• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //===--- Preprocess.cpp - C Language Family Preprocessor Implementation ---===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 //  This file implements the Preprocessor interface.
11 //
12 //===----------------------------------------------------------------------===//
13 //
14 // Options to support:
15 //   -H       - Print the name of each header file used.
16 //   -d[DNI] - Dump various things.
17 //   -fworking-directory - #line's with preprocessor's working dir.
18 //   -fpreprocessed
19 //   -dependency-file,-M,-MM,-MF,-MG,-MP,-MT,-MQ,-MD,-MMD
20 //   -W*
21 //   -w
22 //
23 // Messages to emit:
24 //   "Multiple include guards may be useful for:\n"
25 //
26 //===----------------------------------------------------------------------===//
27 
28 #include "clang/Lex/Preprocessor.h"
29 #include "MacroArgs.h"
30 #include "clang/Lex/ExternalPreprocessorSource.h"
31 #include "clang/Lex/HeaderSearch.h"
32 #include "clang/Lex/MacroInfo.h"
33 #include "clang/Lex/Pragma.h"
34 #include "clang/Lex/PreprocessingRecord.h"
35 #include "clang/Lex/ScratchBuffer.h"
36 #include "clang/Lex/LexDiagnostic.h"
37 #include "clang/Lex/CodeCompletionHandler.h"
38 #include "clang/Lex/ModuleLoader.h"
39 #include "clang/Basic/SourceManager.h"
40 #include "clang/Basic/FileManager.h"
41 #include "clang/Basic/TargetInfo.h"
42 #include "llvm/ADT/APFloat.h"
43 #include "llvm/ADT/SmallString.h"
44 #include "llvm/Support/MemoryBuffer.h"
45 #include "llvm/Support/raw_ostream.h"
46 #include "llvm/Support/Capacity.h"
47 using namespace clang;
48 
49 //===----------------------------------------------------------------------===//
~ExternalPreprocessorSource()50 ExternalPreprocessorSource::~ExternalPreprocessorSource() { }
51 
Preprocessor(DiagnosticsEngine & diags,LangOptions & opts,const TargetInfo * target,SourceManager & SM,HeaderSearch & Headers,ModuleLoader & TheModuleLoader,IdentifierInfoLookup * IILookup,bool OwnsHeaders,bool DelayInitialization,bool IncrProcessing)52 Preprocessor::Preprocessor(DiagnosticsEngine &diags, LangOptions &opts,
53                            const TargetInfo *target, SourceManager &SM,
54                            HeaderSearch &Headers, ModuleLoader &TheModuleLoader,
55                            IdentifierInfoLookup* IILookup,
56                            bool OwnsHeaders,
57                            bool DelayInitialization,
58                            bool IncrProcessing)
59   : Diags(&diags), LangOpts(opts), Target(target),FileMgr(Headers.getFileMgr()),
60     SourceMgr(SM), HeaderInfo(Headers), TheModuleLoader(TheModuleLoader),
61     ExternalSource(0), Identifiers(opts, IILookup),
62     IncrementalProcessing(IncrProcessing), CodeComplete(0),
63     CodeCompletionFile(0), CodeCompletionOffset(0), CodeCompletionReached(0),
64     SkipMainFilePreamble(0, true), CurPPLexer(0),
65     CurDirLookup(0), CurLexerKind(CLK_Lexer), Callbacks(0), MacroArgCache(0),
66     Record(0), MIChainHead(0), MICache(0)
67 {
68   OwnsHeaderSearch = OwnsHeaders;
69 
70   ScratchBuf = new ScratchBuffer(SourceMgr);
71   CounterValue = 0; // __COUNTER__ starts at 0.
72 
73   // Clear stats.
74   NumDirectives = NumDefined = NumUndefined = NumPragma = 0;
75   NumIf = NumElse = NumEndif = 0;
76   NumEnteredSourceFiles = 0;
77   NumMacroExpanded = NumFnMacroExpanded = NumBuiltinMacroExpanded = 0;
78   NumFastMacroExpanded = NumTokenPaste = NumFastTokenPaste = 0;
79   MaxIncludeStackDepth = 0;
80   NumSkipped = 0;
81 
82   // Default to discarding comments.
