• 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 "clang/Basic/FileManager.h"
30 #include "clang/Basic/SourceManager.h"
31 #include "clang/Basic/TargetInfo.h"
32 #include "clang/Lex/CodeCompletionHandler.h"
33 #include "clang/Lex/ExternalPreprocessorSource.h"
34 #include "clang/Lex/HeaderSearch.h"
35 #include "clang/Lex/LexDiagnostic.h"
36 #include "clang/Lex/LiteralSupport.h"
37 #include "clang/Lex/MacroArgs.h"
38 #include "clang/Lex/MacroInfo.h"
39 #include "clang/Lex/ModuleLoader.h"
40 #include "clang/Lex/Pragma.h"
41 #include "clang/Lex/PreprocessingRecord.h"
42 #include "clang/Lex/PreprocessorOptions.h"
43 #include "clang/Lex/ScratchBuffer.h"
44 #include "llvm/ADT/APFloat.h"
45 #include "llvm/ADT/STLExtras.h"
46 #include "llvm/ADT/SmallString.h"
47 #include "llvm/ADT/StringExtras.h"
48 #include "llvm/Support/Capacity.h"
49 #include "llvm/Support/ConvertUTF.h"
50 #include "llvm/Support/MemoryBuffer.h"
51 #include "llvm/Support/raw_ostream.h"
52 using namespace clang;
53 
54 //===----------------------------------------------------------------------===//
~ExternalPreprocessorSource()55 ExternalPreprocessorSource::~ExternalPreprocessorSource() { }
56 
Preprocessor(IntrusiveRefCntPtr<PreprocessorOptions> PPOpts,DiagnosticsEngine & diags,LangOptions & opts,SourceManager & SM,HeaderSearch & Headers,ModuleLoader & TheModuleLoader,IdentifierInfoLookup * IILookup,bool OwnsHeaders,TranslationUnitKind TUKind)57 Preprocessor::Preprocessor(IntrusiveRefCntPtr<PreprocessorOptions> PPOpts,
58                            DiagnosticsEngine &diags, LangOptions &opts,
59                            SourceManager &SM, HeaderSearch &Headers,
60                            ModuleLoader &TheModuleLoader,
61                            IdentifierInfoLookup *IILookup, bool OwnsHeaders,
62                            TranslationUnitKind TUKind)
63     : PPOpts(PPOpts), Diags(&diags), LangOpts(opts), Target(nullptr),
64       FileMgr(Headers.getFileMgr()), SourceMgr(SM), HeaderInfo(Headers),
65       TheModuleLoader(TheModuleLoader), ExternalSource(nullptr),
66       Identifiers(opts, IILookup), IncrementalProcessing(false), TUKind(TUKind),
67       CodeComplete(nullptr), CodeCompletionFile(nullptr),
68       CodeCompletionOffset(0), LastTokenWasAt(false),
69       ModuleImportExpectsIdentifier(false), CodeCompletionReached(0),
70       SkipMainFilePreamble(0, true), CurPPLexer(nullptr),
71       CurDirLookup(nullptr), CurLexerKind(CLK_Lexer), CurSubmodule(nullptr),
72       Callbacks(nullptr), MacroArgCache(nullptr), Record(nullptr),
73       MIChainHead(nullptr), MICache(nullptr), DeserialMIChainHead(nullptr) {
74   OwnsHeaderSearch = OwnsHeaders;
75 
76   ScratchBuf = new ScratchBuffer(SourceMgr);
77   CounterValue = 0; // __COUNTER__ starts at 0.
78 
79   // Clear stats.
80   NumDirectives = NumDefined = NumUndefined = NumPragma = 0;
81   NumIf = NumElse = NumEndif = 0;
82   NumEnteredSourceFiles = 0;
83   NumMacroExpanded = NumFnMacroExpanded = NumBuiltinMacroExpanded = 0;
84   NumFastMacroExpanded = NumTokenPaste = NumFastTokenPaste = 0;
85   MaxIncludeStackDepth = 0;
86   NumSkipped = 0;
87 
88   // Default to discarding comments.
89   KeepComments = false;
90   KeepMacroComments = false;
91   SuppressIncludeNotFoundError = false;
92 
93   // Macro expansion is enabled.
