nam |
| 6134 | { -1, Done, ExtID, ExtID }, // Ws2 |
| 6135 | { Ws3, Done, -1, -1 }, // ExtID |
| 6136 | { Ws3, Done, -1, -1 }, // ExtIDR |
| 6137 | { -1, Done, -1, -1 } // Ws3 |
| 6138 | }; |
| 6139 | signed char state; |
| 6140 | signed char input; |
| 6141 | |
| 6142 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 6143 | state = Init; |
| 6144 | } else { |
| 6145 | state = parseStack->pop().state; |
| 6146 | #if defined(QT_QXML_DEBUG) |
| 6147 | qDebug("QXmlSimpleReader: parseNotationDecl (cont) in state %d" , state); |
| 6148 | #endif |
| 6149 | if (!parseStack->isEmpty()) { |
| 6150 | ParseFunction function = parseStack->top().function; |
| 6151 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 6152 | parseStack->pop(); |
| 6153 | #if defined(QT_QXML_DEBUG) |
| 6154 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 6155 | #endif |
| 6156 | } |
| 6157 | if (!(this->*function)()) { |
| 6158 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6159 | return false; |
| 6160 | } |
| 6161 | } |
| 6162 | } |
| 6163 | |
| 6164 | for (;;) { |
| 6165 | switch (state) { |
| 6166 | case ExtID: |
| 6167 | // call the handler |
| 6168 | if (dtdHnd) { |
| 6169 | if (!dtdHnd->notationDecl(name: name(), publicId, systemId)) { |
| 6170 | reportParseError(error: dtdHnd->errorString()); |
| 6171 | return false; |
| 6172 | } |
| 6173 | } |
| 6174 | state = ExtIDR; |
| 6175 | break; |
| 6176 | case Done: |
| 6177 | return true; |
| 6178 | case -1: |
| 6179 | // Error |
| 6180 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 6181 | return false; |
| 6182 | } |
| 6183 | |
| 6184 | if (atEnd()) { |
| 6185 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6186 | return false; |
| 6187 | } |
| 6188 | if (is_S(ch: c)) { |
| 6189 | input = InpWs; |
| 6190 | } else if (c == QLatin1Char('>')) { |
| 6191 | input = InpGt; |
| 6192 | } else if (c == QLatin1Char('N')) { |
| 6193 | input = InpN; |
| 6194 | } else { |
| 6195 | input = InpUnknown; |
| 6196 | } |
| 6197 | state = table[state][input]; |
| 6198 | |
| 6199 | switch (state) { |
| 6200 | case Not: |
| 6201 | parseString_s = QLatin1String("NOTATION" ); |
| 6202 | if (!parseString()) { |
| 6203 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6204 | return false; |
| 6205 | } |
| 6206 | break; |
| 6207 | case Ws1: |
| 6208 | if (!eat_ws()) { |
| 6209 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6210 | return false; |
| 6211 | } |
| 6212 | break; |
| 6213 | case Nam: |
| 6214 | parseName_useRef = false; |
| 6215 | if (!parseName()) { |
| 6216 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6217 | return false; |
| 6218 | } |
| 6219 | break; |
| 6220 | case Ws2: |
| 6221 | if (!eat_ws()) { |
| 6222 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6223 | return false; |
| 6224 | } |
| 6225 | break; |
| 6226 | case ExtID: |
| 6227 | case ExtIDR: |
| 6228 | parseExternalID_allowPublicID = true; |
| 6229 | if (!parseExternalID()) { |
| 6230 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6231 | return false; |
| 6232 | } |
| 6233 | break; |
| 6234 | case Ws3: |
| 6235 | if (!eat_ws()) { |
| 6236 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNotationDecl, state); |
| 6237 | return false; |
| 6238 | } |
| 6239 | break; |
| 6240 | case Done: |
| 6241 | next(); |
| 6242 | break; |
| 6243 | } |
| 6244 | } |
| 6245 | return false; |
| 6246 | } |
| 6247 | |
| 6248 | /* |
| 6249 | Parse choice [49] or seq [50]. |
| 6250 | |
| 6251 | Precondition: the beginning '('S? is already read and the head |
| 6252 | stands on the first non-whitespace character after it. |
| 6253 | */ |
| 6254 | bool QXmlSimpleReaderPrivate::parseChoiceSeq() |
| 6255 | { |
| 6256 | const signed char Init = 0; |
| 6257 | const signed char Ws1 = 1; // eat whitespace |
| 6258 | const signed char CoS = 2; // choice or set |
| 6259 | const signed char Ws2 = 3; // eat whitespace |
| 6260 | const signed char More = 4; // more cp to read |
| 6261 | const signed char Name = 5; // read name |
| 6262 | const signed char Done = 6; // |
| 6263 | |
| 6264 | const signed char InpWs = 0; // S |
| 6265 | const signed char InpOp = 1; // ( |
| 6266 | const signed char InpCp = 2; //) |
| 6267 | const signed char InpQm = 3; // ? |
| 6268 | const signed char InpAst = 4; // * |
| 6269 | const signed char InpPlus = 5; // + |
| 6270 | const signed char InpPipe = 6; // | |
| 6271 | const signed char InpComm = 7; // , |
| 6272 | const signed char InpUnknown = 8; |
| 6273 | |
| 6274 | static const signed char table[6][9] = { |
| 6275 | /* InpWs InpOp InpCp InpQm InpAst InpPlus InpPipe InpComm InpUnknown */ |
| 6276 | { -1, Ws1, -1, -1, -1, -1, -1, -1, Name }, // Init |
| 6277 | { -1, CoS, -1, -1, -1, -1, -1, -1, CoS }, // Ws1 |
| 6278 | { Ws2, -1, Done, Ws2, Ws2, Ws2, More, More, -1 }, // CS |
| 6279 | { -1, -1, Done, -1, -1, -1, More, More, -1 }, // Ws2 |
| 6280 | { -1, Ws1, -1, -1, -1, -1, -1, -1, Name }, // More (same as Init) |
| 6281 | { Ws2, -1, Done, Ws2, Ws2, Ws2, More, More, -1 } // Name (same as CS) |
| 6282 | }; |
| 6283 | signed char state; |
| 6284 | signed char input; |
| 6285 | |
| 6286 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 6287 | state = Init; |
| 6288 | } else { |
| 6289 | state = parseStack->pop().state; |
| 6290 | #if defined(QT_QXML_DEBUG) |
| 6291 | qDebug("QXmlSimpleReader: parseChoiceSeq (cont) in state %d" , state); |
| 6292 | #endif |
| 6293 | if (!parseStack->isEmpty()) { |
| 6294 | ParseFunction function = parseStack->top().function; |
| 6295 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 6296 | parseStack->pop(); |
| 6297 | #if defined(QT_QXML_DEBUG) |
| 6298 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 6299 | #endif |
| 6300 | } |
| 6301 | if (!(this->*function)()) { |
| 6302 | parseFailed(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6303 | return false; |
| 6304 | } |
| 6305 | } |
| 6306 | } |
| 6307 | |
| 6308 | for (;;) { |
| 6309 | switch (state) { |
| 6310 | case Done: |
| 6311 | return true; |
| 6312 | case -1: |
| 6313 | // Error |
| 6314 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 6315 | return false; |
| 6316 | } |
| 6317 | |
| 6318 | if (atEnd()) { |
| 6319 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6320 | return false; |
| 6321 | } |
| 6322 | if (is_S(ch: c)) { |
| 6323 | input = InpWs; |
| 6324 | } else if (c == QLatin1Char('(')) { |
| 6325 | input = InpOp; |
| 6326 | } else if (c == QLatin1Char(')')) { |
| 6327 | input = InpCp; |
| 6328 | } else if (c == QLatin1Char('?')) { |
| 6329 | input = InpQm; |
| 6330 | } else if (c == QLatin1Char('*')) { |
| 6331 | input = InpAst; |
| 6332 | } else if (c == QLatin1Char('+')) { |
| 6333 | input = InpPlus; |
| 6334 | } else if (c == QLatin1Char('|')) { |
| 6335 | input = InpPipe; |
| 6336 | } else if (c == QLatin1Char(',')) { |
| 6337 | input = InpComm; |
| 6338 | } else { |
| 6339 | input = InpUnknown; |
| 6340 | } |
| 6341 | state = table[state][input]; |
| 6342 | |
| 6343 | switch (state) { |
| 6344 | case Ws1: |
| 6345 | if (!next_eat_ws()) { |
| 6346 | parseFailed(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6347 | return false; |
| 6348 | } |
| 6349 | break; |
| 6350 | case CoS: |
| 6351 | if (!parseChoiceSeq()) { |
| 6352 | parseFailed(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6353 | return false; |
| 6354 | } |
| 6355 | break; |
| 6356 | case Ws2: |
| 6357 | if (!next_eat_ws()) { |
| 6358 | parseFailed(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6359 | return false; |
| 6360 | } |
| 6361 | break; |
| 6362 | case More: |
| 6363 | if (!next_eat_ws()) { |
| 6364 | parseFailed(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6365 | return false; |
| 6366 | } |
| 6367 | break; |
| 6368 | case Name: |
| 6369 | parseName_useRef = false; |
| 6370 | if (!parseName()) { |
| 6371 | parseFailed(where: &QXmlSimpleReaderPrivate::parseChoiceSeq, state); |
| 6372 | return false; |
| 6373 | } |
| 6374 | break; |
| 6375 | case Done: |
| 6376 | next(); |
| 6377 | break; |
| 6378 | } |
| 6379 | } |
| 6380 | return false; |
| 6381 | } |
| 6382 | |
| 6383 | bool QXmlSimpleReaderPrivate::isExpandedEntityValueTooLarge(QString *errorMessage) |