83   KeepComments = false;
84   KeepMacroComments = false;
85   SuppressIncludeNotFoundError = false;
86 
87   // Macro expansion is enabled.
88   DisableMacroExpansion = false;
89   MacroExpansionInDirectivesOverride = false;
90   InMacroArgs = false;
91   InMacroArgPreExpansion = false;
92   NumCachedTokenLexers = 0;
93   PragmasEnabled = true;
94 
95   CachedLexPos = 0;
96 
97   // We haven't read anything from the external source.
98   ReadMacrosFromExternalSource = false;
99 
100   // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro.
101   // This gets unpoisoned where it is allowed.
102   (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned();
103   SetPoisonReason(Ident__VA_ARGS__,diag::ext_pp_bad_vaargs_use);
104 
105   // Initialize the pragma handlers.
106   PragmaHandlers = new PragmaNamespace(StringRef());
107   RegisterBuiltinPragmas();
108 
109   // Initialize builtin macros like __LINE__ and friends.
110   RegisterBuiltinMacros();
111 
112   if(LangOpts.Borland) {
113     Ident__exception_info        = getIdentifierInfo("_exception_info");
114     Ident___exception_info       = getIdentifierInfo("__exception_info");
115     Ident_GetExceptionInfo       = getIdentifierInfo("GetExceptionInformation");
116     Ident__exception_code        = getIdentifierInfo("_exception_code");
117     Ident___exception_code       = getIdentifierInfo("__exception_code");
118     Ident_GetExceptionCode       = getIdentifierInfo("GetExceptionCode");
119     Ident__abnormal_termination  = getIdentifierInfo("_abnormal_termination");
120     Ident___abnormal_termination = getIdentifierInfo("__abnormal_termination");
121     Ident_AbnormalTermination    = getIdentifierInfo("AbnormalTermination");
122   } else {
123     Ident__exception_info = Ident__exception_code = Ident__abnormal_termination = 0;
124     Ident___exception_info = Ident___exception_code = Ident___abnormal_termination = 0;
125     Ident_GetExceptionInfo = Ident_GetExceptionCode = Ident_AbnormalTermination = 0;
126   }
127 
128   if (!DelayInitialization) {
129     assert(Target && "Must provide target information for PP initialization");
130     Initialize(*Target);
131   }
132 }
133 
~Preprocessor()134 Preprocessor::~Preprocessor() {
135   assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!");
136 
137   while (!IncludeMacroStack.empty()) {
138     delete IncludeMacroStack.back().TheLexer;
139     delete IncludeMacroStack.back().TheTokenLexer;
140     IncludeMacroStack.pop_back();
141   }
142 
143   // Free any macro definitions.
144   for (MacroInfoChain *I = MIChainHead ; I ; I = I->Next)
145     I->MI.Destroy();
146 
147   // Free any cached macro expanders.
148   for (unsigned i = 0, e = NumCachedTokenLexers; i != e; ++i)
149     delete TokenLexerCache[i];
150 
151   // Free any cached MacroArgs.
152   for (MacroArgs *ArgList = MacroArgCache; ArgList; )
153     ArgList = ArgList->deallocate();
154 
155   // Release pragma information.
156   delete PragmaHandlers;
157 
158   // Delete the scratch buffer info.
159   delete ScratchBuf;
160 
161   // Delete the header search info, if we own it.
162   if (OwnsHeaderSearch)
163     delete &HeaderInfo;
164 
165   delete Callbacks;
166 }
167 
Initialize(const TargetInfo & Target)168 void Preprocessor::Initialize(const TargetInfo &Target) {
169   assert((!this->Target || this->Target == &Target) &&
170          "Invalid override of target information");
171   this->Target = &Target;
172 
173   // Initialize information about built-ins.
