1 | // Copyright (C) 2016 The Qt Company Ltd. |
2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR GPL-3.0-only WITH Qt-GPL-exception-1.0 |
3 | |
4 | #include "cpp.h" |
5 | |
6 | #include <translator.h> |
7 | #include <QtCore/QBitArray> |
8 | #include <QtCore/QStack> |
9 | #include <QtCore/QTextStream> |
10 | #include <QtCore/QRegularExpression> |
11 | |
12 | QT_BEGIN_NAMESPACE |
13 | |
14 | |
15 | /* qmake ignore Q_OBJECT */ |
16 | |
17 | using namespace Qt::StringLiterals; |
18 | |
19 | static const QString = u"TRANSLATOR"_s ; |
20 | |
21 | //#define DIAGNOSE_RETRANSLATABILITY // FIXME: should make a runtime option of this |
22 | |
23 | size_t qHash(const HashString &str) |
24 | { |
25 | if (str.m_hash & 0x80000000) |
26 | str.m_hash = qHash(key: str.m_str) & 0x7fffffff; |
27 | return str.m_hash; |
28 | } |
29 | |
30 | size_t qHash(const HashStringList &list) |
31 | { |
32 | if (list.m_hash & 0x80000000) { |
33 | uint hash = 0; |
34 | for (const HashString &qs : list.m_list) { |
35 | hash ^= qHash(str: qs) ^ 0x6ad9f526; |
36 | hash = ((hash << 13) & 0x7fffffff) | (hash >> 18); |
37 | } |
38 | list.m_hash = hash; |
39 | } |
40 | return list.m_hash; |
41 | } |
42 | |
43 | static int nextFileId; |
44 | |
45 | class VisitRecorder { |
46 | public: |
47 | VisitRecorder() |
48 | { |
49 | m_ba.resize(size: nextFileId); |
50 | } |
51 | bool tryVisit(int fileId) |
52 | { |
53 | if (m_ba.at(i: fileId)) |
54 | return false; |
55 | m_ba[fileId] = true; |
56 | return true; |
57 | } |
58 | private: |
59 | QBitArray m_ba; |
60 | }; |
61 | |
62 | struct CppParserState |
63 | { |
64 | NamespaceList namespaces; |
65 | QStack<qsizetype> namespaceDepths; |
66 | NamespaceList functionContext; |
67 | QString functionContextUnresolved; |
68 | QString pendingContext; |
69 | }; |
70 | |
71 | class CppParser : private CppParserState { |
72 | |
73 | public: |
74 | CppParser(ParseResults *results = 0); |
75 | void setInput(const QString &in); |
76 | void setInput(QTextStream &ts, const QString &fileName); |
77 | void setTranslator(Translator *_tor) { tor = _tor; } |
78 | void parse(ConversionData &cd, const QStringList &includeStack, QSet<QString> &inclusions); |
79 | void parseInternal(ConversionData &cd, const QStringList &includeStack, QSet<QString> &inclusions); |
80 | const ParseResults *recordResults(bool ); |
81 | void deleteResults() { delete results; } |
82 | |
83 | private: |
84 | struct IfdefState { |
85 | IfdefState() {} |
86 | IfdefState(int _bracketDepth, int _braceDepth, int _parenDepth) : |
87 | bracketDepth(_bracketDepth), |
88 | braceDepth(_braceDepth), |
89 | parenDepth(_parenDepth), |
90 | elseLine(-1) |
91 | {} |
92 | |
93 | CppParserState state; |
94 | int bracketDepth, bracketDepth1st; |
95 | int braceDepth, braceDepth1st; |
96 | int parenDepth, parenDepth1st; |
97 | int elseLine; |
98 | }; |
99 | |
100 | enum TokenType { |
101 | Tok_Eof, Tok_class, Tok_enum, Tok_friend, Tok_namespace, Tok_using, Tok_return, |
102 | Tok_decltype, Tok_Q_OBJECT, Tok_Access, Tok_Cancel, |
103 | Tok_Ident, Tok_String, Tok_RawString, Tok_Arrow, Tok_Colon, Tok_ColonColon, |
104 | Tok_Equals, Tok_LeftBracket, Tok_RightBracket, Tok_AngleBracket, Tok_QuestionMark, |
105 | Tok_LeftBrace, Tok_RightBrace, Tok_LeftParen, Tok_RightParen, Tok_Comma, Tok_Semicolon, |
106 | Tok_Null, Tok_Integer, |
107 | Tok_QuotedInclude, Tok_AngledInclude |
108 | }; |
109 | |
110 | std::ostream &yyMsg(int line = 0); |
111 | |
112 | int getChar(); |
113 | TokenType lookAheadToSemicolonOrLeftBrace(); |
114 | TokenType getToken(); |
115 | |
116 | void processComment(); |
117 | |
118 | bool match(TokenType t); |
119 | bool matchString(QString *s); |
120 | bool matchEncoding(); |
121 | bool matchStringOrNull(QString *s); |
122 | bool matchExpression(); |
123 | |
124 | void recordMessage( |
125 | int line, const QString &context, const QString &text, const QString &, |
126 | const QString &, const QString &msgid, const TranslatorMessage::ExtraData &, |
127 | bool plural); |
128 | |
129 | void handleTr(QString &prefix, bool plural); |
130 | void handleTranslate(bool plural); |
131 | void handleTrId(bool plural); |
132 | void handleDeclareTrFunctions(); |
133 | |
134 | void processInclude(const QString &file, ConversionData &cd, |
135 | const QStringList &includeStack, QSet<QString> &inclusions); |
136 | |
137 | void saveState(CppParserState *state); |
138 | void loadState(const CppParserState &state); |
139 | |
140 | static QString stringifyNamespace(int start, const NamespaceList &namespaces); |
141 | static QString stringifyNamespace(const NamespaceList &namespaces) |
142 | { return stringifyNamespace(start: 1, namespaces); } |
143 | static QString joinNamespaces(const QString &one, const QString &two); |
144 | typedef bool (CppParser::*VisitNamespaceCallback)(const Namespace *ns, void *context) const; |
145 | bool visitNamespace(const NamespaceList &namespaces, int nsCount, |
146 | VisitNamespaceCallback callback, void *context, |
147 | VisitRecorder &vr, const ParseResults *rslt) const; |
148 | bool visitNamespace(const NamespaceList &namespaces, int nsCount, |
149 | VisitNamespaceCallback callback, void *context) const; |
150 | bool qualifyOneCallbackOwn(const Namespace *ns, void *context) const; |
151 | bool qualifyOneCallbackUsing(const Namespace *ns, void *context) const; |
152 | bool qualifyOne(const NamespaceList &namespaces, int nsCnt, const HashString &segment, |
153 | NamespaceList *resolved, QSet<HashStringList> *visitedUsings) const; |
154 | bool qualifyOne(const NamespaceList &namespaces, int nsCnt, const HashString &segment, |
155 | NamespaceList *resolved) const; |
156 | bool fullyQualify(const NamespaceList &namespaces, int nsCnt, |
157 | const NamespaceList &segments, bool isDeclaration, |
158 | NamespaceList *resolved, NamespaceList *unresolved) const; |
159 | bool fullyQualify(const NamespaceList &namespaces, |
160 | const NamespaceList &segments, bool isDeclaration, |
161 | NamespaceList *resolved, NamespaceList *unresolved) const; |
162 | bool fullyQualify(const NamespaceList &namespaces, |
163 | const QString &segments, bool isDeclaration, |
164 | NamespaceList *resolved, NamespaceList *unresolved) const; |
165 | bool findNamespaceCallback(const Namespace *ns, void *context) const; |
166 | const Namespace *findNamespace(const NamespaceList &namespaces, int nsCount = -1) const; |
167 | void enterNamespace(NamespaceList *namespaces, const HashString &name); |
168 | void truncateNamespaces(NamespaceList *namespaces, int lenght); |
169 | Namespace *modifyNamespace(NamespaceList *namespaces, bool haveLast = true); |
170 | |
171 | // Tokenizer state |
172 | QString yyFileName; |
173 | int yyCh; |
174 | bool yyAtNewline; |
175 | QString yyWord; |
176 | qsizetype yyWordInitialCapacity = 0; |
177 | QStack<IfdefState> yyIfdefStack; |
178 | int yyBracketDepth; |
179 | int yyBraceDepth; |
180 | int yyParenDepth; |
181 | int yyLineNo; |
182 | int yyCurLineNo; |
183 | int yyBracketLineNo; |
184 | int yyBraceLineNo; |
185 | int yyParenLineNo; |
186 | |
187 | // the string to read from and current position in the string |
188 | QStringConverter::Encoding yySourceEncoding = QStringConverter::Utf8; |
189 | QString yyInStr; |
190 | const ushort *yyInPtr; |
191 | |
192 | // Parser state |
193 | TokenType yyTok; |
194 | |
195 | bool metaExpected; |
196 | QString context; |
197 | QString text; |
198 | QString ; |
199 | QString ; |
200 | QString msgid; |
201 | QString sourcetext; |
202 | TranslatorMessage::ExtraData ; |
203 | |
204 | QString prospectiveContext; |
205 | ParseResults *results; |
206 | Translator *tor; |
207 | bool directInclude; |
208 | |
209 | CppParserState savedState; |
210 | int yyMinBraceDepth; |
211 | bool inDefine; |
212 | }; |
213 | |
214 | CppParser::CppParser(ParseResults *_results) |
215 | { |
216 | tor = 0; |
217 | if (_results) { |
218 | results = _results; |
219 | directInclude = true; |
220 | } else { |
221 | results = new ParseResults; |
222 | directInclude = false; |
223 | } |
224 | yyBracketDepth = 0; |
225 | yyBraceDepth = 0; |
226 | yyParenDepth = 0; |
227 | yyCurLineNo = 1; |
228 | yyBracketLineNo = 1; |
229 | yyBraceLineNo = 1; |
230 | yyParenLineNo = 1; |
231 | yyAtNewline = true; |
232 | yyMinBraceDepth = 0; |
233 | inDefine = false; |
234 | } |
235 | |
236 | |
237 | std::ostream &CppParser::yyMsg(int line) |
238 | { |
239 | return std::cerr << qPrintable(yyFileName) << ':' << (line ? line : yyLineNo) << ": " ; |
240 | } |
241 | |
242 | void CppParser::setInput(const QString &in) |
243 | { |
244 | yyInStr = in; |
245 | yyFileName = QString(); |
246 | yySourceEncoding = QStringConverter::Utf8; |
247 | } |
248 | |
249 | void CppParser::setInput(QTextStream &ts, const QString &fileName) |
250 | { |
251 | yyInStr = ts.readAll(); |
252 | yyFileName = fileName; |
253 | yySourceEncoding = ts.encoding(); |
254 | } |
255 | |
256 | /* |
257 | The first part of this source file is the C++ tokenizer. We skip |
258 | most of C++; the only tokens that interest us are defined here. |
259 | Thus, the code fragment |
260 | |
261 | int main() |
262 | { |
263 | printf("Hello, world!\n"); |
264 | return 0; |
265 | } |
266 | |
267 | is broken down into the following tokens (Tok_ omitted): |
268 | |
269 | Ident Ident LeftParen RightParen |
270 | LeftBrace |
271 | Ident LeftParen String RightParen Semicolon |
272 | return Semicolon |
273 | RightBrace. |
274 | |
275 | The 0 doesn't produce any token. |
276 | */ |
277 | |
278 | int CppParser::getChar() |
279 | { |
280 | const ushort *uc = yyInPtr; |
281 | forever { |
282 | ushort c = *uc; |
283 | if (!c) { |
284 | yyInPtr = uc; |
285 | return EOF; |
286 | } |
287 | ++uc; |
288 | if (c == '\\') { |
289 | ushort cc = *uc; |
290 | if (cc == '\n') { |
291 | ++yyCurLineNo; |
292 | ++uc; |
293 | continue; |
294 | } |
295 | if (cc == '\r') { |
296 | ++yyCurLineNo; |
297 | ++uc; |