94   DisableMacroExpansion = false;
95   MacroExpansionInDirectivesOverride = false;
96   InMacroArgs = false;
97   InMacroArgPreExpansion = false;
98   NumCachedTokenLexers = 0;
99   PragmasEnabled = true;
100   ParsingIfOrElifDirective = false;
101   PreprocessedOutput = false;
102 
103   CachedLexPos = 0;
104 
105   // We haven't read anything from the external source.
106   ReadMacrosFromExternalSource = false;
107 
108   // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro.
109   // This gets unpoisoned where it is allowed.
110   (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned();
111   SetPoisonReason(Ident__VA_ARGS__,diag::ext_pp_bad_vaargs_use);
112 
113   // Initialize the pragma handlers.
114   PragmaHandlers = new PragmaNamespace(StringRef());
115   RegisterBuiltinPragmas();
116 
117   // Initialize builtin macros like __LINE__ and friends.
118   RegisterBuiltinMacros();
119 
120   if(LangOpts.Borland) {
121     Ident__exception_info        = getIdentifierInfo("_exception_info");
122     Ident___exception_info       = getIdentifierInfo("__exception_info");
123     Ident_GetExceptionInfo       = getIdentifierInfo("GetExceptionInformation");
124     Ident__exception_code        = getIdentifierInfo("_exception_code");
125     Ident___exception_code       = getIdentifierInfo("__exception_code");
126     Ident_GetExceptionCode       = getIdentifierInfo("GetExceptionCode");
127     Ident__abnormal_termination  = getIdentifierInfo("_abnormal_termination");
128     Ident___abnormal_termination = getIdentifierInfo("__abnormal_termination");
129     Ident_AbnormalTermination    = getIdentifierInfo("AbnormalTermination");
130   } else {
131     Ident__exception_info = Ident__exception_code = nullptr;
132     Ident__abnormal_termination = Ident___exception_info = nullptr;
133     Ident___exception_code = Ident___abnormal_termination = nullptr;
134     Ident_GetExceptionInfo = Ident_GetExceptionCode = nullptr;
135     Ident_AbnormalTermination = nullptr;
136   }
137 }
138 
~Preprocessor()139 Preprocessor::~Preprocessor() {
140   assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!");
141 
142   IncludeMacroStack.clear();
143 
144   // Free any macro definitions.
145   for (MacroInfoChain *I = MIChainHead ; I ; I = I->Next)
146     I->MI.Destroy();
147 
148   // Free any cached macro expanders.
149   // This populates MacroArgCache, so all TokenLexers need to be destroyed
150   // before the code below that frees up the MacroArgCache list.
151   for (unsigned i = 0, e = NumCachedTokenLexers; i != e; ++i)
152     delete TokenLexerCache[i];
153   CurTokenLexer.reset();
154 
155   for (DeserializedMacroInfoChain *I = DeserialMIChainHead ; I ; I = I->Next)
156     I->MI.Destroy();
157 
158   // Free any cached MacroArgs.
159   for (MacroArgs *ArgList = MacroArgCache; ArgList;)
160     ArgList = ArgList->deallocate();
161 
162   // Release pragma information.
163   delete PragmaHandlers;
164 
165   // Delete the scratch buffer info.
166   delete ScratchBuf;
167 
168   // Delete the header search info, if we own it.
169   if (OwnsHeaderSearch)
170     delete &HeaderInfo;
171 
172   delete Callbacks;
173 }
174 
Initialize(const TargetInfo & Target)175 void Preprocessor::Initialize(const TargetInfo &Target) {
176   assert((!this->Target || this->Target == &Target) &&
177          "Invalid override of target information");
178   this->Target = &Target;
179 
180   // Initialize information about built-ins.