174   BuiltinInfo.InitializeTarget(Target);
175   HeaderInfo.setTarget(Target);
176 }
177 
setPTHManager(PTHManager * pm)178 void Preprocessor::setPTHManager(PTHManager* pm) {
179   PTH.reset(pm);
180   FileMgr.addStatCache(PTH->createStatCache());
181 }
182 
DumpToken(const Token & Tok,bool DumpFlags) const183 void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const {
184   llvm::errs() << tok::getTokenName(Tok.getKind()) << " '"
185                << getSpelling(Tok) << "'";
186 
187   if (!DumpFlags) return;
188 
189   llvm::errs() << "\t";
190   if (Tok.isAtStartOfLine())
191     llvm::errs() << " [StartOfLine]";
192   if (Tok.hasLeadingSpace())
193     llvm::errs() << " [LeadingSpace]";
194   if (Tok.isExpandDisabled())
195     llvm::errs() << " [ExpandDisabled]";
196   if (Tok.needsCleaning()) {
197     const char *Start = SourceMgr.getCharacterData(Tok.getLocation());
198     llvm::errs() << " [UnClean='" << StringRef(Start, Tok.getLength())
199                  << "']";
200   }
201 
202   llvm::errs() << "\tLoc=<";
203   DumpLocation(Tok.getLocation());
204   llvm::errs() << ">";
205 }
206 
DumpLocation(SourceLocation Loc) const207 void Preprocessor::DumpLocation(SourceLocation Loc) const {
208   Loc.dump(SourceMgr);
209 }
210 
DumpMacro(const MacroInfo & MI) const211 void Preprocessor::DumpMacro(const MacroInfo &MI) const {
212   llvm::errs() << "MACRO: ";
213   for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) {
214     DumpToken(MI.getReplacementToken(i));
215     llvm::errs() << "  ";
216   }
217   llvm::errs() << "\n";
218 }
219 
PrintStats()220 void Preprocessor::PrintStats() {
221   llvm::errs() << "\n*** Preprocessor Stats:\n";
222   llvm::errs() << NumDirectives << " directives found:\n";
223   llvm::errs() << "  " << NumDefined << " #define.\n";
224   llvm::errs() << "  " << NumUndefined << " #undef.\n";
225   llvm::errs() << "  #include/#include_next/#import:\n";
226   llvm::errs() << "    " << NumEnteredSourceFiles << " source files entered.\n";
227   llvm::errs() << "    " << MaxIncludeStackDepth << " max include stack depth\n";
228   llvm::errs() << "  " << NumIf << " #if/#ifndef/#ifdef.\n";
229   llvm::errs() << "  " << NumElse << " #else/#elif.\n";
230   llvm::errs() << "  " << NumEndif << " #endif.\n";
231   llvm::errs() << "  " << NumPragma << " #pragma.\n";
232   llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n";
233 
234   llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/"
235              << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, "
236              << NumFastMacroExpanded << " on the fast path.\n";
237   llvm::errs() << (NumFastTokenPaste+NumTokenPaste)
238              << " token paste (##) operations performed, "
239              << NumFastTokenPaste << " on the fast path.\n";
240 
241   llvm::errs() << "\nPreprocessor Memory: " << getTotalMemory() << "B total";
242 
243   llvm::errs() << "\n  BumpPtr: " << BP.getTotalMemory();
244   llvm::errs() << "\n  Macro Expanded Tokens: "
245                << llvm::capacity_in_bytes(MacroExpandedTokens);
246   llvm::errs() << "\n  Predefines Buffer: " << Predefines.capacity();
247   llvm::errs() << "\n  Macros: " << llvm::capacity_in_bytes(Macros);
248   llvm::errs() << "\n  #pragma push_macro Info: "
249                << llvm::capacity_in_bytes(PragmaPushMacroInfo);
250   llvm::errs() << "\n  Poison Reasons: "
251                << llvm::capacity_in_bytes(PoisonReasons);
252   llvm::errs() << "\n  Comment Handlers: "
253                << llvm::capacity_in_bytes(CommentHandlers) << "\n";
254 }
255 
256 Preprocessor::macro_iterator
macro_begin(bool IncludeExternalMacros) const257 Preprocessor::macro_begin(bool IncludeExternalMacros) const {