298 | if (*uc == '\n') |
299 | ++uc; |
300 | continue; |
301 | } |
302 | } |
303 | if (c == '\r') { |
304 | if (*uc == '\n') |
305 | ++uc; |
306 | c = '\n'; |
307 | ++yyCurLineNo; |
308 | yyAtNewline = true; |
309 | } else if (c == '\n') { |
310 | ++yyCurLineNo; |
311 | yyAtNewline = true; |
312 | } else if (c != ' ' && c != '\t' && c != '#') { |
313 | yyAtNewline = false; |
314 | } |
315 | yyInPtr = uc; |
316 | return int(c); |
317 | } |
318 | } |
319 | |
320 | CppParser::TokenType CppParser::lookAheadToSemicolonOrLeftBrace() |
321 | { |
322 | if (*yyInPtr == 0) |
323 | return Tok_Eof; |
324 | const ushort *uc = yyInPtr + 1; |
325 | forever { |
326 | ushort c = *uc; |
327 | if (!c) |
328 | return Tok_Eof; |
329 | if (c == ';') |
330 | return Tok_Semicolon; |
331 | if (c == '{') |
332 | return Tok_LeftBrace; |
333 | ++uc; |
334 | } |
335 | } |
336 | |
337 | static bool isStringLiteralPrefix(const QStringView s) |
338 | { |
339 | return s == u"L"_s |
340 | || s == u"U"_s |
341 | || s == u"u"_s |
342 | || s == u"u8"_s ; |
343 | } |
344 | |
345 | static bool isRawStringLiteralPrefix(QStringView s) |
346 | { |
347 | if (s.endsWith(c: u'R')) { |
348 | s.chop(n: 1); |
349 | return s.isEmpty() || isStringLiteralPrefix(s); |
350 | } |
351 | return false; |
352 | } |
353 | |
354 | static const QString strQ_OBJECT = u"Q_OBJECT"_s ; |
355 | static const QString strclass = u"class"_s ; |
356 | static const QString strdecltype = u"decltype"_s ; |
357 | static const QString strenum = u"enum"_s ; |
358 | static const QString strfinal = u"final"_s ; |
359 | static const QString strfriend = u"friend"_s ; |
360 | static const QString strnamespace = u"namespace"_s ; |
361 | static const QString strnullptr = u"nullptr"_s ; |
362 | static const QString strQ_NULLPTR = u"Q_NULLPTR"_s ; |
363 | static const QString strNULL = u"NULL"_s ; |
364 | static const QString stroperator = u"operator"_s ; |
365 | static const QString strreturn = u"return"_s ; |
366 | static const QString strstruct = u"struct"_s ; |
367 | static const QString strusing = u"using"_s ; |
368 | static const QString strprivate = u"private"_s ; |
369 | static const QString strprotected = u"protected"_s ; |
370 | static const QString strpublic = u"public"_s ; |
371 | static const QString strslots = u"slots"_s ; |
372 | static const QString strsignals = u"signals"_s ; |
373 | static const QString strQ_SLOTS = u"Q_SLOTS"_s ; |
374 | static const QString strQ_SIGNALS = u"Q_SIGNALS"_s ; |
375 | |
376 | CppParser::TokenType CppParser::getToken() |
377 | { |
378 | restart: |
379 | // Failing this assertion would mean losing the preallocated buffer. |
380 | Q_ASSERT(yyWord.capacity() == yyWordInitialCapacity); |
381 | |
382 | while (yyCh != EOF) { |
383 | yyLineNo = yyCurLineNo; |
384 | |
385 | if (yyCh == '#' && yyAtNewline) { |
386 | /* |
387 | Early versions of lupdate complained about |
388 | unbalanced braces in the following code: |
389 | |
390 | #ifdef ALPHA |
391 | while (beta) { |
392 | #else |
393 | while (gamma) { |
394 | #endif |
395 | delta; |
396 | } |
397 | |
398 | The code contains, indeed, two opening braces for |
399 | one closing brace; yet there's no reason to panic. |
400 | |
401 | The solution is to remember yyBraceDepth as it was |
402 | when #if, #ifdef or #ifndef was met, and to set |
403 | yyBraceDepth to that value when meeting #elif or |
404 | #else. |
405 | */ |
406 | do { |
407 | yyCh = getChar(); |
408 | } while (isspace(yyCh) && yyCh != '\n'); |
409 | |
410 | switch (yyCh) { |
411 | case 'd': // define |
412 | // Skip over the name of the define to avoid it being interpreted as c++ code |
413 | do { // Rest of "define" |
414 | yyCh = getChar(); |
415 | if (yyCh == EOF) |
416 | return Tok_Eof; |
417 | if (yyCh == '\n') |
418 | goto restart; |
419 | } while (!isspace(yyCh)); |
420 | do { // Space beween "define" and macro name |
421 | yyCh = getChar(); |
422 | if (yyCh == EOF) |
423 | return Tok_Eof; |
424 | if (yyCh == '\n') |
425 | goto restart; |
426 | } while (isspace(yyCh)); |
427 | do { // Macro name |
428 | if (yyCh == '(') { |
429 | // Argument list. Follows the name without a space, and no |
430 | // paren nesting is possible. |
431 | do { |
432 | yyCh = getChar(); |
433 | if (yyCh == EOF) |
434 | return Tok_Eof; |
435 | if (yyCh == '\n') |
436 | goto restart; |
437 | } while (yyCh != ')'); |
438 | break; |
439 | } |
440 | yyCh = getChar(); |
441 | if (yyCh == EOF) |
442 | return Tok_Eof; |
443 | if (yyCh == '\n') |
444 | goto restart; |
445 | } while (!isspace(yyCh)); |
446 | do { // Shortcut the immediate newline case if no comments follow. |
447 | yyCh = getChar(); |
448 | if (yyCh == EOF) |
449 | return Tok_Eof; |
450 | if (yyCh == '\n') |
451 | goto restart; |
452 | } while (isspace(yyCh)); |
453 | |
454 | saveState(state: &savedState); |
455 | yyMinBraceDepth = yyBraceDepth; |
456 | inDefine = true; |
457 | goto restart; |
458 | case 'i': |
459 | yyCh = getChar(); |
460 | if (yyCh == 'f') { |
461 | // if, ifdef, ifndef |
462 | yyIfdefStack.push(t: IfdefState(yyBracketDepth, yyBraceDepth, yyParenDepth)); |
463 | yyCh = getChar(); |
464 | } else if (yyCh == 'n') { |
465 | // include |
466 | do { |
467 | yyCh = getChar(); |
468 | } while (yyCh != EOF && !isspace(yyCh) && yyCh != '"' && yyCh != '<' ); |
469 | while (isspace(yyCh)) |
470 | yyCh = getChar(); |
471 | int tChar; |
472 | if (yyCh == '"') |
473 | tChar = '"'; |
474 | else if (yyCh == '<') |
475 | tChar = '>'; |
476 | else |
477 | break; |
478 | ushort *ptr = (ushort *)yyWord.unicode(); |
479 | forever { |
480 | yyCh = getChar(); |
481 | if (yyCh == EOF || yyCh == '\n') |
482 | break; |
483 | if (yyCh == tChar) { |
484 | yyCh = getChar(); |
485 | break; |
486 | } |
487 | *ptr++ = yyCh; |
488 | } |
489 | yyWord.resize(size: ptr - (ushort *)yyWord.unicode()); |
490 | return (tChar == '"') ? Tok_QuotedInclude : Tok_AngledInclude; |
491 | } |
492 | break; |
493 | case 'e': |
494 | yyCh = getChar(); |
495 | if (yyCh == 'l') { |
496 | // elif, else |
497 | if (!yyIfdefStack.isEmpty()) { |
498 | IfdefState &is = yyIfdefStack.top(); |
499 | if (is.elseLine != -1) { |
500 | if (yyBracketDepth != is.bracketDepth1st |
501 | || yyBraceDepth != is.braceDepth1st |
502 | || yyParenDepth != is.parenDepth1st) |
503 | yyMsg(line: is.elseLine) |
504 | << "Parenthesis/bracket/brace mismatch between " |
505 | "#if and #else branches; using #if branch\n" ; |
506 | } else { |
507 | is.bracketDepth1st = yyBracketDepth; |
508 | is.braceDepth1st = yyBraceDepth; |
509 | is.parenDepth1st = yyParenDepth; |
510 | saveState(state: &is.state); |
511 | } |
512 | is.elseLine = yyLineNo; |
513 | yyBracketDepth = is.bracketDepth; |
514 | yyBraceDepth = is.braceDepth; |
515 | yyParenDepth = is.parenDepth; |
516 | } |
517 | yyCh = getChar(); |
518 | } else if (yyCh == 'n') { |
519 | // endif |
520 | if (!yyIfdefStack.isEmpty()) { |
521 | IfdefState is = yyIfdefStack.pop(); |
522 | if (is.elseLine != -1) { |
523 | if (yyBracketDepth != is.bracketDepth1st |
524 | || yyBraceDepth != is.braceDepth1st |
525 | || yyParenDepth != is.parenDepth1st) |
526 | yyMsg(line: is.elseLine) |
527 | << "Parenthesis/brace mismatch between " |
528 | "#if and #else branches; using #if branch\n" ; |
529 | yyBracketDepth = is.bracketDepth1st; |
530 | yyBraceDepth = is.braceDepth1st; |
531 | yyParenDepth = is.parenDepth1st; |
532 | loadState(state: is.state); |
533 | } |
534 | } |
535 | yyCh = getChar(); |
536 | } |
537 | break; |
538 | } |
539 | // Optimization: skip over rest of preprocessor directive |
540 | do { |
541 | if (yyCh == '/') { |
542 | yyCh = getChar(); |
543 | if (yyCh == '/') { |
544 | do { |
545 | yyCh = getChar(); |
546 | } while (yyCh != EOF && yyCh != '\n'); |
547 | break; |
548 | } else if (yyCh == '*') { |
549 | bool metAster = false; |
550 | |
551 | forever { |
552 | yyCh = getChar(); |
553 | if (yyCh == EOF) { |
554 | yyMsg() << "Unterminated C++ comment\n" ; |
555 | break; |
556 | } |
557 | |
558 | if (yyCh == '*') { |
559 | metAster = true; |
560 | } else if (metAster && yyCh == '/') { |
561 | yyCh = getChar(); |
562 | break; |
563 | } else { |
564 | metAster = false; |
565 | } |
566 | } |
567 | } |
568 | } else { |
569 | yyCh = getChar(); |
570 | } |
571 | } while (yyCh != '\n' && yyCh != EOF); |
572 | yyCh = getChar(); |
573 | } else if ((yyCh >= 'A' && yyCh <= 'Z') || (yyCh >= 'a' && yyCh <= 'z') || yyCh == '_') { |
574 | ushort *ptr = (ushort *)yyWord.unicode(); |
575 | do { |
576 | *ptr++ = yyCh; |
577 | yyCh = getChar(); |
578 | } while ((yyCh >= 'A' && yyCh <= 'Z') || (yyCh >= 'a' && yyCh <= 'z') |
579 | || (yyCh >= '0' && yyCh <= '9') || yyCh == '_'); |
580 | yyWord.resize(size: ptr - (ushort *)yyWord.unicode()); |
581 | |
582 | //qDebug() << "IDENT: " << yyWord; |
583 | |
584 | if (yyCh == '"' && isStringLiteralPrefix(s: yyWord)) { |
585 | // Handle prefixed string literals as ordinary string literals. |
586 | continue; |
587 | } |
588 | |
589 | switch (yyWord.unicode()[0].unicode()) { |
590 | case 'N': |
591 | if (yyWord == strNULL) |
592 | return Tok_Null; |
593 | break; |
594 | case 'Q': |
595 | if (yyWord == strQ_NULLPTR) |
596 | return Tok_Null; |
597 | if (yyWord == strQ_OBJECT) |
598 | return Tok_Q_OBJECT; |
599 | if (yyWord == strQ_SLOTS || yyWord == strQ_SIGNALS) |
600 | return Tok_Access; |
601 | break; |
602 | case 'c': |
603 | if (yyWord == strclass) |
604 | return Tok_class; |
605 | break; |
606 | case 'd': |
607 | if (yyWord == strdecltype) |
608 | return Tok_decltype; |
609 | break; |
610 | case 'e': |
611 | if (yyWord == strenum) |
612 | return Tok_enum; |
613 | break; |
614 | case 'f': |
615 | if (yyWord == strfriend) |
616 | return Tok_friend; |
617 | break; |
618 | case 'n': |
619 | if (yyWord == strnamespace) |
620 | return Tok_namespace; |
621 | if (yyWord == strnullptr) |
622 | return Tok_Null; |
623 | break; |
624 | case 'o': |
625 | if (yyWord == stroperator) { |