| 6384 | { |
| 6385 | QString entityNameBuffer; |
| 6386 | |
| 6387 | // For every entity, check how many times all entity names were referenced in its value. |
| 6388 | for (QMap<QString,QString>::const_iterator toSearchIt = entities.constBegin(); |
| 6389 | toSearchIt != entities.constEnd(); |
| 6390 | ++toSearchIt) { |
| 6391 | const QString &toSearch = toSearchIt.key(); |
| 6392 | |
| 6393 | // Don't check the same entities twice. |
| 6394 | if (!literalEntitySizes.contains(key: toSearch)) { |
| 6395 | // The amount of characters that weren't entity names, but literals, like 'X'. |
| 6396 | QString leftOvers = entities.value(key: toSearch); |
| 6397 | // How many times was entityName referenced by toSearch? |
| 6398 | for (QMap<QString,QString>::const_iterator referencedIt = entities.constBegin(); |
| 6399 | referencedIt != entities.constEnd(); |
| 6400 | ++referencedIt) { |
| 6401 | const QString &entityName = referencedIt.key(); |
| 6402 | |
| 6403 | for (int i = 0; i < leftOvers.size() && i != -1; ) { |
| 6404 | entityNameBuffer = QLatin1Char('&') + entityName + QLatin1Char(';'); |
| 6405 | |
| 6406 | i = leftOvers.indexOf(s: entityNameBuffer, from: i); |
| 6407 | if (i != -1) { |
| 6408 | leftOvers.remove(i, len: entityName.size() + 2); |
| 6409 | // The entityName we're currently trying to find was matched in this string; increase our count. |
| 6410 | ++referencesToOtherEntities[toSearch][entityName]; |
| 6411 | } |
| 6412 | } |
| 6413 | } |
| 6414 | literalEntitySizes[toSearch] = leftOvers.size(); |
| 6415 | } |
| 6416 | } |
| 6417 | |
| 6418 | for (QHash<QString, QHash<QString, int> >::const_iterator entityIt = referencesToOtherEntities.constBegin(); |
| 6419 | entityIt != referencesToOtherEntities.constEnd(); |
| 6420 | ++entityIt) { |
| 6421 | const QString &entity = entityIt.key(); |
| 6422 | |
| 6423 | QHash<QString, int>::iterator expandedIt = expandedSizes.find(key: entity); |
| 6424 | if (expandedIt == expandedSizes.end()) { |
| 6425 | expandedIt = expandedSizes.insert(key: entity, value: literalEntitySizes.value(key: entity)); |
| 6426 | for (QHash<QString, int>::const_iterator referenceIt = entityIt->constBegin(); |
| 6427 | referenceIt != entityIt->constEnd(); |
| 6428 | ++referenceIt) { |
| 6429 | const QString &referenceTo = referenceIt.key(); |
| 6430 | const int references = referencesToOtherEntities.value(key: entity).value(key: referenceTo); |
| 6431 | // The total size of an entity's value is the expanded size of all of its referenced entities, plus its literal size. |
| 6432 | *expandedIt += expandedSizes.value(key: referenceTo) * references + literalEntitySizes.value(key: referenceTo) * references; |
| 6433 | } |
| 6434 | |
| 6435 | if (*expandedIt > entityCharacterLimit) { |
| 6436 | if (errorMessage) { |
| 6437 | *errorMessage = QString::fromLatin1(ba: "The XML entity \"%1\" expands to a string that is too large to process (%2 characters > %3)." ) |
| 6438 | .arg(args: entity, args: QString::number(*expandedIt), args: QString::number(entityCharacterLimit)); |
| 6439 | } |
| 6440 | return true; |
| 6441 | } |
| 6442 | } |
| 6443 | } |
| 6444 | return false; |
| 6445 | } |
| 6446 | |
| 6447 | /* |
| 6448 | Parse a EntityDecl [70]. |
| 6449 | |
| 6450 | Precondition: the beginning '<!E' is already read and the head |
| 6451 | stand on the 'N' of '<!ENTITY' |
| 6452 | */ |
| 6453 | bool QXmlSimpleReaderPrivate::parseEntityDecl() |
| 6454 | { |
| 6455 | const signed char Init = 0; |
| 6456 | const signed char Ent = 1; // parse "ENTITY" |
| 6457 | const signed char Ws1 = 2; // white space read |
| 6458 | const signed char Name = 3; // parse name |
| 6459 | const signed char Ws2 = 4; // white space read |
| 6460 | const signed char EValue = 5; // parse entity value |
| 6461 | const signed char EValueR = 6; // same as EValue, but already reported |
| 6462 | const signed char ExtID = 7; // parse ExternalID |
| 6463 | const signed char Ws3 = 8; // white space read |
| 6464 | const signed char Ndata = 9; // parse "NDATA" |
| 6465 | const signed char Ws4 = 10; // white space read |
| 6466 | const signed char NNam = 11; // parse name |
| 6467 | const signed char NNamR = 12; // same as NNam, but already reported |
| 6468 | const signed char PEDec = 13; // parse PEDecl |
| 6469 | const signed char Ws6 = 14; // white space read |
| 6470 | const signed char PENam = 15; // parse name |
| 6471 | const signed char Ws7 = 16; // white space read |
| 6472 | const signed char PEVal = 17; // parse entity value |
| 6473 | const signed char PEValR = 18; // same as PEVal, but already reported |
| 6474 | const signed char PEEID = 19; // parse ExternalID |
| 6475 | const signed char PEEIDR = 20; // same as PEEID, but already reported |
| 6476 | const signed char WsE = 21; // white space read |
| 6477 | const signed char Done = 22; |
| 6478 | const signed char EDDone = 23; // done, but also report an external, unparsed entity decl |
| 6479 | |
| 6480 | const signed char InpWs = 0; // white space |
| 6481 | const signed char InpPer = 1; // % |
| 6482 | const signed char InpQuot = 2; // " or ' |
| 6483 | const signed char InpGt = 3; // > |
| 6484 | const signed char InpN = 4; // N |
| 6485 | const signed char InpUnknown = 5; |
| 6486 | |
| 6487 | static const signed char table[22][6] = { |
| 6488 | /* InpWs InpPer InpQuot InpGt InpN InpUnknown */ |
| 6489 | { -1, -1, -1, -1, Ent, -1 }, // Init |
| 6490 | { Ws1, -1, -1, -1, -1, -1 }, // Ent |
| 6491 | { -1, PEDec, -1, -1, Name, Name }, // Ws1 |
| 6492 | { Ws2, -1, -1, -1, -1, -1 }, // Name |
| 6493 | { -1, -1, EValue, -1, -1, ExtID }, // Ws2 |
| 6494 | { WsE, -1, -1, Done, -1, -1 }, // EValue |
| 6495 | { WsE, -1, -1, Done, -1, -1 }, // EValueR |
| 6496 | { Ws3, -1, -1, EDDone,-1, -1 }, // ExtID |
| 6497 | { -1, -1, -1, EDDone,Ndata, -1 }, // Ws3 |
| 6498 | { Ws4, -1, -1, -1, -1, -1 }, // Ndata |
| 6499 | { -1, -1, -1, -1, NNam, NNam }, // Ws4 |
| 6500 | { WsE, -1, -1, Done, -1, -1 }, // NNam |
| 6501 | { WsE, -1, -1, Done, -1, -1 }, // NNamR |
| 6502 | { Ws6, -1, -1, -1, -1, -1 }, // PEDec |
| 6503 | { -1, -1, -1, -1, PENam, PENam }, // Ws6 |
| 6504 | { Ws7, -1, -1, -1, -1, -1 }, // PENam |
| 6505 | { -1, -1, PEVal, -1, -1, PEEID }, // Ws7 |
| 6506 | { WsE, -1, -1, Done, -1, -1 }, // PEVal |
| 6507 | { WsE, -1, -1, Done, -1, -1 }, // PEValR |
| 6508 | { WsE, -1, -1, Done, -1, -1 }, // PEEID |
| 6509 | { WsE, -1, -1, Done, -1, -1 }, // PEEIDR |
| 6510 | { -1, -1, -1, Done, -1, -1 } // WsE |
| 6511 | }; |
| 6512 | signed char state; |
| 6513 | signed char input; |
| 6514 | |
| 6515 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 6516 | state = Init; |
| 6517 | } else { |
| 6518 | state = parseStack->pop().state; |
| 6519 | #if defined(QT_QXML_DEBUG) |
| 6520 | qDebug("QXmlSimpleReader: parseEntityDecl (cont) in state %d" , state); |
| 6521 | #endif |
| 6522 | if (!parseStack->isEmpty()) { |
| 6523 | ParseFunction function = parseStack->top().function; |
| 6524 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 6525 | parseStack->pop(); |
| 6526 | #if defined(QT_QXML_DEBUG) |
| 6527 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 6528 | #endif |
| 6529 | } |
| 6530 | if (!(this->*function)()) { |
| 6531 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6532 | return false; |
| 6533 | } |
| 6534 | } |
| 6535 | } |
| 6536 | |
| 6537 | for (;;) { |
| 6538 | switch (state) { |
| 6539 | case EValue: |
| 6540 | if ( !entityExist(name())) { |
| 6541 | QString errorMessage; |
| 6542 | if (isExpandedEntityValueTooLarge(errorMessage: &errorMessage)) { |
| 6543 | reportParseError(error: errorMessage); |
| 6544 | return false; |
| 6545 | } |
| 6546 | |
| 6547 | entities.insert(key: name(), value: string()); |
| 6548 | if (declHnd) { |
| 6549 | if (!declHnd->internalEntityDecl(name: name(), value: string())) { |
| 6550 | reportParseError(error: declHnd->errorString()); |
| 6551 | return false; |
| 6552 | } |
| 6553 | } |
| 6554 | } |
| 6555 | state = EValueR; |
| 6556 | break; |
| 6557 | case NNam: |
| 6558 | if ( !entityExist(name())) { |
| 6559 | externEntities.insert(key: name(), value: QXmlSimpleReaderPrivate::ExternEntity(publicId, systemId, ref())); |