258   if (IncludeExternalMacros && ExternalSource &&
259       !ReadMacrosFromExternalSource) {
260     ReadMacrosFromExternalSource = true;
261     ExternalSource->ReadDefinedMacros();
262   }
263 
264   return Macros.begin();
265 }
266 
getTotalMemory() const267 size_t Preprocessor::getTotalMemory() const {
268   return BP.getTotalMemory()
269     + llvm::capacity_in_bytes(MacroExpandedTokens)
270     + Predefines.capacity() /* Predefines buffer. */
271     + llvm::capacity_in_bytes(Macros)
272     + llvm::capacity_in_bytes(PragmaPushMacroInfo)
273     + llvm::capacity_in_bytes(PoisonReasons)
274     + llvm::capacity_in_bytes(CommentHandlers);
275 }
276 
277 Preprocessor::macro_iterator
macro_end(bool IncludeExternalMacros) const278 Preprocessor::macro_end(bool IncludeExternalMacros) const {
279   if (IncludeExternalMacros && ExternalSource &&
280       !ReadMacrosFromExternalSource) {
281     ReadMacrosFromExternalSource = true;
282     ExternalSource->ReadDefinedMacros();
283   }
284 
285   return Macros.end();
286 }
287 
recomputeCurLexerKind()288 void Preprocessor::recomputeCurLexerKind() {
289   if (CurLexer)
290     CurLexerKind = CLK_Lexer;
291   else if (CurPTHLexer)
292     CurLexerKind = CLK_PTHLexer;
293   else if (CurTokenLexer)
294     CurLexerKind = CLK_TokenLexer;
295   else
296     CurLexerKind = CLK_CachingLexer;
297 }
298 
SetCodeCompletionPoint(const FileEntry * File,unsigned CompleteLine,unsigned CompleteColumn)299 bool Preprocessor::SetCodeCompletionPoint(const FileEntry *File,
300                                           unsigned CompleteLine,
301                                           unsigned CompleteColumn) {
302   assert(File);
303   assert(CompleteLine && CompleteColumn && "Starts from 1:1");
304   assert(!CodeCompletionFile && "Already set");
305 
306   using llvm::MemoryBuffer;
307 
308   // Load the actual file's contents.
309   bool Invalid = false;
310   const MemoryBuffer *Buffer = SourceMgr.getMemoryBufferForFile(File, &Invalid);
311   if (Invalid)
312     return true;
313 
314   // Find the byte position of the truncation point.
315   const char *Position = Buffer->getBufferStart();
316   for (unsigned Line = 1; Line < CompleteLine; ++Line) {
317     for (; *Position; ++Position) {
318       if (*Position != '\r' && *Position != '\n')
319         continue;
320 
321       // Eat \r\n or \n\r as a single line.
322       if ((Position[1] == '\r' || Position[1] == '\n') &&
323           Position[0] != Position[1])
324         ++Position;
325       ++Position;
326       break;
327     }
328   }
329 
330   Position += CompleteColumn - 1;
331 
332   // Insert '\0' at the code-completion point.
333   if (Position < Buffer->getBufferEnd()) {
334     CodeCompletionFile = File;
335     CodeCompletionOffset = Position - Buffer->getBufferStart();
336 
337     MemoryBuffer *NewBuffer =
338         MemoryBuffer::getNewUninitMemBuffer(Buffer->getBufferSize() + 1,
339                                             Buffer->getBufferIdentifier());
340     char *NewBuf = const_cast<char*>(NewBuffer->getBufferStart());
341     char *NewPos = std::copy(Buffer->getBufferStart(), Position, NewBuf);
342     *NewPos = '\0';
343     std::copy(Position, Buffer->getBufferEnd(), NewPos+1);
344     SourceMgr.overrideFileContents(File, NewBuffer);
345   }
346 
347   return false;
348 }
349 
CodeCompleteNaturalLanguage()350 void Preprocessor::CodeCompleteNaturalLanguage() {
351   if (CodeComplete)
352     CodeComplete->CodeCompleteNaturalLanguage();
353   setCodeCompletionReached();
354 }
355 
356 /// getSpelling - This method is used to get the spelling of a token into a
357 /// SmallVector. Note that the returned StringRef may not point to the
358 /// supplied buffer if a copy can be avoided.