626 | // Operator overload declaration/definition. |
627 | // We need to prevent those characters from confusing the followup |
628 | // parsing. Actually using them does not add value, so just eat them. |
629 | while (isspace(yyCh)) |
630 | yyCh = getChar(); |
631 | while (yyCh == '+' || yyCh == '-' || yyCh == '*' || yyCh == '/' || yyCh == '%' |
632 | || yyCh == '=' || yyCh == '<' || yyCh == '>' || yyCh == '!' |
633 | || yyCh == '&' || yyCh == '|' || yyCh == '~' || yyCh == '^' |
634 | || yyCh == '[' || yyCh == ']') |
635 | yyCh = getChar(); |
636 | } |
637 | break; |
638 | case 'p': |
639 | if (yyWord == strpublic || yyWord == strprotected || yyWord == strprivate) |
640 | return Tok_Access; |
641 | break; |
642 | case 'r': |
643 | if (yyWord == strreturn) |
644 | return Tok_return; |
645 | break; |
646 | case 's': |
647 | if (yyWord == strstruct) |
648 | return Tok_class; |
649 | if (yyWord == strslots || yyWord == strsignals) |
650 | return Tok_Access; |
651 | break; |
652 | case 'u': |
653 | if (yyWord == strusing) |
654 | return Tok_using; |
655 | break; |
656 | } |
657 | |
658 | // a C++11 raw string literal? |
659 | if (yyCh == '"' && isRawStringLiteralPrefix(s: yyWord)) { |
660 | ptr = reinterpret_cast<ushort *>(const_cast<QChar *>(yyWord.unicode())); |
661 | //get delimiter |
662 | QString delimiter; |
663 | for (yyCh = getChar(); yyCh != EOF && yyCh != '('; yyCh = getChar()) |
664 | delimiter += QLatin1Char(yyCh); |
665 | if (yyCh != EOF) |
666 | yyCh = getChar(); // throw away the opening parentheses |
667 | bool is_end = false; |
668 | ushort *ptr_past_end = nullptr; |
669 | while (yyCh != EOF && !is_end) { |
670 | *ptr++ = yyCh; |
671 | if (ptr_past_end != nullptr) { |
672 | if (delimiter.size() == ptr - ptr_past_end |
673 | && memcmp(s1: delimiter.unicode(), s2: ptr_past_end, n: (ptr - ptr_past_end) * sizeof (ushort)) == 0 |
674 | ) { |
675 | // we've got the delimiter, check if " follows |
676 | yyCh = getChar(); |
677 | if (yyCh == '"') |
678 | is_end = true; |
679 | else |
680 | ptr_past_end = nullptr; |
681 | continue; |
682 | } |
683 | } |
684 | if (yyCh == ')') { |
685 | ptr_past_end = ptr; |
686 | if (delimiter.isEmpty()) { |
687 | // no delimiter, check if " follows |
688 | yyCh = getChar(); |
689 | if (yyCh == '"') |
690 | is_end = true; |
691 | else |
692 | ptr_past_end = nullptr; |
693 | continue; |
694 | } |
695 | } |
696 | yyCh = getChar(); |
697 | } |
698 | if (is_end) |
699 | yyWord.resize(size: ptr_past_end - 1 - reinterpret_cast<const ushort *>(yyWord.unicode())); |
700 | else |
701 | yyWord.resize(size: ptr - reinterpret_cast<const ushort *>(yyWord.unicode())); |
702 | if (yyCh != '"') |
703 | yyMsg() << "Unterminated/mismatched C++ Raw string\n" ; |
704 | else |
705 | yyCh = getChar(); |
706 | return Tok_RawString; |
707 | } |
708 | |
709 | return Tok_Ident; |
710 | } else { |
711 | switch (yyCh) { |
712 | case '\n': |
713 | if (inDefine) { |
714 | loadState(state: savedState); |
715 | prospectiveContext.clear(); |
716 | yyBraceDepth = yyMinBraceDepth; |
717 | yyMinBraceDepth = 0; |
718 | inDefine = false; |
719 | metaExpected = true; |
720 | yyCh = getChar(); |
721 | return Tok_Cancel; // Break out of any multi-token constructs |
722 | } |
723 | yyCh = getChar(); |
724 | break; |
725 | case '/': |
726 | yyCh = getChar(); |
727 | if (yyCh == '/') { |
728 | ushort *ptr = (ushort *)yyWord.unicode(); |
729 | do { |
730 | yyCh = getChar(); |
731 | if (yyCh == EOF) |
732 | break; |
733 | *ptr++ = yyCh; |
734 | } while (yyCh != '\n'); |
735 | yyWord.resize(size: ptr - (ushort *)yyWord.unicode()); |
736 | processComment(); |
737 | } else if (yyCh == '*') { |
738 | bool metAster = false; |
739 | ushort *ptr = (ushort *)yyWord.unicode(); |
740 | |
741 | forever { |
742 | yyCh = getChar(); |
743 | if (yyCh == EOF) { |
744 | yyMsg() << "Unterminated C++ comment\n" ; |
745 | break; |
746 | } |
747 | *ptr++ = yyCh; |
748 | |
749 | if (yyCh == '*') |
750 | metAster = true; |
751 | else if (metAster && yyCh == '/') |
752 | break; |
753 | else |
754 | metAster = false; |
755 | } |
756 | yyWord.resize(size: ptr - (ushort *)yyWord.unicode() - 2); |
757 | processComment(); |
758 | |
759 | yyCh = getChar(); |
760 | } |
761 | break; |
762 | case '"': { |
763 | ushort *ptr = (ushort *)yyWord.unicode(); |
764 | yyCh = getChar(); |
765 | while (yyCh != EOF && yyCh != '\n' && yyCh != '"') { |
766 | if (yyCh == '\\') { |
767 | yyCh = getChar(); |
768 | if (yyCh == EOF || yyCh == '\n') |
769 | break; |
770 | *ptr++ = '\\'; |
771 | } |
772 | *ptr++ = yyCh; |
773 | yyCh = getChar(); |
774 | } |
775 | yyWord.resize(size: ptr - (ushort *)yyWord.unicode()); |
776 | |
777 | if (yyCh != '"') |
778 | yyMsg() << "Unterminated C++ string\n" ; |
779 | else |
780 | yyCh = getChar(); |
781 | return Tok_String; |
782 | } |
783 | case '-': |
784 | yyCh = getChar(); |
785 | if (yyCh == '>') { |
786 | yyCh = getChar(); |
787 | return Tok_Arrow; |
788 | } |
789 | break; |
790 | case ':': |
791 | yyCh = getChar(); |
792 | if (yyCh == ':') { |
793 | yyCh = getChar(); |
794 | return Tok_ColonColon; |
795 | } |
796 | return Tok_Colon; |
797 | // Incomplete: '<' might be part of '<=' or of template syntax. |
798 | // The main intent of not completely ignoring it is to break |
799 | // parsing of things like std::cout << QObject::tr() as |
800 | // context std::cout::QObject (see Task 161106) |
801 | case '=': |
802 | yyCh = getChar(); |
803 | return Tok_Equals; |
804 | case '>': |
805 | case '<': |
806 | yyCh = getChar(); |
807 | return Tok_AngleBracket; |
808 | case '\'': |
809 | yyCh = getChar(); |
810 | if (yyCh == '\\') |
811 | yyCh = getChar(); |
812 | |
813 | forever { |
814 | if (yyCh == EOF || yyCh == '\n') { |
815 | yyMsg() << "Unterminated C++ character\n" ; |
816 | break; |
817 | } |
818 | yyCh = getChar(); |
819 | if (yyCh == '\'') { |
820 | yyCh = getChar(); |
821 | break; |
822 | } |
823 | } |
824 | break; |
825 | case '{': |
826 | if (yyBraceDepth == 0) |
827 | yyBraceLineNo = yyCurLineNo; |
828 | yyBraceDepth++; |
829 | yyCh = getChar(); |
830 | return Tok_LeftBrace; |
831 | case '}': |
832 | if (yyBraceDepth == yyMinBraceDepth) { |
833 | if (!inDefine) |
834 | yyMsg(line: yyCurLineNo) |
835 | << "Excess closing brace in C++ code" |
836 | " (or abuse of the C++ preprocessor)\n" ; |
837 | // Avoid things getting messed up even more |
838 | yyCh = getChar(); |
839 | return Tok_Semicolon; |
840 | } |
841 | yyBraceDepth--; |
842 | yyCh = getChar(); |
843 | return Tok_RightBrace; |
844 | case '(': |
845 | if (yyParenDepth == 0) |
846 | yyParenLineNo = yyCurLineNo; |
847 | yyParenDepth++; |
848 | yyCh = getChar(); |
849 | return Tok_LeftParen; |
850 | case ')': |
851 | if (yyParenDepth == 0) |
852 | yyMsg(line: yyCurLineNo) |
853 | << "Excess closing parenthesis in C++ code" |
854 | " (or abuse of the C++ preprocessor)\n" ; |
855 | else |
856 | yyParenDepth--; |
857 | yyCh = getChar(); |
858 | return Tok_RightParen; |
859 | case '[': |
860 | if (yyBracketDepth == 0) |
861 | yyBracketLineNo = yyCurLineNo; |
862 | yyBracketDepth++; |
863 | yyCh = getChar(); |
864 | return Tok_LeftBracket; |
865 | case ']': |
866 | if (yyBracketDepth == 0) |
867 | yyMsg(line: yyCurLineNo) |
868 | << "Excess closing bracket in C++ code" |
869 | " (or abuse of the C++ preprocessor)\n" ; |
870 | else |
871 | yyBracketDepth--; |
872 | yyCh = getChar(); |
873 | return Tok_RightBracket; |
874 | case ',': |
875 | yyCh = getChar(); |
876 | return Tok_Comma; |
877 | case ';': |
878 | yyCh = getChar(); |
879 | return Tok_Semicolon; |
880 | case '?': |
881 | yyCh = getChar(); |
882 | return Tok_QuestionMark; |
883 | case '0': |
884 | yyCh = getChar(); |
885 | if (yyCh == 'x' || yyCh == 'X') { |
886 | do { |
887 | yyCh = getChar(); |
888 | } while ((yyCh >= '0' && yyCh <= '9') || yyCh == '\'' |
889 | || (yyCh >= 'a' && yyCh <= 'f') || (yyCh >= 'A' && yyCh <= 'F')); |
890 | return Tok_Integer; |
891 | } |
892 | if (yyCh < '0' || yyCh > '9') |
893 | return Tok_Null; |
894 | Q_FALLTHROUGH(); |
895 | case '1': |
896 | case '2': |
897 | case '3': |
898 | case '4': |
899 | case '5': |
900 | case '6': |
901 | case '7': |
902 | case '8': |
903 | case '9': |
904 | do { |
905 | yyCh = getChar(); |
906 | } while ((yyCh >= '0' && yyCh <= '9') || yyCh == '\''); |
907 | return Tok_Integer; |
908 | default: |
909 | yyCh = getChar(); |
910 | break; |
911 | } |
912 | } |
913 | } |
914 | return Tok_Eof; |
915 | } |
916 | |
917 | /* |
918 | The second part of this source file are namespace/class related |
919 | utilities for the third part. |
920 | */ |
921 | |
922 | void CppParser::saveState(CppParserState *state) |
923 | { |
924 | *state = *this; |
925 | } |
926 | |
927 | void CppParser::loadState(const CppParserState &state) |
928 | { |
929 | *static_cast<CppParserState *>(this) = state; |
930 | } |
931 | |
932 | Namespace *CppParser::modifyNamespace(NamespaceList *namespaces, bool haveLast) |
933 | { |
934 | Namespace *pns, *ns = &results->rootNamespace; |
935 | for (int i = 1; i < namespaces->size(); ++i) { |
936 | pns = ns; |
937 | if (!(ns = pns->children.value(key: namespaces->at(i)))) { |
938 | do { |
939 | ns = new Namespace; |
940 | if (haveLast || i < namespaces->size() - 1) |
941 | if (const Namespace *ons = findNamespace(namespaces: *namespaces, nsCount: i + 1)) |
942 | ns->classDef = ons->classDef; |
943 | pns->children.insert(key: namespaces->at(i), value: ns); |
944 | pns = ns; |
945 | } while (++i < namespaces->size()); |
946 | break; |
947 | } |
948 | } |
949 | return ns; |
950 | } |
951 | |
952 | QString CppParser::stringifyNamespace(int start, const NamespaceList &namespaces) |
953 | { |
954 | QString ret; |
955 | int l = 0; |
956 | for (int j = start; j < namespaces.size(); ++j) |
957 | l += namespaces.at(i: j).value().size(); |
958 | ret.reserve(asize: l + qMax(a: 0, b: (namespaces.size() - start - 1)) * 2); |