| 6560 | if (dtdHnd) { |
| 6561 | if (!dtdHnd->unparsedEntityDecl(name: name(), publicId, systemId, notationName: ref())) { |
| 6562 | reportParseError(error: declHnd->errorString()); |
| 6563 | return false; |
| 6564 | } |
| 6565 | } |
| 6566 | } |
| 6567 | state = NNamR; |
| 6568 | break; |
| 6569 | case PEVal: |
| 6570 | if ( !entityExist(name())) { |
| 6571 | parameterEntities.insert(key: name(), value: string()); |
| 6572 | if (declHnd) { |
| 6573 | if (!declHnd->internalEntityDecl(name: QLatin1Char('%') + name(), value: string())) { |
| 6574 | reportParseError(error: declHnd->errorString()); |
| 6575 | return false; |
| 6576 | } |
| 6577 | } |
| 6578 | } |
| 6579 | state = PEValR; |
| 6580 | break; |
| 6581 | case PEEID: |
| 6582 | if ( !entityExist(name())) { |
| 6583 | externParameterEntities.insert(key: name(), value: QXmlSimpleReaderPrivate::ExternParameterEntity(publicId, systemId)); |
| 6584 | if (declHnd) { |
| 6585 | if (!declHnd->externalEntityDecl(name: QLatin1Char('%') + name(), publicId, systemId)) { |
| 6586 | reportParseError(error: declHnd->errorString()); |
| 6587 | return false; |
| 6588 | } |
| 6589 | } |
| 6590 | } |
| 6591 | state = PEEIDR; |
| 6592 | break; |
| 6593 | case EDDone: |
| 6594 | if ( !entityExist(name())) { |
| 6595 | externEntities.insert(key: name(), value: QXmlSimpleReaderPrivate::ExternEntity(publicId, systemId, QString())); |
| 6596 | if (declHnd) { |
| 6597 | if (!declHnd->externalEntityDecl(name: name(), publicId, systemId)) { |
| 6598 | reportParseError(error: declHnd->errorString()); |
| 6599 | return false; |
| 6600 | } |
| 6601 | } |
| 6602 | } |
| 6603 | return true; |
| 6604 | case Done: |
| 6605 | return true; |
| 6606 | case -1: |
| 6607 | // Error |
| 6608 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 6609 | return false; |
| 6610 | } |
| 6611 | |
| 6612 | if (atEnd()) { |
| 6613 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6614 | return false; |
| 6615 | } |
| 6616 | if (is_S(ch: c)) { |
| 6617 | input = InpWs; |
| 6618 | } else if (c == QLatin1Char('%')) { |
| 6619 | input = InpPer; |
| 6620 | } else if (c == QLatin1Char('"') || c == QLatin1Char('\'')) { |
| 6621 | input = InpQuot; |
| 6622 | } else if (c == QLatin1Char('>')) { |
| 6623 | input = InpGt; |
| 6624 | } else if (c == QLatin1Char('N')) { |
| 6625 | input = InpN; |
| 6626 | } else { |
| 6627 | input = InpUnknown; |
| 6628 | } |
| 6629 | state = table[state][input]; |
| 6630 | |
| 6631 | switch (state) { |
| 6632 | case Ent: |
| 6633 | parseString_s = QLatin1String("NTITY" ); |
| 6634 | if (!parseString()) { |
| 6635 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6636 | return false; |
| 6637 | } |
| 6638 | break; |
| 6639 | case Ws1: |
| 6640 | if (!eat_ws()) { |
| 6641 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6642 | return false; |
| 6643 | } |
| 6644 | break; |
| 6645 | case Name: |
| 6646 | parseName_useRef = false; |
| 6647 | if (!parseName()) { |
| 6648 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6649 | return false; |
| 6650 | } |
| 6651 | break; |
| 6652 | case Ws2: |
| 6653 | if (!eat_ws()) { |
| 6654 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6655 | return false; |
| 6656 | } |
| 6657 | break; |
| 6658 | case EValue: |
| 6659 | case EValueR: |
| 6660 | if (!parseEntityValue()) { |
| 6661 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6662 | return false; |
| 6663 | } |
| 6664 | break; |
| 6665 | case ExtID: |
| 6666 | parseExternalID_allowPublicID = false; |
| 6667 | if (!parseExternalID()) { |
| 6668 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6669 | return false; |
| 6670 | } |
| 6671 | break; |
| 6672 | case Ws3: |
| 6673 | if (!eat_ws()) { |
| 6674 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6675 | return false; |
| 6676 | } |
| 6677 | break; |
| 6678 | case Ndata: |
| 6679 | parseString_s = QLatin1String("NDATA" ); |
| 6680 | if (!parseString()) { |
| 6681 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6682 | return false; |
| 6683 | } |
| 6684 | break; |
| 6685 | case Ws4: |
| 6686 | if (!eat_ws()) { |
| 6687 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6688 | return false; |
| 6689 | } |
| 6690 | break; |
| 6691 | case NNam: |
| 6692 | case NNamR: |
| 6693 | parseName_useRef = true; |
| 6694 | if (!parseName()) { |
| 6695 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6696 | return false; |
| 6697 | } |
| 6698 | break; |
| 6699 | case PEDec: |
| 6700 | next(); |
| 6701 | break; |
| 6702 | case Ws6: |
| 6703 | if (!eat_ws()) { |
| 6704 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6705 | return false; |
| 6706 | } |
| 6707 | break; |
| 6708 | case PENam: |
| 6709 | parseName_useRef = false; |
| 6710 | if (!parseName()) { |
| 6711 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6712 | return false; |
| 6713 | } |
| 6714 | break; |
| 6715 | case Ws7: |
| 6716 | if (!eat_ws()) { |
| 6717 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6718 | return false; |
| 6719 | } |
| 6720 | break; |
| 6721 | case PEVal: |
| 6722 | case PEValR: |
| 6723 | if (!parseEntityValue()) { |
| 6724 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6725 | return false; |
| 6726 | } |
| 6727 | break; |
| 6728 | case PEEID: |
| 6729 | case PEEIDR: |
| 6730 | parseExternalID_allowPublicID = false; |
| 6731 | if (!parseExternalID()) { |
| 6732 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6733 | return false; |
| 6734 | } |
| 6735 | break; |
| 6736 | case WsE: |
| 6737 | if (!eat_ws()) { |
| 6738 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityDecl, state); |
| 6739 | return false; |
| 6740 | } |
| 6741 | break; |
| 6742 | case EDDone: |
| 6743 | next(); |
| 6744 | break; |
| 6745 | case Done: |
| 6746 | next(); |
| 6747 | break; |
| 6748 | } |
| 6749 | } |
| 6750 | return false; |
| 6751 | } |
| 6752 | |
| 6753 | /* |
| 6754 | Parse a EntityValue [9] |
| 6755 | */ |
| 6756 | bool QXmlSimpleReaderPrivate::parseEntityValue() |
| 6757 | { |
| 6758 | const signed char Init = 0; |
| 6759 | const signed char Dq = 1; // EntityValue is double quoted |
| 6760 | const signed char DqC = 2; // signed character |
| 6761 | const signed char DqPER = 3; // PERefence |
| 6762 | const signed char DqRef = 4; // Reference |
| 6763 | const signed char Sq = 5; // EntityValue is double quoted |
| 6764 | const signed char SqC = 6; // signed character |
| 6765 | const signed char SqPER = 7; // PERefence |
| 6766 | const signed char SqRef = 8; // Reference |
| 6767 | const signed char Done = 9; |
| 6768 | |
| 6769 | const signed char InpDq = 0; // " |
| 6770 | const signed char InpSq = 1; // ' |
| 6771 | const signed char InpAmp = 2; // & |
| 6772 | const signed char InpPer = 3; // % |
| 6773 | const signed char InpUnknown = 4; |
| 6774 | |
| 6775 | static const signed char table[9][5] = { |
| 6776 | /* InpDq InpSq InpAmp InpPer InpUnknown */ |
| 6777 | { Dq, Sq, -1, -1, -1 }, // Init |
| 6778 | { Done, DqC, DqRef, DqPER, DqC }, // Dq |
| 6779 | { Done, DqC, DqRef, DqPER, DqC }, // DqC |
| 6780 | { Done, DqC, DqRef, DqPER, DqC }, // DqPER |
| 6781 | { Done, DqC, DqRef, DqPER, DqC }, // DqRef |
| 6782 | { SqC, Done, SqRef, SqPER, SqC }, // Sq |
| 6783 | { SqC, Done, SqRef, SqPER, SqC }, // SqC |
| 6784 | { SqC, Done, SqRef, SqPER, SqC }, // SqPER |
| 6785 | { SqC, Done, SqRef, SqPER, SqC } // SqRef |
| 6786 | }; |
| 6787 | signed char state; |
| 6788 | signed char input; |
| 6789 | |
| 6790 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 6791 | state = Init; |
| 6792 | } else { |
| 6793 | state = parseStack->pop().state; |
| 6794 | #if defined(QT_QXML_DEBUG) |
| 6795 | qDebug("QXmlSimpleReader: parseEntityValue (cont) in state %d" , state); |
| 6796 | #endif |
| 6797 | if (!parseStack->isEmpty()) { |
| 6798 | ParseFunction function = parseStack->top().function; |
| 6799 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 6800 | parseStack->pop(); |
| 6801 | #if defined(QT_QXML_DEBUG) |
| 6802 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 6803 | #endif |
| 6804 | } |
| 6805 | if (!(this->*function)()) { |