getSpelling(const Token & Tok,SmallVectorImpl<char> & Buffer,bool * Invalid) const359 StringRef Preprocessor::getSpelling(const Token &Tok,
360                                           SmallVectorImpl<char> &Buffer,
361                                           bool *Invalid) const {
362   // NOTE: this has to be checked *before* testing for an IdentifierInfo.
363   if (Tok.isNot(tok::raw_identifier)) {
364     // Try the fast path.
365     if (const IdentifierInfo *II = Tok.getIdentifierInfo())
366       return II->getName();
367   }
368 
369   // Resize the buffer if we need to copy into it.
370   if (Tok.needsCleaning())
371     Buffer.resize(Tok.getLength());
372 
373   const char *Ptr = Buffer.data();
374   unsigned Len = getSpelling(Tok, Ptr, Invalid);
375   return StringRef(Ptr, Len);
376 }
377 
378 /// CreateString - Plop the specified string into a scratch buffer and return a
379 /// location for it.  If specified, the source location provides a source
380 /// location for the token.
CreateString(const char * Buf,unsigned Len,Token & Tok,SourceLocation ExpansionLocStart,SourceLocation ExpansionLocEnd)381 void Preprocessor::CreateString(const char *Buf, unsigned Len, Token &Tok,
382                                 SourceLocation ExpansionLocStart,
383                                 SourceLocation ExpansionLocEnd) {
384   Tok.setLength(Len);
385 
386   const char *DestPtr;
387   SourceLocation Loc = ScratchBuf->getToken(Buf, Len, DestPtr);
388 
389   if (ExpansionLocStart.isValid())
390     Loc = SourceMgr.createExpansionLoc(Loc, ExpansionLocStart,
391                                        ExpansionLocEnd, Len);
392   Tok.setLocation(Loc);
393 
394   // If this is a raw identifier or a literal token, set the pointer data.
395   if (Tok.is(tok::raw_identifier))
396     Tok.setRawIdentifierData(DestPtr);
397   else if (Tok.isLiteral())
398     Tok.setLiteralData(DestPtr);
399 }
400 
getCurrentModule()401 Module *Preprocessor::getCurrentModule() {
402   if (getLangOpts().CurrentModule.empty())
403     return 0;
404 
405   return getHeaderSearchInfo().lookupModule(getLangOpts().CurrentModule);
406 }
407 
408 //===----------------------------------------------------------------------===//
409 // Preprocessor Initialization Methods
410 //===----------------------------------------------------------------------===//
411 
412 
413 /// EnterMainSourceFile - Enter the specified FileID as the main source file,
414 /// which implicitly adds the builtin defines etc.
EnterMainSourceFile()415 void Preprocessor::EnterMainSourceFile() {
416   // We do not allow the preprocessor to reenter the main file.  Doing so will
417   // cause FileID's to accumulate information from both runs (e.g. #line
418   // information) and predefined macros aren't guaranteed to be set properly.
419   assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!");
420   FileID MainFileID = SourceMgr.getMainFileID();
421 
422   // If MainFileID is loaded it means we loaded an AST file, no need to enter
423   // a main file.
424   if (!SourceMgr.isLoadedFileID(MainFileID)) {
425     // Enter the main file source buffer.
426     EnterSourceFile(MainFileID, 0, SourceLocation());
427 
428     // If we've been asked to skip bytes in the main file (e.g., as part of a
429     // precompiled preamble), do so now.