959 | for (int i = start; i < namespaces.size(); ++i) { |
960 | if (i > start) |
961 | ret += QLatin1String("::" ); |
962 | ret += namespaces.at(i).value(); |
963 | } |
964 | return ret; |
965 | } |
966 | |
967 | QString CppParser::joinNamespaces(const QString &one, const QString &two) |
968 | { |
969 | return two.isEmpty() ? one : one.isEmpty() ? two : one + QStringLiteral("::" ) + two; |
970 | } |
971 | |
972 | bool CppParser::visitNamespace(const NamespaceList &namespaces, int nsCount, |
973 | VisitNamespaceCallback callback, void *context, |
974 | VisitRecorder &vr, const ParseResults *rslt) const |
975 | { |
976 | const Namespace *ns = &rslt->rootNamespace; |
977 | for (int i = 1; i < nsCount; ++i) |
978 | if (!(ns = ns->children.value(key: namespaces.at(i)))) |
979 | goto supers; |
980 | if ((this->*callback)(ns, context)) |
981 | return true; |
982 | supers: |
983 | for (const ParseResults *sup : rslt->includes) |
984 | if (vr.tryVisit(fileId: sup->fileId) |
985 | && visitNamespace(namespaces, nsCount, callback, context, vr, rslt: sup)) |
986 | return true; |
987 | return false; |
988 | } |
989 | |
990 | bool CppParser::visitNamespace(const NamespaceList &namespaces, int nsCount, |
991 | VisitNamespaceCallback callback, void *context) const |
992 | { |
993 | VisitRecorder vr; |
994 | return visitNamespace(namespaces, nsCount, callback, context, vr, rslt: results); |
995 | } |
996 | |
997 | struct QualifyOneData { |
998 | QualifyOneData(const NamespaceList &ns, int nsc, const HashString &seg, NamespaceList *rslvd, |
999 | QSet<HashStringList> *visited) |
1000 | : namespaces(ns), nsCount(nsc), segment(seg), resolved(rslvd), visitedUsings(visited) |
1001 | {} |
1002 | |
1003 | const NamespaceList &namespaces; |
1004 | int nsCount; |
1005 | const HashString &segment; |
1006 | NamespaceList *resolved; |
1007 | QSet<HashStringList> *visitedUsings; |
1008 | }; |
1009 | |
1010 | bool CppParser::qualifyOneCallbackOwn(const Namespace *ns, void *context) const |
1011 | { |
1012 | QualifyOneData *data = (QualifyOneData *)context; |
1013 | if (ns->children.contains(key: data->segment)) { |
1014 | *data->resolved = data->namespaces.mid(pos: 0, len: data->nsCount); |
1015 | *data->resolved << data->segment; |
1016 | return true; |
1017 | } |
1018 | auto nsai = ns->aliases.constFind(key: data->segment); |
1019 | if (nsai != ns->aliases.constEnd()) { |
1020 | const NamespaceList &nsl = *nsai; |
1021 | if (nsl.last().value().isEmpty()) { // Delayed alias resolution |
1022 | NamespaceList &nslIn = *const_cast<NamespaceList *>(&nsl); |
1023 | nslIn.removeLast(); |
1024 | NamespaceList nslOut; |
1025 | if (!fullyQualify(namespaces: data->namespaces, nsCnt: data->nsCount, segments: nslIn, isDeclaration: false, resolved: &nslOut, unresolved: 0)) { |
1026 | const_cast<Namespace *>(ns)->aliases.remove(key: data->segment); |
1027 | return false; |
1028 | } |
1029 | nslIn = nslOut; |
1030 | } |
1031 | *data->resolved = nsl; |
1032 | return true; |
1033 | } |
1034 | return false; |
1035 | } |
1036 | |
1037 | bool CppParser::qualifyOneCallbackUsing(const Namespace *ns, void *context) const |
1038 | { |
1039 | QualifyOneData *data = (QualifyOneData *)context; |
1040 | for (const HashStringList &use : ns->usings) |
1041 | if (!data->visitedUsings->contains(value: use)) { |
1042 | data->visitedUsings->insert(value: use); |
1043 | if (qualifyOne(namespaces: use.value(), nsCnt: use.value().size(), segment: data->segment, resolved: data->resolved, |
1044 | visitedUsings: data->visitedUsings)) |
1045 | return true; |
1046 | } |
1047 | return false; |
1048 | } |
1049 | |
1050 | bool CppParser::qualifyOne(const NamespaceList &namespaces, int nsCnt, const HashString &segment, |
1051 | NamespaceList *resolved, QSet<HashStringList> *visitedUsings) const |
1052 | { |
1053 | QualifyOneData data(namespaces, nsCnt, segment, resolved, visitedUsings); |
1054 | |
1055 | if (visitNamespace(namespaces, nsCount: nsCnt, callback: &CppParser::qualifyOneCallbackOwn, context: &data)) |
1056 | return true; |
1057 | |
1058 | return visitNamespace(namespaces, nsCount: nsCnt, callback: &CppParser::qualifyOneCallbackUsing, context: &data); |
1059 | } |
1060 | |
1061 | bool CppParser::qualifyOne(const NamespaceList &namespaces, int nsCnt, const HashString &segment, |
1062 | NamespaceList *resolved) const |
1063 | { |
1064 | QSet<HashStringList> visitedUsings; |
1065 | |
1066 | return qualifyOne(namespaces, nsCnt, segment, resolved, visitedUsings: &visitedUsings); |
1067 | } |
1068 | |
1069 | bool CppParser::fullyQualify(const NamespaceList &namespaces, int nsCnt, |
1070 | const NamespaceList &segments, bool isDeclaration, |
1071 | NamespaceList *resolved, NamespaceList *unresolved) const |
1072 | { |
1073 | int nsIdx; |
1074 | int initSegIdx; |
1075 | |
1076 | if (segments.first().value().isEmpty()) { |
1077 | // fully qualified |
1078 | if (segments.size() == 1) { |
1079 | resolved->clear(); |
1080 | *resolved << HashString(QString()); |
1081 | return true; |
1082 | } |
1083 | initSegIdx = 1; |
1084 | nsIdx = 0; |
1085 | } else { |
1086 | initSegIdx = 0; |
1087 | nsIdx = nsCnt - 1; |
1088 | } |
1089 | |
1090 | do { |
1091 | if (qualifyOne(namespaces, nsCnt: nsIdx + 1, segment: segments[initSegIdx], resolved)) { |
1092 | int segIdx = initSegIdx; |
1093 | while (++segIdx < segments.size()) { |
1094 | if (!qualifyOne(namespaces: *resolved, nsCnt: resolved->size(), segment: segments[segIdx], resolved)) { |
1095 | if (unresolved) |
1096 | *unresolved = segments.mid(pos: segIdx); |
1097 | return false; |
1098 | } |
1099 | } |
1100 | return true; |
1101 | } |
1102 | } while (!isDeclaration && --nsIdx >= 0); |
1103 | resolved->clear(); |
1104 | *resolved << HashString(QString()); |
1105 | if (unresolved) |
1106 | *unresolved = segments.mid(pos: initSegIdx); |
1107 | return false; |
1108 | } |
1109 | |
1110 | bool CppParser::fullyQualify(const NamespaceList &namespaces, |
1111 | const NamespaceList &segments, bool isDeclaration, |
1112 | NamespaceList *resolved, NamespaceList *unresolved) const |
1113 | { |
1114 | return fullyQualify(namespaces, nsCnt: namespaces.size(), |
1115 | segments, isDeclaration, resolved, unresolved); |
1116 | } |
1117 | |
1118 | bool CppParser::fullyQualify(const NamespaceList &namespaces, |
1119 | const QString &quali, bool isDeclaration, |
1120 | NamespaceList *resolved, NamespaceList *unresolved) const |
1121 | { |
1122 | static QString strColons(QLatin1String("::" )); |
1123 | |
1124 | NamespaceList segments; |
1125 | for (const QString &str : quali.split(sep: strColons)) // XXX slow, but needs to be fast(?) |
1126 | segments << HashString(str); |
1127 | return fullyQualify(namespaces, segments, isDeclaration, resolved, unresolved); |
1128 | } |
1129 | |
1130 | bool CppParser::findNamespaceCallback(const Namespace *ns, void *context) const |
1131 | { |
1132 | *((const Namespace **)context) = ns; |
1133 | return true; |
1134 | } |
1135 | |
1136 | const Namespace *CppParser::findNamespace(const NamespaceList &namespaces, int nsCount) const |
1137 | { |
1138 | const Namespace *ns = 0; |
1139 | if (nsCount == -1) |
1140 | nsCount = namespaces.size(); |
1141 | visitNamespace(namespaces, nsCount, callback: &CppParser::findNamespaceCallback, context: &ns); |
1142 | return ns; |
1143 | } |
1144 | |
1145 | void CppParser::enterNamespace(NamespaceList *namespaces, const HashString &name) |
1146 | { |
1147 | *namespaces << name; |
1148 | if (!findNamespace(namespaces: *namespaces)) |
1149 | modifyNamespace(namespaces, haveLast: false); |
1150 | } |
1151 | |
1152 | void CppParser::truncateNamespaces(NamespaceList *namespaces, int length) |
1153 | { |
1154 | if (namespaces->size() > length) |
1155 | namespaces->erase(abegin: namespaces->begin() + length, aend: namespaces->end()); |
1156 | } |
1157 | |
1158 | |
1159 | /* |
1160 | Functions for processing include files. |
1161 | */ |
1162 | |
1163 | IncludeCycleHash &CppFiles::includeCycles() |
1164 | { |
1165 | static IncludeCycleHash cycles; |
1166 | |
1167 | return cycles; |
1168 | } |
1169 | |
1170 | TranslatorHash &CppFiles::translatedFiles() |
1171 | { |
1172 | static TranslatorHash tors; |
1173 | |
1174 | return tors; |
1175 | } |
1176 | |
1177 | QSet<QString> &CppFiles::blacklistedFiles() |
1178 | { |
1179 | static QSet<QString> blacklisted; |
1180 | |
1181 | return blacklisted; |
1182 | } |
1183 | |
1184 | QSet<const ParseResults *> CppFiles::getResults(const QString &cleanFile) |
1185 | { |
1186 | IncludeCycle * const cycle = includeCycles().value(key: cleanFile); |
1187 | |
1188 | if (cycle) |
1189 | return cycle->results; |
1190 | else |
1191 | return QSet<const ParseResults *>(); |
1192 | } |
1193 | |
1194 | void CppFiles::setResults(const QString &cleanFile, const ParseResults *results) |
1195 | { |
1196 | IncludeCycle *cycle = includeCycles().value(key: cleanFile); |
1197 | |
1198 | if (!cycle) { |
1199 | cycle = new IncludeCycle; |
1200 | includeCycles().insert(key: cleanFile, value: cycle); |
1201 | } |
1202 | |
1203 | cycle->fileNames.insert(value: cleanFile); |
1204 | cycle->results.insert(value: results); |
1205 | } |
1206 | |
1207 | const Translator *CppFiles::getTranslator(const QString &cleanFile) |
1208 | { |
1209 | return translatedFiles().value(key: cleanFile); |
1210 | } |
1211 | |
1212 | void CppFiles::setTranslator(const QString &cleanFile, const Translator *tor) |
1213 | { |
1214 | translatedFiles().insert(key: cleanFile, value: tor); |
1215 | } |
1216 | |
1217 | bool CppFiles::isBlacklisted(const QString &cleanFile) |
1218 | { |
1219 | return blacklistedFiles().contains(value: cleanFile); |
1220 | } |
1221 | |
1222 | void CppFiles::setBlacklisted(const QString &cleanFile) |
1223 | { |
1224 | blacklistedFiles().insert(value: cleanFile); |
1225 | } |
1226 | |
1227 | void CppFiles::addIncludeCycle(const QSet<QString> &fileNames) |
1228 | { |