| 6806 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityValue, state); |
| 6807 | return false; |
| 6808 | } |
| 6809 | } |
| 6810 | } |
| 6811 | |
| 6812 | for (;;) { |
| 6813 | switch (state) { |
| 6814 | case Done: |
| 6815 | return true; |
| 6816 | case -1: |
| 6817 | // Error |
| 6818 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 6819 | return false; |
| 6820 | } |
| 6821 | |
| 6822 | if (atEnd()) { |
| 6823 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseEntityValue, state); |
| 6824 | return false; |
| 6825 | } |
| 6826 | if (c == QLatin1Char('"')) { |
| 6827 | input = InpDq; |
| 6828 | } else if (c == QLatin1Char('\'')) { |
| 6829 | input = InpSq; |
| 6830 | } else if (c == QLatin1Char('&')) { |
| 6831 | input = InpAmp; |
| 6832 | } else if (c == QLatin1Char('%')) { |
| 6833 | input = InpPer; |
| 6834 | } else { |
| 6835 | input = InpUnknown; |
| 6836 | } |
| 6837 | state = table[state][input]; |
| 6838 | |
| 6839 | switch (state) { |
| 6840 | case Dq: |
| 6841 | case Sq: |
| 6842 | stringClear(); |
| 6843 | next(); |
| 6844 | break; |
| 6845 | case DqC: |
| 6846 | case SqC: |
| 6847 | stringAddC(); |
| 6848 | next(); |
| 6849 | break; |
| 6850 | case DqPER: |
| 6851 | case SqPER: |
| 6852 | parsePEReference_context = InEntityValue; |
| 6853 | if (!parsePEReference()) { |
| 6854 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityValue, state); |
| 6855 | return false; |
| 6856 | } |
| 6857 | break; |
| 6858 | case DqRef: |
| 6859 | case SqRef: |
| 6860 | parseReference_context = InEntityValue; |
| 6861 | if (!parseReference()) { |
| 6862 | parseFailed(where: &QXmlSimpleReaderPrivate::parseEntityValue, state); |
| 6863 | return false; |
| 6864 | } |
| 6865 | break; |
| 6866 | case Done: |
| 6867 | next(); |
| 6868 | break; |
| 6869 | } |
| 6870 | } |
| 6871 | return false; |
| 6872 | } |
| 6873 | |
| 6874 | /* |
| 6875 | Parse a comment [15]. |
| 6876 | |
| 6877 | Precondition: the beginning '<!' of the comment is already read and the head |
| 6878 | stands on the first '-' of '<!--'. |
| 6879 | |
| 6880 | If this funktion was successful, the head-position is on the first |
| 6881 | character after the comment. |
| 6882 | */ |
| 6883 | bool QXmlSimpleReaderPrivate::() |
| 6884 | { |
| 6885 | const signed char Init = 0; |
| 6886 | const signed char Dash1 = 1; // the first dash was read |
| 6887 | const signed char Dash2 = 2; // the second dash was read |
| 6888 | const signed char Com = 3; // read comment |
| 6889 | const signed char Com2 = 4; // read comment (help state) |
| 6890 | const signed char ComE = 5; // finished reading comment |
| 6891 | const signed char Done = 6; |
| 6892 | |
| 6893 | const signed char InpDash = 0; // - |
| 6894 | const signed char InpGt = 1; // > |
| 6895 | const signed char InpUnknown = 2; |
| 6896 | |
| 6897 | static const signed char table[6][3] = { |
| 6898 | /* InpDash InpGt InpUnknown */ |
| 6899 | { Dash1, -1, -1 }, // Init |
| 6900 | { Dash2, -1, -1 }, // Dash1 |
| 6901 | { Com2, Com, Com }, // Dash2 |
| 6902 | { Com2, Com, Com }, // Com |
| 6903 | { ComE, Com, Com }, // Com2 |
| 6904 | { -1, Done, -1 } // ComE |
| 6905 | }; |
| 6906 | signed char state; |
| 6907 | signed char input; |
| 6908 | |
| 6909 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 6910 | state = Init; |
| 6911 | } else { |
| 6912 | state = parseStack->pop().state; |
| 6913 | #if defined(QT_QXML_DEBUG) |
| 6914 | qDebug("QXmlSimpleReader: parseComment (cont) in state %d" , state); |
| 6915 | #endif |
| 6916 | if (!parseStack->isEmpty()) { |
| 6917 | ParseFunction function = parseStack->top().function; |
| 6918 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 6919 | parseStack->pop(); |
| 6920 | #if defined(QT_QXML_DEBUG) |
| 6921 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 6922 | #endif |
| 6923 | } |
| 6924 | if (!(this->*function)()) { |
| 6925 | parseFailed(where: &QXmlSimpleReaderPrivate::parseComment, state); |
| 6926 | return false; |
| 6927 | } |
| 6928 | } |
| 6929 | } |
| 6930 | |
| 6931 | for (;;) { |
| 6932 | switch (state) { |
| 6933 | case Dash2: |
| 6934 | stringClear(); |
| 6935 | break; |
| 6936 | case Com2: |
| 6937 | // if next character is not a dash than don't skip it |
| 6938 | if (!atEnd() && c != QLatin1Char('-')) |
| 6939 | stringAddC(QLatin1Char('-')); |
| 6940 | break; |
| 6941 | case Done: |
| 6942 | return true; |
| 6943 | case -1: |
| 6944 | // Error |
| 6945 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGCOMMENT)); |
| 6946 | return false; |
| 6947 | } |
| 6948 | |
| 6949 | if (atEnd()) { |
| 6950 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseComment, state); |
| 6951 | return false; |
| 6952 | } |
| 6953 | if (c == QLatin1Char('-')) { |
| 6954 | input = InpDash; |
| 6955 | } else if (c == QLatin1Char('>')) { |
| 6956 | input = InpGt; |
| 6957 | } else { |
| 6958 | input = InpUnknown; |
| 6959 | } |
| 6960 | state = table[state][input]; |
| 6961 | |
| 6962 | switch (state) { |
| 6963 | case Dash1: |
| 6964 | next(); |
| 6965 | break; |
| 6966 | case Dash2: |
| 6967 | next(); |
| 6968 | break; |
| 6969 | case Com: |
| 6970 | stringAddC(); |
| 6971 | next(); |
| 6972 | break; |
| 6973 | case Com2: |
| 6974 | next(); |
| 6975 | break; |
| 6976 | case ComE: |
| 6977 | next(); |
| 6978 | break; |
| 6979 | case Done: |
| 6980 | next(); |
| 6981 | break; |
| 6982 | } |
| 6983 | } |
| 6984 | return false; |
| 6985 | } |
| 6986 | |
| 6987 | /* |
| 6988 | Parse an Attribute [41]. |
| 6989 | |
| 6990 | Precondition: the head stands on the first character of the name |
| 6991 | of the attribute (i.e. all whitespaces are already parsed). |
| 6992 | |
| 6993 | The head stand on the next character after the end quotes. The |
| 6994 | variable name contains the name of the attribute and the variable |
| 6995 | string contains the value of the attribute. |
| 6996 | */ |
| 6997 | bool QXmlSimpleReaderPrivate::parseAttribute() |
| 6998 | { |
| 6999 | const int Init = 0; |
| 7000 | const int PName = 1; // parse name |
| 7001 | const int Ws = 2; // eat ws |
| 7002 | const int Eq = 3; // the '=' was read |
| 7003 | const int Quotes = 4; // " or ' were read |
| 7004 | |
| 7005 | const int InpNameBe = 0; |
| 7006 | const int InpEq = 1; // = |
| 7007 | const int InpDq = 2; // " |
| 7008 | const int InpSq = 3; // ' |
| 7009 | const int InpUnknown = 4; |
| 7010 | |
| 7011 | static const int table[4][5] = { |
| 7012 | /* InpNameBe InpEq InpDq InpSq InpUnknown */ |
| 7013 | { PName, -1, -1, -1, -1 }, // Init |
| 7014 | { -1, Eq, -1, -1, Ws }, // PName |
| 7015 | { -1, Eq, -1, -1, -1 }, // Ws |
| 7016 | { -1, -1, Quotes, Quotes, -1 } // Eq |
| 7017 | }; |
| 7018 | int state; |
| 7019 | int input; |
| 7020 | |
| 7021 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 7022 | state = Init; |
| 7023 | } else { |
| 7024 | state = parseStack->pop().state; |
| 7025 | #if defined(QT_QXML_DEBUG) |
| 7026 | qDebug("QXmlSimpleReader: parseAttribute (cont) in state %d" , state); |
| 7027 | #endif |
| 7028 | if (!parseStack->isEmpty()) { |
| 7029 | ParseFunction function = parseStack->top().function; |
| 7030 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 7031 | parseStack->pop(); |
| 7032 | #if defined(QT_QXML_DEBUG) |
| 7033 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 7034 | #endif |
| 7035 | } |
| 7036 | if (!(this->*function)()) { |
| 7037 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttribute, state); |
| 7038 | return false; |
| 7039 | } |
| 7040 | } |
| 7041 | } |
| 7042 | |
| 7043 | for (;;) { |
| 7044 | switch (state) { |
| 7045 | case Quotes: |
| 7046 | // Done |
| 7047 | return true; |
| 7048 | case -1: |
| 7049 | // Error |
| 7050 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 7051 | return false; |
| 7052 | } |
| 7053 | |
| 7054 | if (atEnd()) { |
| 7055 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseAttribute, state); |
| 7056 | return false; |
| 7057 | } |
| 7058 | if (determineNameChar(ch: c) == NameBeginning) { |
| 7059 | input = InpNameBe; |
| 7060 | } else if (c == QLatin1Char('=')) { |
| 7061 | input = InpEq; |
| 7062 | } else if (c == QLatin1Char('"')) { |