430     if (SkipMainFilePreamble.first > 0)
431       CurLexer->SkipBytes(SkipMainFilePreamble.first,
432                           SkipMainFilePreamble.second);
433 
434     // Tell the header info that the main file was entered.  If the file is later
435     // #imported, it won't be re-entered.
436     if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID))
437       HeaderInfo.IncrementIncludeCount(FE);
438   }
439 
440   // Preprocess Predefines to populate the initial preprocessor state.
441   llvm::MemoryBuffer *SB =
442     llvm::MemoryBuffer::getMemBufferCopy(Predefines, "<built-in>");
443   assert(SB && "Cannot create predefined source buffer");
444   FileID FID = SourceMgr.createFileIDForMemBuffer(SB);
445   assert(!FID.isInvalid() && "Could not create FileID for predefines?");
446 
447   // Start parsing the predefines.
448   EnterSourceFile(FID, 0, SourceLocation());
449 }
450 
EndSourceFile()451 void Preprocessor::EndSourceFile() {
452   // Notify the client that we reached the end of the source file.
453   if (Callbacks)
454     Callbacks->EndOfMainFile();
455 }
456 
457 //===----------------------------------------------------------------------===//
458 // Lexer Event Handling.
459 //===----------------------------------------------------------------------===//
460 
461 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the
462 /// identifier information for the token and install it into the token,
463 /// updating the token kind accordingly.
LookUpIdentifierInfo(Token & Identifier) const464 IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier) const {
465   assert(Identifier.getRawIdentifierData() != 0 && "No raw identifier data!");
466 
467   // Look up this token, see if it is a macro, or if it is a language keyword.
468   IdentifierInfo *II;
469   if (!Identifier.needsCleaning()) {
470     // No cleaning needed, just use the characters from the lexed buffer.
471     II = getIdentifierInfo(StringRef(Identifier.getRawIdentifierData(),
472                                            Identifier.getLength()));
473   } else {
474     // Cleaning needed, alloca a buffer, clean into it, then use the buffer.
475     SmallString<64> IdentifierBuffer;
476     StringRef CleanedStr = getSpelling(Identifier, IdentifierBuffer);
477     II = getIdentifierInfo(CleanedStr);
478   }
479 
480   // Update the token info (identifier info and appropriate token kind).
481   Identifier.setIdentifierInfo(II);
482   Identifier.setKind(II->getTokenID());
483 
484   return II;
485 }
486 
SetPoisonReason(IdentifierInfo * II,unsigned DiagID)487 void Preprocessor::SetPoisonReason(IdentifierInfo *II, unsigned DiagID) {
488   PoisonReasons[II] = DiagID;
489 }
490 
PoisonSEHIdentifiers(bool Poison)491 void Preprocessor::PoisonSEHIdentifiers(bool Poison) {
492   assert(Ident__exception_code && Ident__exception_info);
493   assert(Ident___exception_code && Ident___exception_info);
494   Ident__exception_code->setIsPoisoned(Poison);
495   Ident___exception_code->setIsPoisoned(Poison);
496   Ident_GetExceptionCode->setIsPoisoned(Poison);
497   Ident__exception_info->setIsPoisoned(Poison);
498   Ident___exception_info->setIsPoisoned(Poison);
499   Ident_GetExceptionInfo->setIsPoisoned(Poison);
500   Ident__abnormal_termination->setIsPoisoned(Poison);
501   Ident___abnormal_termination->setIsPoisoned(Poison);
502   Ident_AbnormalTermination->setIsPoisoned(Poison);
503 }
504 
HandlePoisonedIdentifier(Token & Identifier)505 void Preprocessor::HandlePoisonedIdentifier(Token & Identifier) {
506   assert(Identifier.getIdentifierInfo() &&
507          "Can't handle identifiers without identifier info!");
508   llvm::DenseMap<IdentifierInfo*,unsigned>::const_iterator it =
509     PoisonReasons.find(Identifier.getIdentifierInfo());
510   if(it == PoisonReasons.end())
511     Diag(Identifier, diag::err_pp_used_poisoned_id);
512   else
513     Diag(Identifier,it->second) << Identifier.getIdentifierInfo();
514 }
515 
516 /// HandleIdentifier - This callback is invoked when the lexer reads an
517 /// identifier.  This callback looks up the identifier in the map and/or
518 /// potentially macro expands it or turns it into a named token (like 'for').