1229 | IncludeCycle * const cycle = new IncludeCycle; |
1230 | cycle->fileNames = fileNames; |
1231 | |
1232 | QSet<IncludeCycle *> intersectingCycles; |
1233 | for (const QString &fileName : fileNames) { |
1234 | IncludeCycle *intersectingCycle = includeCycles().value(key: fileName); |
1235 | |
1236 | if (intersectingCycle && !intersectingCycles.contains(value: intersectingCycle)) { |
1237 | intersectingCycles.insert(value: intersectingCycle); |
1238 | |
1239 | cycle->fileNames.unite(other: intersectingCycle->fileNames); |
1240 | cycle->results.unite(other: intersectingCycle->results); |
1241 | } |
1242 | } |
1243 | qDeleteAll(c: intersectingCycles); |
1244 | |
1245 | for (const QString &fileName : std::as_const(t&: cycle->fileNames)) |
1246 | includeCycles().insert(key: fileName, value: cycle); |
1247 | } |
1248 | |
1249 | static bool (const QString &name) |
1250 | { |
1251 | QString fileExt = QFileInfo(name).suffix(); |
1252 | return fileExt.isEmpty() || fileExt.startsWith(c: QLatin1Char('h'), cs: Qt::CaseInsensitive); |
1253 | } |
1254 | |
1255 | void CppParser::processInclude(const QString &file, ConversionData &cd, const QStringList &includeStack, |
1256 | QSet<QString> &inclusions) |
1257 | { |
1258 | QString cleanFile = QDir::cleanPath(path: file); |
1259 | |
1260 | for (const QString &ex : std::as_const(t&: cd.m_excludes)) { |
1261 | QRegularExpression rx(QRegularExpression::wildcardToRegularExpression(str: ex)); |
1262 | if (rx.match(subject: cleanFile).hasMatch()) |
1263 | return; |
1264 | } |
1265 | |
1266 | const int index = includeStack.indexOf(str: cleanFile); |
1267 | if (index != -1) { |
1268 | CppFiles::addIncludeCycle(fileNames: QSet<QString>(includeStack.cbegin() + index, includeStack.cend())); |
1269 | return; |
1270 | } |
1271 | |
1272 | // If the #include is in any kind of namespace, has been blacklisted previously, |
1273 | // or is not a header file (stdc++ extensionless or *.h*), then really include |
1274 | // it. Otherwise it is safe to process it stand-alone and re-use the parsed |
1275 | // namespace data for inclusion into other files. |
1276 | bool isIndirect = false; |
1277 | if (namespaces.size() == 1 && functionContext.size() == 1 |
1278 | && functionContextUnresolved.isEmpty() && pendingContext.isEmpty() |
1279 | && !CppFiles::isBlacklisted(cleanFile) |
1280 | && isHeader(name: cleanFile)) { |
1281 | |
1282 | QSet<const ParseResults *> res = CppFiles::getResults(cleanFile); |
1283 | if (!res.isEmpty()) { |
1284 | results->includes.unite(other: res); |
1285 | return; |
1286 | } |
1287 | |
1288 | isIndirect = true; |
1289 | } |
1290 | |
1291 | QFile f(cleanFile); |
1292 | if (!f.open(flags: QIODevice::ReadOnly)) { |
1293 | yyMsg() << qPrintable( |
1294 | QStringLiteral("Cannot open %1: %2\n" ).arg(cleanFile, f.errorString())); |
1295 | return; |
1296 | } |
1297 | |
1298 | QTextStream ts(&f); |
1299 | ts.setEncoding(yySourceEncoding); |
1300 | ts.setAutoDetectUnicode(true); |
1301 | |
1302 | inclusions.insert(value: cleanFile); |
1303 | if (isIndirect) { |
1304 | CppParser parser; |
1305 | for (const QString &projectRoot : std::as_const(t&: cd.m_projectRoots)) |
1306 | if (cleanFile.startsWith(s: projectRoot)) { |
1307 | parser.setTranslator(new Translator); |
1308 | break; |
1309 | } |
1310 | parser.setInput(ts, fileName: cleanFile); |
1311 | QStringList stack = includeStack; |
1312 | stack << cleanFile; |
1313 | parser.parse(cd, includeStack: stack, inclusions); |
1314 | results->includes.insert(value: parser.recordResults(isHeader: true)); |
1315 | } else { |
1316 | CppParser parser(results); |
1317 | parser.namespaces = namespaces; |
1318 | parser.functionContext = functionContext; |
1319 | parser.functionContextUnresolved = functionContextUnresolved; |
1320 | parser.setInput(ts, fileName: cleanFile); |
1321 | parser.setTranslator(tor); |
1322 | QStringList stack = includeStack; |
1323 | stack << cleanFile; |
1324 | parser.parseInternal(cd, includeStack: stack, inclusions); |
1325 | // Avoid that messages obtained by direct scanning are used |
1326 | CppFiles::setBlacklisted(cleanFile); |
1327 | } |
1328 | inclusions.remove(value: cleanFile); |
1329 | |
1330 | prospectiveContext.clear(); |
1331 | pendingContext.clear(); |
1332 | } |
1333 | |
1334 | /* |
1335 | The third part of this source file is the parser. It accomplishes |
1336 | a very easy task: It finds all strings inside a tr() or translate() |
1337 | call, and possibly finds out the context of the call. It supports |
1338 | three cases: (1) the context is specified, as in |
1339 | FunnyDialog::tr("Hello") or translate("FunnyDialog", "Hello"); |
1340 | (2) the call appears within an inlined function; (3) the call |
1341 | appears within a function defined outside the class definition. |
1342 | */ |
1343 | |
1344 | bool CppParser::match(TokenType t) |
1345 | { |
1346 | bool matches = (yyTok == t); |
1347 | if (matches) |
1348 | yyTok = getToken(); |
1349 | return matches; |
1350 | } |
1351 | |
1352 | bool CppParser::matchString(QString *s) |
1353 | { |
1354 | bool matches = false; |
1355 | s->clear(); |
1356 | forever { |
1357 | if (yyTok != Tok_String && yyTok != Tok_RawString) |
1358 | return matches; |
1359 | matches = true; |
1360 | if (yyTok == Tok_String) |
1361 | *s += ParserTool::transcode(str: yyWord); |
1362 | else |
1363 | *s += yyWord; |
1364 | s->detach(); |
1365 | yyTok = getToken(); |
1366 | } |
1367 | } |
1368 | |
1369 | static const QString strQApplication = u"QApplication"_s ; |
1370 | static const QString strQCoreApplication = u"QCoreApplication"_s ; |
1371 | static const QString strUnicodeUTF8 = u"UnicodeUTF8"_s ; |
1372 | static const QString strDefaultCodec = u"DefaultCodec"_s ; |
1373 | static const QString strCodecForTr = u"CodecForTr"_s ; |
1374 | static const QString strLatin1 = u"Latin1"_s ; |
1375 | |
1376 | bool CppParser::matchEncoding() |
1377 | { |
1378 | if (yyTok != Tok_Ident) |
1379 | return false; |
1380 | if (yyWord == strQApplication || yyWord == strQCoreApplication) { |
1381 | yyTok = getToken(); |
1382 | if (yyTok == Tok_ColonColon) |
1383 | yyTok = getToken(); |
1384 | } |
1385 | if (yyWord == strUnicodeUTF8) { |
1386 | yyTok = getToken(); |
1387 | return true; |
1388 | } |
1389 | if (yyWord == strLatin1 || yyWord == strDefaultCodec || yyWord == strCodecForTr) |
1390 | yyMsg() << "Unsupported encoding Latin1/DefaultCodec/CodecForTr\n" ; |
1391 | return false; |
1392 | } |
1393 | |
1394 | bool CppParser::matchStringOrNull(QString *s) |
1395 | { |
1396 | return matchString(s) || match(t: Tok_Null); |
1397 | } |
1398 | |
1399 | /* |
1400 | * match any expression that can return a number, which can be |
1401 | * 1. Literal number (e.g. '11') |
1402 | * 2. simple identifier (e.g. 'm_count') |
1403 | * 3. simple function call (e.g. 'size()' ) |
1404 | * 4. function call on an object (e.g. 'list.size()') |
1405 | * 5. function call on an object (e.g. 'list->size()') |
1406 | * |
1407 | * Other cases: |
1408 | * size(2,4) |
1409 | * list().size() |
1410 | * list(a,b).size(2,4) |
1411 | * etc... |
1412 | */ |
1413 | bool CppParser::matchExpression() |
1414 | { |
1415 | if (match(t: Tok_Null) || match(t: Tok_Integer)) |
1416 | return true; |
1417 | |
1418 | int parenlevel = 0; |
1419 | while (match(t: Tok_Ident) || parenlevel > 0) { |
1420 | if (yyTok == Tok_RightParen) { |
1421 | if (parenlevel == 0) break; |
1422 | --parenlevel; |
1423 | yyTok = getToken(); |
1424 | } else if (yyTok == Tok_LeftParen) { |
1425 | yyTok = getToken(); |
1426 | if (yyTok == Tok_RightParen) { |
1427 | yyTok = getToken(); |
1428 | } else { |
1429 | ++parenlevel; |
1430 | } |
1431 | } else if (yyTok == Tok_Ident) { |
1432 | continue; |
1433 | } else if (yyTok == Tok_Arrow) { |
1434 | yyTok = getToken(); |
1435 | } else if (parenlevel == 0 || yyTok == Tok_Cancel) { |
1436 | return false; |
1437 | } |
1438 | } |
1439 | return true; |
1440 | } |
1441 | |
1442 | void CppParser::recordMessage(int line, const QString &context, const QString &text, const QString &, |
1443 | const QString &, const QString &msgid, const TranslatorMessage::ExtraData &, bool plural) |
1444 | { |
1445 | TranslatorMessage msg( |
1446 | ParserTool::transcode(str: context), text, ParserTool::transcode(str: comment), QString(), |
1447 | yyFileName, line, QStringList(), |
1448 | TranslatorMessage::Unfinished, plural); |
1449 | msg.setExtraComment(ParserTool::transcode(str: extracomment.simplified())); |
1450 | msg.setId(msgid); |
1451 | msg.setExtras(extra); |
1452 | tor->append(msg); |
1453 | } |
1454 | |
1455 | void CppParser::handleTr(QString &prefix, bool plural) |
1456 | { |
1457 | if (!sourcetext.isEmpty()) |
1458 | yyMsg() << "//% cannot be used with tr() / QT_TR_NOOP(). Ignoring\n" ; |
1459 | int line = yyLineNo; |
1460 | yyTok = getToken(); |
1461 | if (matchString(s: &text) && !text.isEmpty()) { |
1462 | comment.clear(); |
1463 | |
1464 | if (yyTok == Tok_RightParen) { |
1465 | // no comment |
1466 | } else if (match(t: Tok_Comma) && matchStringOrNull(s: &comment)) { //comment |
1467 | if (yyTok == Tok_RightParen) { |
1468 | // ok, |
1469 | } else if (match(t: Tok_Comma)) { |
1470 | plural = true; |
1471 | } |
1472 | } |
1473 | if (!pendingContext.isEmpty() && !prefix.startsWith(s: QLatin1String("::" ))) { |
1474 | NamespaceList unresolved; |
1475 | if (!fullyQualify(namespaces, quali: pendingContext, isDeclaration: true, resolved: &functionContext, unresolved: &unresolved)) { |
1476 | functionContextUnresolved = stringifyNamespace(start: 0, namespaces: unresolved); |
1477 | yyMsg() << qPrintable( |
1478 | QStringLiteral("Qualifying with unknown namespace/class %1::%2\n" ) |
1479 | .arg(stringifyNamespace(functionContext)).arg(unresolved.first().value())); |
1480 | } |
1481 | pendingContext.clear(); |
1482 | } |
1483 | if (prefix.isEmpty()) { |
1484 | if (functionContextUnresolved.isEmpty()) { |