| 7063 | input = InpDq; |
| 7064 | } else if (c == QLatin1Char('\'')) { |
| 7065 | input = InpSq; |
| 7066 | } else { |
| 7067 | input = InpUnknown; |
| 7068 | } |
| 7069 | state = table[state][input]; |
| 7070 | |
| 7071 | switch (state) { |
| 7072 | case PName: |
| 7073 | parseName_useRef = false; |
| 7074 | if (!parseName()) { |
| 7075 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttribute, state); |
| 7076 | return false; |
| 7077 | } |
| 7078 | break; |
| 7079 | case Ws: |
| 7080 | if (!eat_ws()) { |
| 7081 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttribute, state); |
| 7082 | return false; |
| 7083 | } |
| 7084 | break; |
| 7085 | case Eq: |
| 7086 | if (!next_eat_ws()) { |
| 7087 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttribute, state); |
| 7088 | return false; |
| 7089 | } |
| 7090 | break; |
| 7091 | case Quotes: |
| 7092 | if (!parseAttValue()) { |
| 7093 | parseFailed(where: &QXmlSimpleReaderPrivate::parseAttribute, state); |
| 7094 | return false; |
| 7095 | } |
| 7096 | break; |
| 7097 | } |
| 7098 | } |
| 7099 | return false; |
| 7100 | } |
| 7101 | |
| 7102 | /* |
| 7103 | Parse a Name [5] and store the name in name or ref (if useRef is true). |
| 7104 | */ |
| 7105 | bool QXmlSimpleReaderPrivate::parseName() |
| 7106 | { |
| 7107 | const int Init = 0; |
| 7108 | const int Name1 = 1; // parse first character of the name |
| 7109 | const int Name = 2; // parse name |
| 7110 | const int Done = 3; |
| 7111 | |
| 7112 | static const int table[3][3] = { |
| 7113 | /* InpNameBe InpNameCh InpUnknown */ |
| 7114 | { Name1, -1, -1 }, // Init |
| 7115 | { Name, Name, Done }, // Name1 |
| 7116 | { Name, Name, Done } // Name |
| 7117 | }; |
| 7118 | int state; |
| 7119 | |
| 7120 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 7121 | state = Init; |
| 7122 | } else { |
| 7123 | state = parseStack->pop().state; |
| 7124 | #if defined(QT_QXML_DEBUG) |
| 7125 | qDebug("QXmlSimpleReader: parseName (cont) in state %d" , state); |
| 7126 | #endif |
| 7127 | if (!parseStack->isEmpty()) { |
| 7128 | ParseFunction function = parseStack->top().function; |
| 7129 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 7130 | parseStack->pop(); |
| 7131 | #if defined(QT_QXML_DEBUG) |
| 7132 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 7133 | #endif |
| 7134 | } |
| 7135 | if (!(this->*function)()) { |
| 7136 | parseFailed(where: &QXmlSimpleReaderPrivate::parseName, state); |
| 7137 | return false; |
| 7138 | } |
| 7139 | } |
| 7140 | } |
| 7141 | |
| 7142 | for (;;) { |
| 7143 | switch (state) { |
| 7144 | case Done: |
| 7145 | return true; |
| 7146 | case -1: |
| 7147 | // Error |
| 7148 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 7149 | return false; |
| 7150 | } |
| 7151 | |
| 7152 | if (atEnd()) { |
| 7153 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseName, state); |
| 7154 | return false; |
| 7155 | } |
| 7156 | |
| 7157 | // we can safely do the (int) cast thanks to the Q_ASSERTs earlier in this function |
| 7158 | state = table[state][(int)fastDetermineNameChar(ch: c)]; |
| 7159 | |
| 7160 | switch (state) { |
| 7161 | case Name1: |
| 7162 | if (parseName_useRef) { |
| 7163 | refClear(); |
| 7164 | refAddC(); |
| 7165 | } else { |
| 7166 | nameClear(); |
| 7167 | nameAddC(); |
| 7168 | } |
| 7169 | next(); |
| 7170 | break; |
| 7171 | case Name: |
| 7172 | if (parseName_useRef) { |
| 7173 | refAddC(); |
| 7174 | } else { |
| 7175 | nameAddC(); |
| 7176 | } |
| 7177 | next(); |
| 7178 | break; |
| 7179 | } |
| 7180 | } |
| 7181 | return false; |
| 7182 | } |
| 7183 | |
| 7184 | /* |
| 7185 | Parse a Nmtoken [7] and store the name in name. |
| 7186 | */ |
| 7187 | bool QXmlSimpleReaderPrivate::parseNmtoken() |
| 7188 | { |
| 7189 | const signed char Init = 0; |
| 7190 | const signed char NameF = 1; |
| 7191 | const signed char Name = 2; |
| 7192 | const signed char Done = 3; |
| 7193 | |
| 7194 | const signed char InpNameCh = 0; // NameChar without InpNameBe |
| 7195 | const signed char InpUnknown = 1; |
| 7196 | |
| 7197 | static const signed char table[3][2] = { |
| 7198 | /* InpNameCh InpUnknown */ |
| 7199 | { NameF, -1 }, // Init |
| 7200 | { Name, Done }, // NameF |
| 7201 | { Name, Done } // Name |
| 7202 | }; |
| 7203 | signed char state; |
| 7204 | signed char input; |
| 7205 | |
| 7206 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 7207 | state = Init; |
| 7208 | } else { |
| 7209 | state = parseStack->pop().state; |
| 7210 | #if defined(QT_QXML_DEBUG) |
| 7211 | qDebug("QXmlSimpleReader: parseNmtoken (cont) in state %d" , state); |
| 7212 | #endif |
| 7213 | if (!parseStack->isEmpty()) { |
| 7214 | ParseFunction function = parseStack->top().function; |
| 7215 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 7216 | parseStack->pop(); |
| 7217 | #if defined(QT_QXML_DEBUG) |
| 7218 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 7219 | #endif |
| 7220 | } |
| 7221 | if (!(this->*function)()) { |
| 7222 | parseFailed(where: &QXmlSimpleReaderPrivate::parseNmtoken, state); |
| 7223 | return false; |
| 7224 | } |
| 7225 | } |
| 7226 | } |
| 7227 | |
| 7228 | for (;;) { |
| 7229 | switch (state) { |
| 7230 | case Done: |
| 7231 | return true; |
| 7232 | case -1: |
| 7233 | // Error |
| 7234 | reportParseError(error: QLatin1String(XMLERR_LETTEREXPECTED)); |
| 7235 | return false; |
| 7236 | } |
| 7237 | |
| 7238 | if (atEnd()) { |
| 7239 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseNmtoken, state); |
| 7240 | return false; |
| 7241 | } |
| 7242 | if (determineNameChar(ch: c) == NotName) { |
| 7243 | input = InpUnknown; |
| 7244 | } else { |
| 7245 | input = InpNameCh; |
| 7246 | } |
| 7247 | state = table[state][input]; |
| 7248 | |
| 7249 | switch (state) { |
| 7250 | case NameF: |
| 7251 | nameClear(); |
| 7252 | nameAddC(); |
| 7253 | next(); |
| 7254 | break; |
| 7255 | case Name: |
| 7256 | nameAddC(); |
| 7257 | next(); |
| 7258 | break; |
| 7259 | } |
| 7260 | } |
| 7261 | return false; |
| 7262 | } |
| 7263 | |
| 7264 | /* |
| 7265 | Parse a Reference [67]. |
| 7266 | |
| 7267 | parseReference_charDataRead is set to true if the reference must not be |
| 7268 | parsed. The character(s) which the reference mapped to are appended to |
| 7269 | string. The head stands on the first character after the reference. |
| 7270 | |
| 7271 | parseReference_charDataRead is set to false if the reference must be parsed. |
| 7272 | The character(s) which the reference mapped to are inserted at the reference |
| 7273 | position. The head stands on the first character of the replacement). |
| 7274 | */ |
| 7275 | bool QXmlSimpleReaderPrivate::parseReference() |
| 7276 | { |
| 7277 | // temporary variables (only used in very local context, so they don't |
| 7278 | // interfere with incremental parsing) |
| 7279 | uint tmp; |
| 7280 | bool ok; |
| 7281 | |
| 7282 | const signed char Init = 0; |
| 7283 | const signed char SRef = 1; // start of a reference |
| 7284 | const signed char ChRef = 2; // parse CharRef |
| 7285 | const signed char ChDec = 3; // parse CharRef decimal |
| 7286 | const signed char ChHexS = 4; // start CharRef hexadecimal |
| 7287 | const signed char ChHex = 5; // parse CharRef hexadecimal |
| 7288 | const signed char Name = 6; // parse name |
| 7289 | const signed char DoneD = 7; // done CharRef decimal |
| 7290 | const signed char DoneH = 8; // done CharRef hexadecimal |
| 7291 | const signed char DoneN = 9; // done EntityRef |
| 7292 | |
| 7293 | const signed char InpAmp = 0; // & |
| 7294 | const signed char InpSemi = 1; // ; |
| 7295 | const signed char InpHash = 2; // # |
| 7296 | const signed char InpX = 3; // x |
| 7297 | const signed char InpNum = 4; // 0-9 |
| 7298 | const signed char InpHex = 5; // a-f A-F |
| 7299 | const signed char InpUnknown = 6; |
| 7300 | |
| 7301 | static const signed char table[8][7] = { |
| 7302 | /* InpAmp InpSemi InpHash InpX InpNum InpHex InpUnknown */ |
| 7303 | { SRef, -1, -1, -1, -1, -1, -1 }, // Init |
| 7304 | { -1, -1, ChRef, Name, Name, Name, Name }, // SRef |
| 7305 | { -1, -1, -1, ChHexS, ChDec, -1, -1 }, // ChRef |