519 ///
520 /// Note that callers of this method are guarded by checking the
521 /// IdentifierInfo's 'isHandleIdentifierCase' bit.  If this method changes, the
522 /// IdentifierInfo methods that compute these properties will need to change to
523 /// match.
HandleIdentifier(Token & Identifier)524 void Preprocessor::HandleIdentifier(Token &Identifier) {
525   assert(Identifier.getIdentifierInfo() &&
526          "Can't handle identifiers without identifier info!");
527 
528   IdentifierInfo &II = *Identifier.getIdentifierInfo();
529 
530   // If the information about this identifier is out of date, update it from
531   // the external source.
532   // We have to treat __VA_ARGS__ in a special way, since it gets
533   // serialized with isPoisoned = true, but our preprocessor may have
534   // unpoisoned it if we're defining a C99 macro.
535   if (II.isOutOfDate()) {
536     bool CurrentIsPoisoned = false;
537     if (&II == Ident__VA_ARGS__)
538       CurrentIsPoisoned = Ident__VA_ARGS__->isPoisoned();
539 
540     ExternalSource->updateOutOfDateIdentifier(II);
541     Identifier.setKind(II.getTokenID());
542 
543     if (&II == Ident__VA_ARGS__)
544       II.setIsPoisoned(CurrentIsPoisoned);
545   }
546 
547   // If this identifier was poisoned, and if it was not produced from a macro
548   // expansion, emit an error.
549   if (II.isPoisoned() && CurPPLexer) {
550     HandlePoisonedIdentifier(Identifier);
551   }
552 
553   // If this is a macro to be expanded, do it.
554   if (MacroInfo *MI = getMacroInfo(&II)) {
555     if (!DisableMacroExpansion) {
556       if (Identifier.isExpandDisabled()) {
557         Diag(Identifier, diag::pp_disabled_macro_expansion);
558       } else if (MI->isEnabled()) {
559         if (!HandleMacroExpandedIdentifier(Identifier, MI))
560           return;
561       } else {
562         // C99 6.10.3.4p2 says that a disabled macro may never again be
563         // expanded, even if it's in a context where it could be expanded in the
564         // future.
565         Identifier.setFlag(Token::DisableExpand);
566         Diag(Identifier, diag::pp_disabled_macro_expansion);
567       }
568     }
569   }
570 
571   // If this identifier is a keyword in C++11, produce a warning. Don't warn if
572   // we're not considering macro expansion, since this identifier might be the
573   // name of a macro.
574   // FIXME: This warning is disabled in cases where it shouldn't be, like
575   //   "#define constexpr constexpr", "int constexpr;"
576   if (II.isCXX11CompatKeyword() & !DisableMacroExpansion) {
577     Diag(Identifier, diag::warn_cxx11_keyword) << II.getName();
578     // Don't diagnose this keyword again in this translation unit.
579     II.setIsCXX11CompatKeyword(false);
580   }
581 
582   // C++ 2.11p2: If this is an alternative representation of a C++ operator,
583   // then we act as if it is the actual operator and not the textual
584   // representation of it.
585   if (II.isCPlusPlusOperatorKeyword())
586     Identifier.setIdentifierInfo(0);
587 
588   // If this is an extension token, diagnose its use.
589   // We avoid diagnosing tokens that originate from macro definitions.
590   // FIXME: This warning is disabled in cases where it shouldn't be,
591   // like "#define TY typeof", "TY(1) x".
592   if (II.isExtensionToken() && !DisableMacroExpansion)
593     Diag(Identifier, diag::ext_token_used);
594 
595   // If this is the '__experimental_modules_import' contextual keyword, note
596   // that the next token indicates a module name.
