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