| 7306 | { -1, DoneD, -1, -1, ChDec, -1, -1 }, // ChDec |
| 7307 | { -1, -1, -1, -1, ChHex, ChHex, -1 }, // ChHexS |
| 7308 | { -1, DoneH, -1, -1, ChHex, ChHex, -1 }, // ChHex |
| 7309 | { -1, DoneN, -1, -1, -1, -1, -1 } // Name |
| 7310 | }; |
| 7311 | signed char state; |
| 7312 | signed char input; |
| 7313 | |
| 7314 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 7315 | parseReference_charDataRead = false; |
| 7316 | state = Init; |
| 7317 | } else { |
| 7318 | state = parseStack->pop().state; |
| 7319 | #if defined(QT_QXML_DEBUG) |
| 7320 | qDebug("QXmlSimpleReader: parseReference (cont) in state %d" , state); |
| 7321 | #endif |
| 7322 | if (!parseStack->isEmpty()) { |
| 7323 | ParseFunction function = parseStack->top().function; |
| 7324 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 7325 | parseStack->pop(); |
| 7326 | #if defined(QT_QXML_DEBUG) |
| 7327 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 7328 | #endif |
| 7329 | } |
| 7330 | if (!(this->*function)()) { |
| 7331 | parseFailed(where: &QXmlSimpleReaderPrivate::parseReference, state); |
| 7332 | return false; |
| 7333 | } |
| 7334 | } |
| 7335 | } |
| 7336 | |
| 7337 | for (;;) { |
| 7338 | switch (state) { |
| 7339 | case DoneD: |
| 7340 | return true; |
| 7341 | case DoneH: |
| 7342 | return true; |
| 7343 | case DoneN: |
| 7344 | return true; |
| 7345 | case -1: |
| 7346 | // Error |
| 7347 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGREFERENCE)); |
| 7348 | return false; |
| 7349 | } |
| 7350 | |
| 7351 | if (atEnd()) { |
| 7352 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseReference, state); |
| 7353 | return false; |
| 7354 | } |
| 7355 | if (c.row()) { |
| 7356 | input = InpUnknown; |
| 7357 | } else if (c.cell() == '&') { |
| 7358 | input = InpAmp; |
| 7359 | } else if (c.cell() == ';') { |
| 7360 | input = InpSemi; |
| 7361 | } else if (c.cell() == '#') { |
| 7362 | input = InpHash; |
| 7363 | } else if (c.cell() == 'x') { |
| 7364 | input = InpX; |
| 7365 | } else if ('0' <= c.cell() && c.cell() <= '9') { |
| 7366 | input = InpNum; |
| 7367 | } else if ('a' <= c.cell() && c.cell() <= 'f') { |
| 7368 | input = InpHex; |
| 7369 | } else if ('A' <= c.cell() && c.cell() <= 'F') { |
| 7370 | input = InpHex; |
| 7371 | } else { |
| 7372 | input = InpUnknown; |
| 7373 | } |
| 7374 | state = table[state][input]; |
| 7375 | |
| 7376 | switch (state) { |
| 7377 | case SRef: |
| 7378 | refClear(); |
| 7379 | next(); |
| 7380 | break; |
| 7381 | case ChRef: |
| 7382 | next(); |
| 7383 | break; |
| 7384 | case ChDec: |
| 7385 | refAddC(); |
| 7386 | next(); |
| 7387 | break; |
| 7388 | case ChHexS: |
| 7389 | next(); |
| 7390 | break; |
| 7391 | case ChHex: |
| 7392 | refAddC(); |
| 7393 | next(); |
| 7394 | break; |
| 7395 | case Name: |
| 7396 | // read the name into the ref |
| 7397 | parseName_useRef = true; |
| 7398 | if (!parseName()) { |
| 7399 | parseFailed(where: &QXmlSimpleReaderPrivate::parseReference, state); |
| 7400 | return false; |
| 7401 | } |
| 7402 | break; |
| 7403 | case DoneD: |
| 7404 | tmp = ref().toUInt(ok: &ok, base: 10); |
| 7405 | if (ok) { |
| 7406 | if (tmp > 0xffff) { |
| 7407 | stringAddC(QChar::highSurrogate(ucs4: tmp)); |
| 7408 | stringAddC(QChar::lowSurrogate(ucs4: tmp)); |
| 7409 | } else { |
| 7410 | stringAddC(QChar(tmp)); |
| 7411 | } |
| 7412 | } else { |
| 7413 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGREFERENCE)); |
| 7414 | return false; |
| 7415 | } |
| 7416 | parseReference_charDataRead = true; |
| 7417 | next(); |
| 7418 | break; |
| 7419 | case DoneH: |
| 7420 | tmp = ref().toUInt(ok: &ok, base: 16); |
| 7421 | if (ok) { |
| 7422 | if (tmp > 0xffff) { |
| 7423 | stringAddC(QChar::highSurrogate(ucs4: tmp)); |
| 7424 | stringAddC(QChar::lowSurrogate(ucs4: tmp)); |
| 7425 | } else { |
| 7426 | stringAddC(QChar(tmp)); |
| 7427 | } |
| 7428 | } else { |
| 7429 | reportParseError(error: QLatin1String(XMLERR_ERRORPARSINGREFERENCE)); |
| 7430 | return false; |
| 7431 | } |
| 7432 | parseReference_charDataRead = true; |
| 7433 | next(); |
| 7434 | break; |
| 7435 | case DoneN: |
| 7436 | if (!processReference()) |
| 7437 | return false; |
| 7438 | next(); |
| 7439 | break; |
| 7440 | } |
| 7441 | } |
| 7442 | return false; |
| 7443 | } |
| 7444 | |
| 7445 | /* |
| 7446 | Helper function for parseReference() |
| 7447 | */ |
| 7448 | bool QXmlSimpleReaderPrivate::processReference() |
| 7449 | { |
| 7450 | QString reference = ref(); |
| 7451 | if (reference == QLatin1String("amp" )) { |
| 7452 | if (parseReference_context == InEntityValue) { |
| 7453 | // Bypassed |
| 7454 | stringAddC(QLatin1Char('&')); stringAddC(QLatin1Char('a')); stringAddC(QLatin1Char('m')); stringAddC(QLatin1Char('p')); stringAddC(QLatin1Char(';')); |
| 7455 | } else { |
| 7456 | // Included or Included in literal |
| 7457 | stringAddC(QLatin1Char('&')); |
| 7458 | } |
| 7459 | parseReference_charDataRead = true; |
| 7460 | } else if (reference == QLatin1String("lt" )) { |
| 7461 | if (parseReference_context == InEntityValue) { |
| 7462 | // Bypassed |
| 7463 | stringAddC(QLatin1Char('&')); stringAddC(QLatin1Char('l')); stringAddC(QLatin1Char('t')); stringAddC(QLatin1Char(';')); |
| 7464 | } else { |
| 7465 | // Included or Included in literal |
| 7466 | stringAddC(QLatin1Char('<')); |
| 7467 | } |
| 7468 | parseReference_charDataRead = true; |
| 7469 | } else if (reference == QLatin1String("gt" )) { |
| 7470 | if (parseReference_context == InEntityValue) { |
| 7471 | // Bypassed |
| 7472 | stringAddC(QLatin1Char('&')); stringAddC(QLatin1Char('g')); stringAddC(QLatin1Char('t')); stringAddC(QLatin1Char(';')); |
| 7473 | } else { |
| 7474 | // Included or Included in literal |
| 7475 | stringAddC(QLatin1Char('>')); |
| 7476 | } |
| 7477 | parseReference_charDataRead = true; |
| 7478 | } else if (reference == QLatin1String("apos" )) { |
| 7479 | if (parseReference_context == InEntityValue) { |
| 7480 | // Bypassed |
| 7481 | stringAddC(QLatin1Char('&')); stringAddC(QLatin1Char('a')); stringAddC(QLatin1Char('p')); stringAddC(QLatin1Char('o')); stringAddC(QLatin1Char('s')); stringAddC(QLatin1Char(';')); |
| 7482 | } else { |
| 7483 | // Included or Included in literal |
| 7484 | stringAddC(QLatin1Char('\'')); |
| 7485 | } |
| 7486 | parseReference_charDataRead = true; |
| 7487 | } else if (reference == QLatin1String("quot" )) { |
| 7488 | if (parseReference_context == InEntityValue) { |
| 7489 | // Bypassed |
| 7490 | stringAddC(QLatin1Char('&')); stringAddC(QLatin1Char('q')); stringAddC(QLatin1Char('u')); stringAddC(QLatin1Char('o')); stringAddC(QLatin1Char('t')); stringAddC(QLatin1Char(';')); |
| 7491 | } else { |
| 7492 | // Included or Included in literal |
| 7493 | stringAddC(QLatin1Char('"')); |
| 7494 | } |
| 7495 | parseReference_charDataRead = true; |
| 7496 | } else { |
| 7497 | QMap<QString,QString>::Iterator it; |
| 7498 | it = entities.find(key: reference); |
| 7499 | if (it != entities.end()) { |
| 7500 | // "Internal General" |
| 7501 | switch (parseReference_context) { |
| 7502 | case InContent: |
| 7503 | // Included |
| 7504 | if (!insertXmlRef(*it, reference, false)) |
| 7505 | return false; |
| 7506 | parseReference_charDataRead = false; |
| 7507 | break; |
| 7508 | case InAttributeValue: |
| 7509 | // Included in literal |
| 7510 | if (!insertXmlRef(*it, reference, true)) |
| 7511 | return false; |
| 7512 | parseReference_charDataRead = false; |
| 7513 | break; |
| 7514 | case InEntityValue: |
| 7515 | { |
| 7516 | // Bypassed |
| 7517 | stringAddC(QLatin1Char('&')); |
| 7518 | for (int i=0; i<(int)reference.size(); i++) { |
| 7519 | stringAddC(reference[i]); |
| 7520 | } |
| 7521 | stringAddC(QLatin1Char(';')); |
| 7522 | parseReference_charDataRead = true; |
| 7523 | } |
| 7524 | break; |
| 7525 | case InDTD: |
| 7526 | // Forbidden |
| 7527 | parseReference_charDataRead = false; |
| 7528 | reportParseError(error: QLatin1String(XMLERR_INTERNALGENERALENTITYINDTD)); |
| 7529 | return false; |
| 7530 | } |
| 7531 | } else { |
| 7532 | QMap<QString,QXmlSimpleReaderPrivate::ExternEntity>::Iterator itExtern; |
| 7533 | itExtern = externEntities.find(key: reference); |