1485 | int idx = functionContext.size(); |
1486 | if (idx < 2) { |
1487 | yyMsg() << "tr() cannot be called without context\n" ; |
1488 | return; |
1489 | } |
1490 | Namespace *fctx; |
1491 | while (!(fctx = findNamespace(namespaces: functionContext, nsCount: idx)->classDef)->hasTrFunctions) { |
1492 | if (idx == 1) { |
1493 | context = stringifyNamespace(namespaces: functionContext); |
1494 | fctx = findNamespace(namespaces: functionContext)->classDef; |
1495 | if (!fctx->complained) { |
1496 | yyMsg() << qPrintable( |
1497 | QStringLiteral("Class '%1' lacks Q_OBJECT macro\n" ).arg(context)); |
1498 | fctx->complained = true; |
1499 | } |
1500 | goto gotctx; |
1501 | } |
1502 | --idx; |
1503 | } |
1504 | if (fctx->trQualification.isEmpty()) { |
1505 | context.clear(); |
1506 | for (int i = 1;;) { |
1507 | context += functionContext.at(i).value(); |
1508 | if (++i == idx) |
1509 | break; |
1510 | context += QLatin1String("::" ); |
1511 | } |
1512 | fctx->trQualification = context; |
1513 | } else { |
1514 | context = fctx->trQualification; |
1515 | } |
1516 | } else { |
1517 | context = joinNamespaces(one: stringifyNamespace(namespaces: functionContext), two: functionContextUnresolved); |
1518 | } |
1519 | } else { |
1520 | #ifdef DIAGNOSE_RETRANSLATABILITY |
1521 | int last = prefix.lastIndexOf(QLatin1String("::" )); |
1522 | QString className = prefix.mid(last == -1 ? 0 : last + 2); |
1523 | if (!className.isEmpty() && className == functionName) { |
1524 | yyMsg() << qPrintable(QStringLiteral("It is not recommended to call tr() from within a constructor '%1::%2'\n" ) |
1525 | .arg(className).arg(functionName)); |
1526 | } |
1527 | #endif |
1528 | prefix.chop(n: 2); |
1529 | NamespaceList nsl; |
1530 | NamespaceList unresolved; |
1531 | if (fullyQualify(namespaces: functionContext, quali: prefix, isDeclaration: false, resolved: &nsl, unresolved: &unresolved)) { |
1532 | Namespace *fctx = findNamespace(namespaces: nsl)->classDef; |
1533 | if (fctx->trQualification.isEmpty()) { |
1534 | context = stringifyNamespace(namespaces: nsl); |
1535 | fctx->trQualification = context; |
1536 | } else { |
1537 | context = fctx->trQualification; |
1538 | } |
1539 | if (!fctx->hasTrFunctions && !fctx->complained) { |
1540 | yyMsg() << qPrintable(QStringLiteral("Class '%1' lacks Q_OBJECT macro\n" ) |
1541 | .arg(context)); |
1542 | fctx->complained = true; |
1543 | } |
1544 | } else { |
1545 | context = joinNamespaces(one: stringifyNamespace(namespaces: nsl), two: stringifyNamespace(start: 0, namespaces: unresolved)); |
1546 | } |
1547 | prefix.clear(); |
1548 | } |
1549 | |
1550 | gotctx: |
1551 | recordMessage(line, context, text, comment, extracomment, msgid, extra, plural); |
1552 | } |
1553 | sourcetext.clear(); // Will have warned about that already |
1554 | extracomment.clear(); |
1555 | msgid.clear(); |
1556 | extra.clear(); |
1557 | metaExpected = false; |
1558 | } |
1559 | |
1560 | void CppParser::handleTranslate(bool plural) |
1561 | { |
1562 | if (!sourcetext.isEmpty()) |
1563 | yyMsg() << "//% cannot be used with translate() / QT_TRANSLATE_NOOP(). Ignoring\n" ; |
1564 | int line = yyLineNo; |
1565 | yyTok = getToken(); |
1566 | if (matchString(s: &context) |
1567 | && match(t: Tok_Comma) |
1568 | && matchString(s: &text) && !text.isEmpty()) |
1569 | { |
1570 | comment.clear(); |
1571 | if (yyTok != Tok_RightParen) { |
1572 | // look for comment |
1573 | if (match(t: Tok_Comma) && matchStringOrNull(s: &comment)) { |
1574 | if (yyTok != Tok_RightParen) { |
1575 | // look for encoding |
1576 | if (match(t: Tok_Comma)) { |
1577 | if (matchEncoding()) { |
1578 | if (yyTok != Tok_RightParen) { |
1579 | // look for the plural quantifier, |
1580 | // this can be a number, an identifier or |
1581 | // a function call, |
1582 | // so for simplicity we mark it as plural if |
1583 | // we know we have a comma instead of an |
1584 | // right parentheses. |
1585 | plural |= match(t: Tok_Comma); |
1586 | } |
1587 | } else { |
1588 | // This can be a QTranslator::translate("context", |
1589 | // "source", "comment", n) plural translation |
1590 | if (matchExpression() && yyTok == Tok_RightParen) { |
1591 | plural = true; |
1592 | } else { |
1593 | return; |
1594 | } |
1595 | } |
1596 | } else { |
1597 | return; |
1598 | } |
1599 | } |
1600 | } else { |
1601 | return; |
1602 | } |
1603 | } |
1604 | recordMessage(line, context, text, comment, extracomment, msgid, extra, plural); |
1605 | } |
1606 | sourcetext.clear(); // Will have warned about that already |
1607 | extracomment.clear(); |
1608 | msgid.clear(); |
1609 | extra.clear(); |
1610 | metaExpected = false; |
1611 | } |
1612 | |
1613 | void CppParser::handleTrId(bool plural) |
1614 | { |
1615 | if (!msgid.isEmpty()) |
1616 | yyMsg() << "//= cannot be used with qtTrId() / QT_TRID_NOOP(). Ignoring\n" ; |
1617 | int line = yyLineNo; |
1618 | yyTok = getToken(); |
1619 | if (matchString(s: &msgid) && !msgid.isEmpty()) { |
1620 | plural |= match(t: Tok_Comma); |
1621 | recordMessage(line, context: QString(), text: ParserTool::transcode(str: sourcetext), comment: QString(), extracomment, |
1622 | msgid, extra, plural); |
1623 | } |
1624 | sourcetext.clear(); |
1625 | extracomment.clear(); |
1626 | msgid.clear(); |
1627 | extra.clear(); |
1628 | metaExpected = false; |
1629 | } |
1630 | |
1631 | void CppParser::handleDeclareTrFunctions() |
1632 | { |
1633 | QString name; |
1634 | forever { |
1635 | yyTok = getToken(); |
1636 | if (yyTok != Tok_Ident) |
1637 | return; |
1638 | name += yyWord; |
1639 | name.detach(); |
1640 | yyTok = getToken(); |
1641 | if (yyTok == Tok_RightParen) |
1642 | break; |
1643 | if (yyTok != Tok_ColonColon) |
1644 | return; |
1645 | name += QLatin1String("::" ); |
1646 | } |
1647 | Namespace *ns = modifyNamespace(namespaces: &namespaces); |
1648 | ns->hasTrFunctions = true; |
1649 | ns->trQualification = name; |
1650 | ns->trQualification.detach(); |
1651 | } |
1652 | |
1653 | void CppParser::parse(ConversionData &cd, const QStringList &includeStack, |
1654 | QSet<QString> &inclusions) |
1655 | { |
1656 | namespaces << HashString(); |
1657 | functionContext = namespaces; |
1658 | functionContextUnresolved.clear(); |
1659 | |
1660 | parseInternal(cd, includeStack, inclusions); |
1661 | } |
1662 | |
1663 | void CppParser::parseInternal(ConversionData &cd, const QStringList &includeStack, QSet<QString> &inclusions) |
1664 | { |
1665 | static QString strColons(QLatin1String("::" )); |
1666 | |
1667 | QString prefix; |
1668 | #ifdef DIAGNOSE_RETRANSLATABILITY |
1669 | QString functionName; |
1670 | #endif |
1671 | bool yyTokColonSeen = false; // Start of c'tor's initializer list |
1672 | bool yyTokIdentSeen = false; // Start of initializer (member or base class) |
1673 | bool maybeInTrailingReturnType = false; |
1674 | metaExpected = true; |
1675 | |
1676 | prospectiveContext.clear(); |
1677 | pendingContext.clear(); |
1678 | |
1679 | yyWord.reserve(asize: yyInStr.size()); // Rather insane. That's because we do no length checking. |
1680 | yyWordInitialCapacity = yyWord.capacity(); |
1681 | yyInPtr = (const ushort *)yyInStr.unicode(); |
1682 | yyCh = getChar(); |
1683 | yyTok = getToken(); |
1684 | while (yyTok != Tok_Eof) { |
1685 | // these are array indexing operations. we ignore them entirely |
1686 | // so they don't confuse our scoping of static initializers. |
1687 | // we enter the loop by either reading a left bracket or by an |
1688 | // #else popping the state. |
1689 | if (yyBracketDepth && yyBraceDepth == namespaceDepths.size()) { |
1690 | yyTok = getToken(); |
1691 | continue; |
1692 | } |
1693 | //qDebug() << "TOKEN: " << yyTok; |
1694 | switch (yyTok) { |
1695 | case Tok_QuotedInclude: { |
1696 | text = QDir(QFileInfo(yyFileName).absolutePath()).absoluteFilePath(fileName: yyWord); |
1697 | text.detach(); |
1698 | if (QFileInfo(text).isFile()) { |
1699 | processInclude(file: text, cd, includeStack, inclusions); |
1700 | yyTok = getToken(); |
1701 | break; |
1702 | } |
1703 | } |
1704 | Q_FALLTHROUGH(); |
1705 | case Tok_AngledInclude: { |
1706 | const QStringList cSources = cd.m_allCSources.values(key: yyWord); |
1707 | if (!cSources.isEmpty()) { |
1708 | for (const QString &cSource : cSources) |
1709 | processInclude(file: cSource, cd, includeStack, inclusions); |
1710 | goto incOk; |
1711 | } |
1712 | for (const QString &incPath : std::as_const(t&: cd.m_includePath)) { |
1713 | text = QDir(incPath).absoluteFilePath(fileName: yyWord); |
1714 | text.detach(); |
1715 | if (QFileInfo(text).isFile()) { |
1716 | processInclude(file: text, cd, includeStack, inclusions); |
1717 | goto incOk; |
1718 | } |
1719 | } |
1720 | incOk: |
1721 | yyTok = getToken(); |
1722 | break; |
1723 | } |
1724 | case Tok_friend: |
1725 | yyTok = getToken(); |
1726 | // These are forward declarations, so ignore them. |
1727 | if (yyTok == Tok_class) |
1728 | yyTok = getToken(); |
1729 | break; |
1730 | case Tok_class: |
1731 | /* |
1732 | Partial support for inlined functions. |
1733 | */ |
1734 | yyTok = getToken(); |
1735 | if (yyBraceDepth == namespaceDepths.size() && yyParenDepth == 0) { |
1736 | NamespaceList quali; |
1737 | HashString fct; |
1738 | |
1739 | // Find class name including qualification |
1740 | forever { |
1741 | text = yyWord; |
1742 | text.detach(); |
1743 | fct.setValue(text); |
1744 | yyTok = getToken(); |
1745 | |
1746 | if (yyTok == Tok_ColonColon) { |
1747 | quali << fct; |
1748 | yyTok = getToken(); |
1749 | } else if (yyTok == Tok_Ident) { |
1750 | if (yyWord == strfinal) { |
1751 | // C++11: final may appear immediately after the name of the class |
1752 | yyTok = getToken(); |
1753 | break; |
1754 | } |
1755 | |
1756 | // Handle impure definitions such as 'class Q_EXPORT QMessageBox', in |
1757 | // which case 'QMessageBox' is the class name, not 'Q_EXPORT', by |
1758 | // abandoning any qualification collected so far. |