597   //
598   // Note that we do not treat '__experimental_modules_import' as a contextual
599   // keyword when we're in a caching lexer, because caching lexers only get
600   // used in contexts where import declarations are disallowed.
601   if (II.isModulesImport() && !InMacroArgs && !DisableMacroExpansion &&
602       getLangOpts().Modules && CurLexerKind != CLK_CachingLexer) {
603     ModuleImportLoc = Identifier.getLocation();
604     ModuleImportPath.clear();
605     ModuleImportExpectsIdentifier = true;
606     CurLexerKind = CLK_LexAfterModuleImport;
607   }
608 }
609 
610 /// \brief Lex a token following the 'import' contextual keyword.
611 ///
LexAfterModuleImport(Token & Result)612 void Preprocessor::LexAfterModuleImport(Token &Result) {
613   // Figure out what kind of lexer we actually have.
614   recomputeCurLexerKind();
615 
616   // Lex the next token.
617   Lex(Result);
618 
619   // The token sequence
620   //
621   //   import identifier (. identifier)*
622   //
623   // indicates a module import directive. We already saw the 'import'
624   // contextual keyword, so now we're looking for the identifiers.
625   if (ModuleImportExpectsIdentifier && Result.getKind() == tok::identifier) {
626     // We expected to see an identifier here, and we did; continue handling
627     // identifiers.
628     ModuleImportPath.push_back(std::make_pair(Result.getIdentifierInfo(),
629                                               Result.getLocation()));
630     ModuleImportExpectsIdentifier = false;
631     CurLexerKind = CLK_LexAfterModuleImport;
632     return;
633   }
634 
635   // If we're expecting a '.' or a ';', and we got a '.', then wait until we
636   // see the next identifier.
637   if (!ModuleImportExpectsIdentifier && Result.getKind() == tok::period) {
638     ModuleImportExpectsIdentifier = true;
639     CurLexerKind = CLK_LexAfterModuleImport;
640     return;
641   }
642 
643   // If we have a non-empty module path, load the named module.
644   if (!ModuleImportPath.empty())
645     (void)TheModuleLoader.loadModule(ModuleImportLoc, ModuleImportPath,
646                                      Module::MacrosVisible,
647                                      /*IsIncludeDirective=*/false);
648 }
649 
addCommentHandler(CommentHandler * Handler)650 void Preprocessor::addCommentHandler(CommentHandler *Handler) {
651   assert(Handler && "NULL comment handler");
652   assert(std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler) ==
653          CommentHandlers.end() && "Comment handler already registered");
654   CommentHandlers.push_back(Handler);
655 }
656 
removeCommentHandler(CommentHandler * Handler)657 void Preprocessor::removeCommentHandler(CommentHandler *Handler) {
658   std::vector<CommentHandler *>::iterator Pos
659   = std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler);
660   assert(Pos != CommentHandlers.end() && "Comment handler not registered");
661   CommentHandlers.erase(Pos);
662 }
663 
HandleComment(Token & result,SourceRange Comment)664 bool Preprocessor::HandleComment(Token &result, SourceRange Comment) {
665   bool AnyPendingTokens = false;
666   for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(),
667        HEnd = CommentHandlers.end();
668        H != HEnd; ++H) {
669     if ((*H)->HandleComment(*this, Comment))
670       AnyPendingTokens = true;
671   }
672   if (!AnyPendingTokens || getCommentRetentionState())
673     return false;
674   Lex(result);
675   return true;
676 }
677 
~ModuleLoader()678 ModuleLoader::~ModuleLoader() { }
679 
~CommentHandler()680 CommentHandler::~CommentHandler() { }
681 
~CodeCompletionHandler()682 CodeCompletionHandler::~CodeCompletionHandler() { }
683 
createPreprocessingRecord(bool RecordConditionalDirectives)684 void Preprocessor::createPreprocessingRecord(bool RecordConditionalDirectives) {
685   if (Record)
686     return;
687 
688   Record = new PreprocessingRecord(getSourceManager(),
689                                    RecordConditionalDirectives);
690   addPPCallbacks(Record);
691 }
692