| 7534 | if (itExtern == externEntities.end()) { |
| 7535 | // entity not declared |
| 7536 | // ### check this case for conformance |
| 7537 | if (parseReference_context == InEntityValue) { |
| 7538 | // Bypassed |
| 7539 | stringAddC(QLatin1Char('&')); |
| 7540 | for (int i=0; i<(int)reference.size(); i++) { |
| 7541 | stringAddC(reference[i]); |
| 7542 | } |
| 7543 | stringAddC(QLatin1Char(';')); |
| 7544 | parseReference_charDataRead = true; |
| 7545 | } else { |
| 7546 | // if we have some char data read, report it now |
| 7547 | if (parseReference_context == InContent) { |
| 7548 | if (contentCharDataRead) { |
| 7549 | if (reportWhitespaceCharData || !string().simplified().isEmpty()) { |
| 7550 | if (contentHnd != nullptr && !contentHnd->characters(ch: string())) { |
| 7551 | reportParseError(error: contentHnd->errorString()); |
| 7552 | return false; |
| 7553 | } |
| 7554 | } |
| 7555 | stringClear(); |
| 7556 | contentCharDataRead = false; |
| 7557 | } |
| 7558 | } |
| 7559 | |
| 7560 | if (contentHnd) { |
| 7561 | skipped_entity_in_content = parseReference_context == InContent; |
| 7562 | if (!contentHnd->skippedEntity(name: reference)) { |
| 7563 | skipped_entity_in_content = false; |
| 7564 | reportParseError(error: contentHnd->errorString()); |
| 7565 | return false; // error |
| 7566 | } |
| 7567 | skipped_entity_in_content = false; |
| 7568 | } |
| 7569 | } |
| 7570 | } else if ((*itExtern).notation.isNull()) { |
| 7571 | // "External Parsed General" |
| 7572 | switch (parseReference_context) { |
| 7573 | case InContent: |
| 7574 | { |
| 7575 | // Included if validating |
| 7576 | bool skipIt = true; |
| 7577 | if (entityRes) { |
| 7578 | QXmlInputSource *ret = nullptr; |
| 7579 | if (!entityRes->resolveEntity(publicId: (*itExtern).publicId, systemId: (*itExtern).systemId, ret)) { |
| 7580 | delete ret; |
| 7581 | reportParseError(error: entityRes->errorString()); |
| 7582 | return false; |
| 7583 | } |
| 7584 | if (ret) { |
| 7585 | QString xmlRefString; |
| 7586 | QString buffer = ret->data(); |
| 7587 | while (!buffer.isEmpty()) { |
| 7588 | xmlRefString += buffer; |
| 7589 | ret->fetchData(); |
| 7590 | buffer = ret->data(); |
| 7591 | } |
| 7592 | |
| 7593 | delete ret; |
| 7594 | if (!stripTextDecl(str&: xmlRefString)) { |
| 7595 | reportParseError(error: QLatin1String(XMLERR_ERRORINTEXTDECL)); |
| 7596 | return false; |
| 7597 | } |
| 7598 | if (!insertXmlRef(xmlRefString, reference, false)) |
| 7599 | return false; |
| 7600 | skipIt = false; |
| 7601 | } |
| 7602 | } |
| 7603 | if (skipIt && contentHnd) { |
| 7604 | skipped_entity_in_content = true; |
| 7605 | if (!contentHnd->skippedEntity(name: reference)) { |
| 7606 | skipped_entity_in_content = false; |
| 7607 | reportParseError(error: contentHnd->errorString()); |
| 7608 | return false; // error |
| 7609 | } |
| 7610 | skipped_entity_in_content = false; |
| 7611 | } |
| 7612 | parseReference_charDataRead = false; |
| 7613 | } break; |
| 7614 | case InAttributeValue: |
| 7615 | // Forbidden |
| 7616 | parseReference_charDataRead = false; |
| 7617 | reportParseError(error: QLatin1String(XMLERR_EXTERNALGENERALENTITYINAV)); |
| 7618 | return false; |
| 7619 | case InEntityValue: |
| 7620 | { |
| 7621 | // Bypassed |
| 7622 | stringAddC(QLatin1Char('&')); |
| 7623 | for (int i=0; i<(int)reference.size(); i++) { |
| 7624 | stringAddC(reference[i]); |
| 7625 | } |
| 7626 | stringAddC(QLatin1Char(';')); |
| 7627 | parseReference_charDataRead = true; |
| 7628 | } |
| 7629 | break; |
| 7630 | case InDTD: |
| 7631 | // Forbidden |
| 7632 | parseReference_charDataRead = false; |
| 7633 | reportParseError(error: QLatin1String(XMLERR_EXTERNALGENERALENTITYINDTD)); |
| 7634 | return false; |
| 7635 | } |
| 7636 | } else { |
| 7637 | // "Unparsed" |
| 7638 | // ### notify for "Occurs as Attribute Value" missing (but this is no reference, anyway) |
| 7639 | // Forbidden |
| 7640 | parseReference_charDataRead = false; |
| 7641 | reportParseError(error: QLatin1String(XMLERR_UNPARSEDENTITYREFERENCE)); |
| 7642 | return false; // error |
| 7643 | } |
| 7644 | } |
| 7645 | } |
| 7646 | return true; // no error |
| 7647 | } |
| 7648 | |
| 7649 | /* |
| 7650 | Parses over a simple string. |
| 7651 | |
| 7652 | After the string was successfully parsed, the head is on the first |
| 7653 | character after the string. |
| 7654 | */ |
| 7655 | bool QXmlSimpleReaderPrivate::parseString() |
| 7656 | { |
| 7657 | const signed char InpCharExpected = 0; // the character that was expected |
| 7658 | const signed char InpUnknown = 1; |
| 7659 | |
| 7660 | signed char state; // state in this function is the position in the string s |
| 7661 | signed char input; |
| 7662 | |
| 7663 | if (parseStack == nullptr || parseStack->isEmpty()) { |
| 7664 | Done = parseString_s.size(); |
| 7665 | state = 0; |
| 7666 | } else { |
| 7667 | state = parseStack->pop().state; |
| 7668 | #if defined(QT_QXML_DEBUG) |
| 7669 | qDebug("QXmlSimpleReader: parseString (cont) in state %d" , state); |
| 7670 | #endif |
| 7671 | if (!parseStack->isEmpty()) { |
| 7672 | ParseFunction function = parseStack->top().function; |
| 7673 | if (function == &QXmlSimpleReaderPrivate::eat_ws) { |
| 7674 | parseStack->pop(); |
| 7675 | #if defined(QT_QXML_DEBUG) |
| 7676 | qDebug("QXmlSimpleReader: eat_ws (cont)" ); |
| 7677 | #endif |
| 7678 | } |
| 7679 | if (!(this->*function)()) { |
| 7680 | parseFailed(where: &QXmlSimpleReaderPrivate::parseString, state); |
| 7681 | return false; |
| 7682 | } |
| 7683 | } |
| 7684 | } |
| 7685 | |
| 7686 | for (;;) { |
| 7687 | if (state == Done) { |
| 7688 | return true; |
| 7689 | } |
| 7690 | |
| 7691 | if (atEnd()) { |
| 7692 | unexpectedEof(where: &QXmlSimpleReaderPrivate::parseString, state); |
| 7693 | return false; |
| 7694 | } |
| 7695 | if (c == parseString_s[(int)state]) { |
| 7696 | input = InpCharExpected; |
| 7697 | } else { |
| 7698 | input = InpUnknown; |
| 7699 | } |
| 7700 | if (input == InpCharExpected) { |
| 7701 | state++; |
| 7702 | } else { |
| 7703 | // Error |
| 7704 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDCHARACTER)); |
| 7705 | return false; |
| 7706 | } |
| 7707 | |
| 7708 | next(); |
| 7709 | } |
| 7710 | return false; |
| 7711 | } |
| 7712 | |
| 7713 | /* |
| 7714 | This private function inserts and reports an entity substitution. The |
| 7715 | substituted string is \a data and the name of the entity reference is \a |
| 7716 | name. If \a inLiteral is true, the entity is IncludedInLiteral (i.e., " and ' |
| 7717 | must be quoted. Otherwise they are not quoted. |
| 7718 | |
| 7719 | This function returns \c false on error. |
| 7720 | */ |
| 7721 | bool QXmlSimpleReaderPrivate::insertXmlRef(const QString &data, const QString &name, bool inLiteral) |
| 7722 | { |
| 7723 | if (inLiteral) { |
| 7724 | QString tmp = data; |
| 7725 | xmlRefStack.push(t: XmlRef(name, tmp.replace(c: QLatin1Char('\"'), |
| 7726 | after: QLatin1String(""" )).replace(c: QLatin1Char('\''), after: QLatin1String("'" )))); |
| 7727 | } else { |
| 7728 | xmlRefStack.push(t: XmlRef(name, data)); |
| 7729 | } |
| 7730 | int n = qMax(a: parameterEntities.size(), b: entities.size()); |
| 7731 | if (xmlRefStack.size() > n+1) { |
| 7732 | // recursive entities |
| 7733 | reportParseError(error: QLatin1String(XMLERR_RECURSIVEENTITIES)); |
| 7734 | return false; |
| 7735 | } |
| 7736 | if (reportEntities && lexicalHnd) { |
| 7737 | if (!lexicalHnd->startEntity(name)) { |
| 7738 | reportParseError(error: lexicalHnd->errorString()); |
| 7739 | return false; |
| 7740 | } |
| 7741 | } |
| 7742 | return true; |
| 7743 | } |
| 7744 | |
| 7745 | /* |
| 7746 | This private function moves the cursor to the next character. |
| 7747 | */ |
| 7748 | void QXmlSimpleReaderPrivate::next() |
| 7749 | { |
| 7750 | int count = xmlRefStack.size(); |
| 7751 | while (count != 0) { |
| 7752 | if (xmlRefStack.top().isEmpty()) { |
| 7753 | xmlRefStack.pop_back(); |
| 7754 | count--; |
| 7755 | } else { |
| 7756 | c = xmlRefStack.top().next(); |
| 7757 | return; |
| 7758 | } |
| 7759 | } |
| 7760 | |
| 7761 | // the following could be written nicer, but since it is a time-critical |