1759 | quali.clear(); |
1760 | } else { |
1761 | break; |
1762 | } |
1763 | } |
1764 | |
1765 | if (yyTok == Tok_Colon || yyTok == Tok_AngleBracket) { |
1766 | // Skip any token until '{' or ';' since we might do things wrong if we find |
1767 | // a '::' or ':' token here. |
1768 | do { |
1769 | yyTok = getToken(); |
1770 | if (yyTok == Tok_Eof) |
1771 | goto goteof; |
1772 | if (yyTok == Tok_Cancel) |
1773 | goto case_default; |
1774 | } while (yyTok != Tok_LeftBrace && yyTok != Tok_Semicolon); |
1775 | } else { |
1776 | if (yyTok != Tok_LeftBrace) { |
1777 | // Obviously a forward declaration. We skip those, as they |
1778 | // don't create actually usable namespaces. |
1779 | break; |
1780 | } |
1781 | } |
1782 | |
1783 | if (!quali.isEmpty()) { |
1784 | // Forward-declared class definitions can be namespaced. |
1785 | NamespaceList nsl; |
1786 | if (!fullyQualify(namespaces, segments: quali, isDeclaration: true, resolved: &nsl, unresolved: 0)) { |
1787 | yyMsg() << "Ignoring definition of undeclared qualified class\n" ; |
1788 | break; |
1789 | } |
1790 | namespaceDepths.push(t: namespaces.size()); |
1791 | namespaces = nsl; |
1792 | } else { |
1793 | namespaceDepths.push(t: namespaces.size()); |
1794 | } |
1795 | enterNamespace(namespaces: &namespaces, name: fct); |
1796 | |
1797 | functionContext = namespaces; |
1798 | functionContextUnresolved.clear(); // Pointless |
1799 | prospectiveContext.clear(); |
1800 | pendingContext.clear(); |
1801 | |
1802 | metaExpected = true; |
1803 | yyTok = getToken(); |
1804 | } |
1805 | break; |
1806 | case Tok_namespace: |
1807 | yyTok = getToken(); |
1808 | if (yyTok == Tok_Ident) { |
1809 | text = yyWord; |
1810 | text.detach(); |
1811 | HashString ns = HashString(text); |
1812 | NamespaceList nestedNamespaces; |
1813 | forever { |
1814 | yyTok = getToken(); |
1815 | if (yyTok != Tok_ColonColon) |
1816 | break; |
1817 | yyTok = getToken(); |
1818 | if (yyTok != Tok_Ident) |
1819 | break; // whoops |
1820 | nestedNamespaces.append(t: ns); |
1821 | text = yyWord; |
1822 | text.detach(); |
1823 | ns = HashString(text); |
1824 | } |
1825 | if (yyTok == Tok_LeftBrace) { |
1826 | namespaceDepths.push(t: namespaces.size()); |
1827 | for (const auto &nns : nestedNamespaces) |
1828 | enterNamespace(namespaces: &namespaces, name: nns); |
1829 | enterNamespace(namespaces: &namespaces, name: ns); |
1830 | |
1831 | functionContext = namespaces; |
1832 | functionContextUnresolved.clear(); |
1833 | prospectiveContext.clear(); |
1834 | pendingContext.clear(); |
1835 | metaExpected = true; |
1836 | yyTok = getToken(); |
1837 | } else if (yyTok == Tok_Equals) { |
1838 | // e.g. namespace Is = OuterSpace::InnerSpace; |
1839 | // Note: 'Is' being qualified is invalid per C++17. |
1840 | NamespaceList fullName; |
1841 | yyTok = getToken(); |
1842 | if (yyTok == Tok_ColonColon) |
1843 | fullName.append(t: HashString(QString())); |
1844 | while (yyTok == Tok_ColonColon || yyTok == Tok_Ident) { |
1845 | if (yyTok == Tok_Ident) { |
1846 | text = yyWord; |
1847 | text.detach(); |
1848 | fullName.append(t: HashString(text)); |
1849 | } |
1850 | yyTok = getToken(); |
1851 | } |
1852 | if (fullName.isEmpty()) |
1853 | break; |
1854 | fullName.append(t: HashString(QString())); // Mark as unresolved |
1855 | modifyNamespace(namespaces: &namespaces)->aliases[ns] = fullName; |
1856 | } |
1857 | } else if (yyTok == Tok_LeftBrace) { |
1858 | // Anonymous namespace |
1859 | namespaceDepths.push(t: namespaces.size()); |
1860 | metaExpected = true; |
1861 | yyTok = getToken(); |
1862 | } |
1863 | break; |
1864 | case Tok_using: |
1865 | yyTok = getToken(); |
1866 | // XXX this should affect only the current scope, not the entire current namespace |
1867 | if (yyTok == Tok_namespace) { |
1868 | NamespaceList fullName; |
1869 | yyTok = getToken(); |
1870 | if (yyTok == Tok_ColonColon) |
1871 | fullName.append(t: HashString(QString())); |
1872 | while (yyTok == Tok_ColonColon || yyTok == Tok_Ident) { |
1873 | if (yyTok == Tok_Ident) { |
1874 | text = yyWord; |
1875 | text.detach(); |
1876 | fullName.append(t: HashString(text)); |
1877 | } |
1878 | yyTok = getToken(); |
1879 | } |
1880 | NamespaceList nsl; |
1881 | if (fullyQualify(namespaces, segments: fullName, isDeclaration: false, resolved: &nsl, unresolved: 0)) |
1882 | modifyNamespace(namespaces: &namespaces)->usings << HashStringList(nsl); |
1883 | } else { |
1884 | NamespaceList fullName; |
1885 | if (yyTok == Tok_ColonColon) |
1886 | fullName.append(t: HashString(QString())); |
1887 | while (yyTok == Tok_ColonColon || yyTok == Tok_Ident) { |
1888 | if (yyTok == Tok_Ident) { |
1889 | text = yyWord; |
1890 | text.detach(); |
1891 | fullName.append(t: HashString(text)); |
1892 | } |
1893 | yyTok = getToken(); |
1894 | } |
1895 | if (fullName.isEmpty()) |
1896 | break; |
1897 | // using-declarations cannot rename classes, so the last element of |
1898 | // fullName is already the resolved name we actually want. |
1899 | // As we do no resolution here, we'll collect useless usings of data |
1900 | // members and methods as well. This is no big deal. |
1901 | fullName.append(t: HashString(QString())); // Mark as unresolved |
1902 | const HashString &ns = *(fullName.constEnd() - 2); |
1903 | modifyNamespace(namespaces: &namespaces)->aliases[ns] = fullName; |
1904 | } |
1905 | break; |
1906 | case Tok_Q_OBJECT: |
1907 | modifyNamespace(namespaces: &namespaces)->hasTrFunctions = true; |
1908 | yyTok = getToken(); |
1909 | break; |
1910 | case Tok_Ident: |
1911 | if (yyTokColonSeen && |
1912 | yyBraceDepth == namespaceDepths.size() && yyParenDepth == 0) { |
1913 | // member or base class identifier |
1914 | yyTokIdentSeen = true; |
1915 | } |
1916 | yyTok = getToken(); |
1917 | if (yyTok == Tok_LeftParen) { |
1918 | bool forcePlural = false; |
1919 | switch (trFunctionAliasManager.trFunctionByName(trFunctionName: yyWord)) { |
1920 | case TrFunctionAliasManager::Function_Q_DECLARE_TR_FUNCTIONS: |
1921 | handleDeclareTrFunctions(); |
1922 | break; |
1923 | case TrFunctionAliasManager::Function_QT_TR_N_NOOP: |
1924 | forcePlural = true; |
1925 | Q_FALLTHROUGH(); |
1926 | case TrFunctionAliasManager::Function_tr: |
1927 | case TrFunctionAliasManager::Function_trUtf8: |
1928 | case TrFunctionAliasManager::Function_QT_TR_NOOP: |
1929 | case TrFunctionAliasManager::Function_QT_TR_NOOP_UTF8: |
1930 | if (tor) |
1931 | handleTr(prefix, plural: forcePlural); |
1932 | break; |
1933 | case TrFunctionAliasManager::Function_QT_TRANSLATE_N_NOOP: |
1934 | case TrFunctionAliasManager::Function_QT_TRANSLATE_N_NOOP3: |
1935 | forcePlural = true; |
1936 | Q_FALLTHROUGH(); |
1937 | case TrFunctionAliasManager::Function_translate: |
1938 | case TrFunctionAliasManager::Function_findMessage: |
1939 | case TrFunctionAliasManager::Function_QT_TRANSLATE_NOOP: |
1940 | case TrFunctionAliasManager::Function_QT_TRANSLATE_NOOP_UTF8: |
1941 | case TrFunctionAliasManager::Function_QT_TRANSLATE_NOOP3: |
1942 | case TrFunctionAliasManager::Function_QT_TRANSLATE_NOOP3_UTF8: |
1943 | if (tor) |
1944 | handleTranslate(plural: forcePlural); |
1945 | break; |
1946 | case TrFunctionAliasManager::Function_QT_TRID_N_NOOP: |
1947 | forcePlural = true; |
1948 | Q_FALLTHROUGH(); |
1949 | case TrFunctionAliasManager::Function_qtTrId: |
1950 | case TrFunctionAliasManager::Function_QT_TRID_NOOP: |
1951 | if (tor) |
1952 | handleTrId(plural: forcePlural); |
1953 | break; |
1954 | default: |
1955 | goto notrfunc; |
1956 | } |
1957 | yyTok = getToken(); |
1958 | break; |
1959 | } |
1960 | if (yyTok == Tok_ColonColon && !maybeInTrailingReturnType) { |
1961 | prefix += yyWord; |
1962 | prefix.detach(); |
1963 | } else { |
1964 | notrfunc: |
1965 | prefix.clear(); |
1966 | } |
1967 | metaExpected = false; |
1968 | break; |
1969 | case Tok_Arrow: |
1970 | if (yyParenDepth == 0 && yyBraceDepth == namespaceDepths.size()) |
1971 | maybeInTrailingReturnType = true; |
1972 | yyTok = getToken(); |
1973 | if (yyTok == Tok_Ident) { |
1974 | switch (trFunctionAliasManager.trFunctionByName(trFunctionName: yyWord)) { |
1975 | case TrFunctionAliasManager::Function_tr: |
1976 | case TrFunctionAliasManager::Function_trUtf8: |
1977 | yyMsg() << "Cannot invoke tr() like this\n" ; |
1978 | break; |
1979 | } |
1980 | } |
1981 | break; |
1982 | case Tok_ColonColon: |
1983 | if (yyTokIdentSeen || maybeInTrailingReturnType) { |
1984 | // member or base class identifier |
1985 | yyTok = getToken(); |
1986 | break; |
1987 | } |
1988 | if (yyBraceDepth == namespaceDepths.size() && yyParenDepth == 0 && !yyTokColonSeen) |
1989 | prospectiveContext = prefix; |
1990 | prefix += strColons; |
1991 | yyTok = getToken(); |
1992 | #ifdef DIAGNOSE_RETRANSLATABILITY |
1993 | if (yyTok == Tok_Ident && yyBraceDepth == namespaceDepths.count() && yyParenDepth == 0) { |
1994 | functionName = yyWord; |
1995 | functionName.detach(); |
1996 | } |
1997 | #endif |
1998 | break; |
1999 | case Tok_RightBrace: |
2000 | if (!yyTokColonSeen) { |
2001 | if (yyBraceDepth + 1 == namespaceDepths.size()) { |
2002 | // class or namespace |
2003 | truncateNamespaces(namespaces: &namespaces, length: namespaceDepths.pop()); |
2004 | } |
2005 | if (yyBraceDepth == namespaceDepths.size()) { |
2006 | // function, class or namespace |
2007 | if (!yyBraceDepth && !directInclude) |
2008 | truncateNamespaces(namespaces: &functionContext, length: 1); |
2009 | else |
2010 | functionContext = namespaces; |
2011 | functionContextUnresolved.clear(); |
2012 | pendingContext.clear(); |
2013 | } |
2014 | } |
2015 | Q_FALLTHROUGH(); |
2016 | case Tok_Semicolon: |
2017 | maybeInTrailingReturnType = false; |
2018 | prospectiveContext.clear(); |
2019 | prefix.clear(); |
2020 | if (!sourcetext.isEmpty() || !extracomment.isEmpty() || !msgid.isEmpty() || !extra.isEmpty()) { |
2021 | yyMsg() << "Discarding unconsumed meta data\n" ; |
2022 | sourcetext.clear(); |