| 7762 | // function, rather optimize for speed |
| 7763 | ushort uc = c.unicode(); |
| 7764 | c = inputSource->next(); |
| 7765 | // If we are not incremental parsing, we just skip over EndOfData chars to give the |
| 7766 | // parser an uninterrupted stream of document chars. |
| 7767 | if (c == QXmlInputSource::EndOfData && parseStack == nullptr) |
| 7768 | c = inputSource->next(); |
| 7769 | if (uc == '\n') { |
| 7770 | lineNr++; |
| 7771 | columnNr = -1; |
| 7772 | } else if (uc == '\r') { |
| 7773 | if (c != QLatin1Char('\n')) { |
| 7774 | lineNr++; |
| 7775 | columnNr = -1; |
| 7776 | } |
| 7777 | } |
| 7778 | ++columnNr; |
| 7779 | } |
| 7780 | |
| 7781 | /* |
| 7782 | This private function moves the cursor to the next non-whitespace character. |
| 7783 | This function does not move the cursor if the actual cursor position is a |
| 7784 | non-whitespace character. |
| 7785 | |
| 7786 | Returns \c false when you use incremental parsing and this function reaches EOF |
| 7787 | with reading only whitespace characters. In this case it also populates the |
| 7788 | parseStack with useful information. In all other cases, this function returns |
| 7789 | true. |
| 7790 | */ |
| 7791 | bool QXmlSimpleReaderPrivate::eat_ws() |
| 7792 | { |
| 7793 | while (!atEnd()) { |
| 7794 | if (!is_S(ch: c)) { |
| 7795 | return true; |
| 7796 | } |
| 7797 | next(); |
| 7798 | } |
| 7799 | if (parseStack != nullptr) { |
| 7800 | unexpectedEof(where: &QXmlSimpleReaderPrivate::eat_ws, state: 0); |
| 7801 | return false; |
| 7802 | } |
| 7803 | return true; |
| 7804 | } |
| 7805 | |
| 7806 | bool QXmlSimpleReaderPrivate::next_eat_ws() |
| 7807 | { |
| 7808 | next(); |
| 7809 | return eat_ws(); |
| 7810 | } |
| 7811 | |
| 7812 | /* |
| 7813 | This private function initializes the reader. \a i is the input source to |
| 7814 | read the data from. |
| 7815 | */ |
| 7816 | void QXmlSimpleReaderPrivate::init(const QXmlInputSource *i) |
| 7817 | { |
| 7818 | lineNr = 0; |
| 7819 | columnNr = -1; |
| 7820 | inputSource = const_cast<QXmlInputSource *>(i); |
| 7821 | initData(); |
| 7822 | |
| 7823 | externParameterEntities.clear(); |
| 7824 | parameterEntities.clear(); |
| 7825 | externEntities.clear(); |
| 7826 | entities.clear(); |
| 7827 | |
| 7828 | clear(c&: tags); |
| 7829 | |
| 7830 | doctype.clear(); |
| 7831 | xmlVersion.clear(); |
| 7832 | encoding.clear(); |
| 7833 | standalone = QXmlSimpleReaderPrivate::Unknown; |
| 7834 | error.clear(); |
| 7835 | } |
| 7836 | |
| 7837 | /* |
| 7838 | This private function initializes the XML data related variables. Especially, |
| 7839 | it reads the data from the input source. |
| 7840 | */ |
| 7841 | void QXmlSimpleReaderPrivate::initData() |
| 7842 | { |
| 7843 | c = QXmlInputSource::EndOfData; |
| 7844 | xmlRefStack.clear(); |
| 7845 | next(); |
| 7846 | } |
| 7847 | |
| 7848 | /* |
| 7849 | Returns \c true if a entity with the name \a e exists, |
| 7850 | otherwise returns \c false. |
| 7851 | */ |
| 7852 | bool QXmlSimpleReaderPrivate::entityExist(const QString& e) const |
| 7853 | { |
| 7854 | if ( parameterEntities.find(key: e) == parameterEntities.end() && |
| 7855 | externParameterEntities.find(key: e) == externParameterEntities.end() && |
| 7856 | externEntities.find(key: e) == externEntities.end() && |
| 7857 | entities.find(key: e) == entities.end()) { |
| 7858 | return false; |
| 7859 | } else { |
| 7860 | return true; |
| 7861 | } |
| 7862 | } |
| 7863 | |
| 7864 | void QXmlSimpleReaderPrivate::reportParseError(const QString& error) |
| 7865 | { |
| 7866 | this->error = error; |
| 7867 | if (errorHnd) { |
| 7868 | if (this->error.isNull()) { |
| 7869 | const QXmlParseException ex(QLatin1String(XMLERR_OK), columnNr+1, lineNr+1, |
| 7870 | thisPublicId, thisSystemId); |
| 7871 | errorHnd->fatalError(exception: ex); |
| 7872 | } else { |
| 7873 | const QXmlParseException ex(this->error, columnNr+1, lineNr+1, |
| 7874 | thisPublicId, thisSystemId); |
| 7875 | errorHnd->fatalError(exception: ex); |
| 7876 | } |
| 7877 | } |
| 7878 | } |
| 7879 | |
| 7880 | /* |
| 7881 | This private function is called when a parsing function encounters an |
| 7882 | unexpected EOF. It decides what to do (depending on incremental parsing or |
| 7883 | not). \a where is a pointer to the function where the error occurred and \a |
| 7884 | state is the parsing state in this function. |
| 7885 | */ |
| 7886 | void QXmlSimpleReaderPrivate::unexpectedEof(ParseFunction where, int state) |
| 7887 | { |
| 7888 | if (parseStack == nullptr) { |
| 7889 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDEOF)); |
| 7890 | } else { |
| 7891 | if (c == QXmlInputSource::EndOfDocument) { |
| 7892 | reportParseError(error: QLatin1String(XMLERR_UNEXPECTEDEOF)); |
| 7893 | } else { |
| 7894 | pushParseState(function: where, state); |
| 7895 | } |
| 7896 | } |
| 7897 | } |
| 7898 | |
| 7899 | /* |
| 7900 | This private function is called when a parse...() function returned false. It |
| 7901 | determines if there was an error or if incremental parsing simply went out of |
| 7902 | data and does the right thing for the case. \a where is a pointer to the |
| 7903 | function where the error occurred and \a state is the parsing state in this |
| 7904 | function. |
| 7905 | */ |
| 7906 | void QXmlSimpleReaderPrivate::parseFailed(ParseFunction where, int state) |
| 7907 | { |
| 7908 | if (parseStack != nullptr && error.isNull()) { |
| 7909 | pushParseState(function: where, state); |
| 7910 | } |
| 7911 | } |
| 7912 | |
| 7913 | /* |
| 7914 | This private function pushes the function pointer \a function and state \a |
| 7915 | state to the parse stack. This is used when you are doing an incremental |
| 7916 | parsing and reach the end of file too early. |
| 7917 | |
| 7918 | Only call this function when d->parseStack!=0. |
| 7919 | */ |
| 7920 | void QXmlSimpleReaderPrivate::pushParseState(ParseFunction function, int state) |
| 7921 | { |
| 7922 | QXmlSimpleReaderPrivate::ParseState ps; |
| 7923 | ps.function = function; |
| 7924 | ps.state = state; |
| 7925 | parseStack->push(t: ps); |
| 7926 | } |
| 7927 | |
| 7928 | inline static void updateValue(QString &value, const QChar *array, int &arrayPos, int &valueLen) |
| 7929 | { |
| 7930 | value.resize(size: valueLen + arrayPos); |
| 7931 | memcpy(dest: value.data() + valueLen, src: array, n: arrayPos * sizeof(QChar)); |
| 7932 | valueLen += arrayPos; |
| 7933 | arrayPos = 0; |
| 7934 | } |
| 7935 | |
| 7936 | // use buffers instead of QString::operator+= when single characters are read |
| 7937 | const QString& QXmlSimpleReaderPrivate::string() |
| 7938 | { |
| 7939 | updateValue(value&: stringValue, array: stringArray, arrayPos&: stringArrayPos, valueLen&: stringValueLen); |
| 7940 | return stringValue; |
| 7941 | } |
| 7942 | const QString& QXmlSimpleReaderPrivate::name() |
| 7943 | { |
| 7944 | updateValue(value&: nameValue, array: nameArray, arrayPos&: nameArrayPos, valueLen&: nameValueLen); |
| 7945 | return nameValue; |
| 7946 | } |
| 7947 | const QString& QXmlSimpleReaderPrivate::ref() |
| 7948 | { |
| 7949 | updateValue(value&: refValue, array: refArray, arrayPos&: refArrayPos, valueLen&: refValueLen); |
| 7950 | return refValue; |
| 7951 | } |
| 7952 | |
| 7953 | void QXmlSimpleReaderPrivate::stringAddC(QChar ch) |
| 7954 | { |
| 7955 | if (stringArrayPos == 256) |
| 7956 | updateValue(value&: stringValue, array: stringArray, arrayPos&: stringArrayPos, valueLen&: stringValueLen); |
| 7957 | stringArray[stringArrayPos++] = ch; |
| 7958 | } |
| 7959 | void QXmlSimpleReaderPrivate::nameAddC(QChar ch) |
| 7960 | { |
| 7961 | if (nameArrayPos == 256) |
| 7962 | updateValue(value&: nameValue, array: nameArray, arrayPos&: nameArrayPos, valueLen&: nameValueLen); |
| 7963 | nameArray[nameArrayPos++] = ch; |
| 7964 | } |
| 7965 | void QXmlSimpleReaderPrivate::refAddC(QChar ch) |
| 7966 | { |
| 7967 | if (refArrayPos == 256) |
| 7968 | updateValue(value&: refValue, array: refArray, arrayPos&: refArrayPos, valueLen&: refValueLen); |
| 7969 | refArray[refArrayPos++] = ch; |
| 7970 | } |
| 7971 | |
| 7972 | QT_END_NAMESPACE |
| 7973 | |