2023 | extracomment.clear(); |
2024 | msgid.clear(); |
2025 | extra.clear(); |
2026 | } |
2027 | metaExpected = true; |
2028 | yyTok = getToken(); |
2029 | break; |
2030 | case Tok_Access: |
2031 | // Eat access specifiers, so their colons are not mistaken for c'tor initializer list starts |
2032 | do { |
2033 | yyTok = getToken(); |
2034 | } while (yyTok == Tok_Access); // Multiple specifiers are possible, e.g. "public slots" |
2035 | metaExpected = true; |
2036 | if (yyTok == Tok_Colon) |
2037 | goto case_default; |
2038 | break; |
2039 | case Tok_Colon: |
2040 | case Tok_Equals: |
2041 | if (yyBraceDepth == namespaceDepths.size() && yyParenDepth == 0) { |
2042 | if (!prospectiveContext.isEmpty()) { |
2043 | pendingContext = prospectiveContext; |
2044 | prospectiveContext.clear(); |
2045 | } |
2046 | //ignore colons for bitfields (are usually followed by a semicolon) |
2047 | if (yyTok == Tok_Colon) { |
2048 | if (lookAheadToSemicolonOrLeftBrace() != Tok_Semicolon) |
2049 | yyTokColonSeen = true; |
2050 | } |
2051 | } |
2052 | metaExpected = true; |
2053 | yyTok = getToken(); |
2054 | break; |
2055 | case Tok_LeftBrace: |
2056 | if (yyBraceDepth == namespaceDepths.size() + 1 && yyParenDepth == 0) { |
2057 | if (!prospectiveContext.isEmpty()) { |
2058 | pendingContext = prospectiveContext; |
2059 | prospectiveContext.clear(); |
2060 | } |
2061 | if (!yyTokIdentSeen) { |
2062 | // Function body |
2063 | yyTokColonSeen = false; |
2064 | } |
2065 | } |
2066 | maybeInTrailingReturnType = false; |
2067 | yyTokIdentSeen = false; |
2068 | metaExpected = true; |
2069 | yyTok = getToken(); |
2070 | break; |
2071 | case Tok_LeftParen: |
2072 | if (!yyTokColonSeen && yyBraceDepth == namespaceDepths.size() && yyParenDepth == 1 |
2073 | && !prospectiveContext.isEmpty()) { |
2074 | pendingContext = prospectiveContext; |
2075 | prospectiveContext.clear(); |
2076 | } |
2077 | yyTokIdentSeen = false; |
2078 | metaExpected = true; |
2079 | yyTok = getToken(); |
2080 | break; |
2081 | case Tok_Comma: |
2082 | case Tok_QuestionMark: |
2083 | metaExpected = true; |
2084 | yyTok = getToken(); |
2085 | break; |
2086 | case Tok_RightParen: |
2087 | if (yyParenDepth == 0) { |
2088 | if (!yyTokColonSeen && !pendingContext.isEmpty() |
2089 | && yyBraceDepth == namespaceDepths.size()) { |
2090 | // Demote the pendingContext to prospectiveContext. |
2091 | prospectiveContext = pendingContext; |
2092 | pendingContext.clear(); |
2093 | } |
2094 | metaExpected = true; |
2095 | } else { |
2096 | metaExpected = false; |
2097 | } |
2098 | yyTok = getToken(); |
2099 | break; |
2100 | case Tok_decltype: |
2101 | { |
2102 | // Save the parentheses depth outside the 'decltype' specifier. |
2103 | auto initialParenDepth = yyParenDepth; |
2104 | |
2105 | // Eat the opening parenthesis that follows 'decltype'. |
2106 | yyTok = getToken(); |
2107 | |
2108 | // Skip over everything within the parentheses that follow 'decltype'. |
2109 | while (yyParenDepth != initialParenDepth && yyTok != Tok_Eof) |
2110 | yyTok = getToken(); |
2111 | } |
2112 | break; |
2113 | case Tok_enum: |
2114 | yyTok = getToken(); |
2115 | // If it is an enum class then ignore |
2116 | if (yyTok == Tok_class) |
2117 | yyTok = getToken(); |
2118 | break; |
2119 | default: |
2120 | if (!yyParenDepth && !maybeInTrailingReturnType) |
2121 | prospectiveContext.clear(); |
2122 | Q_FALLTHROUGH(); |
2123 | case Tok_RightBracket: // ignoring indexing; for static initializers |
2124 | case_default: |
2125 | yyTok = getToken(); |
2126 | break; |
2127 | } |
2128 | } |
2129 | |
2130 | goteof: |
2131 | if (yyBraceDepth != 0) |
2132 | yyMsg(line: yyBraceLineNo) |
2133 | << "Unbalanced opening brace in C++ code (or abuse of the C++ preprocessor)\n" ; |
2134 | else if (yyParenDepth != 0) |
2135 | yyMsg(line: yyParenLineNo) |
2136 | << "Unbalanced opening parenthesis in C++ code" |
2137 | " (or abuse of the C++ preprocessor)\n" ; |
2138 | else if (yyBracketDepth != 0) |
2139 | yyMsg(line: yyBracketLineNo) |
2140 | << "Unbalanced opening bracket in C++ code" |
2141 | " (or abuse of the C++ preprocessor)\n" ; |
2142 | } |
2143 | |
2144 | void CppParser::() |
2145 | { |
2146 | if (!tor || !metaExpected) |
2147 | return; |
2148 | |
2149 | const QChar *ptr = yyWord.unicode(); |
2150 | if (*ptr == QLatin1Char(':') && ptr[1].isSpace()) { |
2151 | yyWord.remove(i: 0, len: 2); |
2152 | extracomment += yyWord; |
2153 | extracomment.detach(); |
2154 | } else if (*ptr == QLatin1Char('=') && ptr[1].isSpace()) { |
2155 | yyWord.remove(i: 0, len: 2); |
2156 | msgid = yyWord.simplified(); |
2157 | msgid.detach(); |
2158 | } else if (*ptr == QLatin1Char('~') && ptr[1].isSpace()) { |
2159 | yyWord.remove(i: 0, len: 2); |
2160 | text = yyWord.trimmed(); |
2161 | int k = text.indexOf(c: QLatin1Char(' ')); |
2162 | if (k > -1) { |
2163 | QString = text.mid(position: k + 1).trimmed(); |
2164 | if (commentvalue.startsWith(c: QLatin1Char('"')) && commentvalue.endsWith(c: QLatin1Char('"')) |
2165 | && commentvalue.size() != 1) { |
2166 | commentvalue = commentvalue.sliced(pos: 1, n: commentvalue.size() - 2); |
2167 | } |
2168 | extra.insert(key: text.left(n: k), value: commentvalue); |
2169 | } |
2170 | text.clear(); |
2171 | } else if (*ptr == QLatin1Char('%') && ptr[1].isSpace()) { |
2172 | sourcetext.reserve(asize: sourcetext.size() + yyWord.size() - 2); |
2173 | ushort *ptr = (ushort *)sourcetext.data() + sourcetext.size(); |
2174 | int p = 2, c; |
2175 | forever { |
2176 | if (p >= yyWord.size()) |
2177 | break; |
2178 | c = yyWord.unicode()[p++].unicode(); |
2179 | if (isspace(c)) |
2180 | continue; |
2181 | if (c != '"') { |
2182 | yyMsg() << "Unexpected character in meta string\n" ; |
2183 | break; |
2184 | } |
2185 | forever { |
2186 | if (p >= yyWord.size()) { |
2187 | whoops: |
2188 | yyMsg() << "Unterminated meta string\n" ; |
2189 | break; |
2190 | } |
2191 | c = yyWord.unicode()[p++].unicode(); |
2192 | if (c == '"') |
2193 | break; |
2194 | if (c == '\\') { |
2195 | if (p >= yyWord.size()) |
2196 | goto whoops; |
2197 | c = yyWord.unicode()[p++].unicode(); |
2198 | if (c == '\n') |
2199 | goto whoops; |
2200 | *ptr++ = '\\'; |
2201 | } |
2202 | *ptr++ = c; |
2203 | } |
2204 | } |
2205 | sourcetext.resize(size: ptr - (ushort *)sourcetext.data()); |
2206 | } else { |
2207 | const ushort *uc = (const ushort *)yyWord.unicode(); // Is zero-terminated |
2208 | int idx = 0; |
2209 | ushort c; |
2210 | while ((c = uc[idx]) == ' ' || c == '\t' || c == '\n') |
2211 | ++idx; |
2212 | if (!memcmp(s1: uc + idx, s2: CppMagicComment.unicode(), n: CppMagicComment.size() * 2)) { |
2213 | idx += CppMagicComment.size(); |
2214 | comment = QString::fromRawData(yyWord.unicode() + idx, |
2215 | size: yyWord.size() - idx).simplified(); |
2216 | int k = comment.indexOf(c: QLatin1Char(' ')); |
2217 | if (k == -1) { |
2218 | context = comment; |
2219 | } else { |
2220 | context = comment.left(n: k); |
2221 | comment.remove(i: 0, len: k + 1); |
2222 | TranslatorMessage msg( |
2223 | ParserTool::transcode(str: context), QString(), |
2224 | ParserTool::transcode(str: comment), QString(), |
2225 | yyFileName, yyLineNo, QStringList(), |
2226 | TranslatorMessage::Finished, false); |
2227 | msg.setExtraComment(ParserTool::transcode(str: extracomment.simplified())); |
2228 | extracomment.clear(); |
2229 | tor->append(msg); |
2230 | tor->setExtras(extra); |
2231 | extra.clear(); |
2232 | } |
2233 | } |
2234 | } |
2235 | } |
2236 | |
2237 | const ParseResults *CppParser::recordResults(bool ) |
2238 | { |
2239 | if (tor) { |
2240 | if (tor->messageCount()) { |
2241 | CppFiles::setTranslator(cleanFile: yyFileName, tor); |
2242 | } else { |
2243 | delete tor; |
2244 | tor = 0; |
2245 | } |
2246 | } |
2247 | if (isHeader) { |
2248 | const ParseResults *pr; |
2249 | if (!tor && results->includes.size() == 1 |
2250 | && results->rootNamespace.children.isEmpty() |
2251 | && results->rootNamespace.aliases.isEmpty() |
2252 | && results->rootNamespace.usings.isEmpty()) { |
2253 | // This is a forwarding header. Slash it. |
2254 | pr = *results->includes.cbegin(); |
2255 | delete results; |
2256 | } else { |
2257 | results->fileId = nextFileId++; |
2258 | pr = results; |
2259 | } |
2260 | CppFiles::setResults(cleanFile: yyFileName, results: pr); |
2261 | return pr; |
2262 | } else { |
2263 | delete results; |
2264 | return 0; |
2265 | } |
2266 | } |
2267 | |
2268 | void loadCPP(Translator &translator, const QStringList &filenames, ConversionData &cd) |
2269 | { |
2270 | QStringConverter::Encoding e = cd.m_sourceIsUtf16 ? QStringConverter::Utf16 : QStringConverter::Utf8; |
2271 | |
2272 | for (const QString &filename : filenames) { |
2273 | if (!CppFiles::getResults(cleanFile: filename).isEmpty() || CppFiles::isBlacklisted(cleanFile: filename)) |
2274 | continue; |
2275 | |
2276 | QFile file(filename); |
2277 | if (!file.open(flags: QIODevice::ReadOnly)) { |
2278 | cd.appendError(QStringLiteral("Cannot open %1: %2" ).arg(args: filename, |
2279 | args: file.errorString())); |
2280 | continue; |
2281 | } |
2282 | |
2283 | CppParser parser; |
2284 | QTextStream ts(&file); |
2285 | ts.setEncoding(e); |
2286 | ts.setAutoDetectUnicode(true); |
2287 | parser.setInput(ts, fileName: filename); |
2288 | Translator *tor = new Translator; |
2289 | parser.setTranslator(tor); |
2290 | QSet<QString> inclusions; |
2291 | parser.parse(cd, includeStack: QStringList(), inclusions); |
2292 | parser.recordResults(isHeader: isHeader(name: filename)); |
2293 | } |
2294 | |
2295 | for (const QString &filename : filenames) { |
2296 | if (!CppFiles::isBlacklisted(cleanFile: filename)) { |
2297 | if (const Translator *tor = CppFiles::getTranslator(cleanFile: filename)) { |
2298 | for (const TranslatorMessage &msg : tor->messages()) |
2299 | translator.extend(msg, cd); |
2300 | } |
2301 | } |
2302 | } |
2303 | } |
2304 | |
2305 | QT_END_NAMESPACE |
2306 | |