181   BuiltinInfo.InitializeTarget(Target);
182   HeaderInfo.setTarget(Target);
183 }
184 
setPTHManager(PTHManager * pm)185 void Preprocessor::setPTHManager(PTHManager* pm) {
186   PTH.reset(pm);
187   FileMgr.addStatCache(PTH->createStatCache());
188 }
189 
DumpToken(const Token & Tok,bool DumpFlags) const190 void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const {
191   llvm::errs() << tok::getTokenName(Tok.getKind()) << " '"
192                << getSpelling(Tok) << "'";
193 
194   if (!DumpFlags) return;
195 
196   llvm::errs() << "\t";
197   if (Tok.isAtStartOfLine())
198     llvm::errs() << " [StartOfLine]";
199   if (Tok.hasLeadingSpace())
200     llvm::errs() << " [LeadingSpace]";
201   if (Tok.isExpandDisabled())
202     llvm::errs() << " [ExpandDisabled]";
203   if (Tok.needsCleaning()) {
204     const char *Start = SourceMgr.getCharacterData(Tok.getLocation());
205     llvm::errs() << " [UnClean='" << StringRef(Start, Tok.getLength())
206                  << "']";
207   }
208 
209   llvm::errs() << "\tLoc=<";
210   DumpLocation(Tok.getLocation());
211   llvm::errs() << ">";
212 }
213 
DumpLocation(SourceLocation Loc) const214 void Preprocessor::DumpLocation(SourceLocation Loc) const {
215   Loc.dump(SourceMgr);
216 }
217 
DumpMacro(const MacroInfo & MI) const218 void Preprocessor::DumpMacro(const MacroInfo &MI) const {
219   llvm::errs() << "MACRO: ";
220   for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) {
221     DumpToken(MI.getReplacementToken(i));
222     llvm::errs() << "  ";
223   }
224   llvm::errs() << "\n";
225 }
226 
PrintStats()227 void Preprocessor::PrintStats() {
228   llvm::errs() << "\n*** Preprocessor Stats:\n";
229   llvm::errs() << NumDirectives << " directives found:\n";
230   llvm::errs() << "  " << NumDefined << " #define.\n";
231   llvm::errs() << "  " << NumUndefined << " #undef.\n";
232   llvm::errs() << "  #include/#include_next/#import:\n";
233   llvm::errs() << "    " << NumEnteredSourceFiles << " source files entered.\n";
234   llvm::errs() << "    " << MaxIncludeStackDepth << " max include stack depth\n";
235   llvm::errs() << "  " << NumIf << " #if/#ifndef/#ifdef.\n";
236   llvm::errs() << "  " << NumElse << " #else/#elif.\n";
237   llvm::errs() << "  " << NumEndif << " #endif.\n";
238   llvm::errs() << "  " << NumPragma << " #pragma.\n";
239   llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n";
240 
241   llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/"
242              << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, "
243              << NumFastMacroExpanded << " on the fast path.\n";
244   llvm::errs() << (NumFastTokenPaste+NumTokenPaste)
245              << " token paste (##) operations performed, "
246              << NumFastTokenPaste << " on the fast path.\n";
247 
248   llvm::errs() << "\nPreprocessor Memory: " << getTotalMemory() << "B total";
249 
250   llvm::errs() << "\n  BumpPtr: " << BP.getTotalMemory();
251   llvm::errs() << "\n  Macro Expanded Tokens: "
252                << llvm::capacity_in_bytes(MacroExpandedTokens);
253   llvm::errs() << "\n  Predefines Buffer: " << Predefines.capacity();
254   llvm::errs() << "\n  Macros: " << llvm::capacity_in_bytes(Macros);
255   llvm::errs() << "\n  #pragma push_macro Info: "
256                << llvm::capacity_in_bytes(PragmaPushMacroInfo);
257   llvm::errs() << "\n  Poison Reasons: "
258                << llvm::capacity_in_bytes(PoisonReasons);
259   llvm::errs() << "\n  Comment Handlers: "
260                << llvm::capacity_in_bytes(CommentHandlers) << "\n";
261 }
262 
263 Preprocessor::macro_iterator
macro_begin(bool IncludeExternalMacros) const264 Preprocessor::macro_begin(bool IncludeExternalMacros) const {
265   if (IncludeExternalMacros && ExternalSource &&
266       !ReadMacrosFromExternalSource) {
267     ReadMacrosFromExternalSource = true;
268     ExternalSource->ReadDefinedMacros();
269   }
270 
271   return Macros.begin();
272 }
273 
getTotalMemory() const274 size_t Preprocessor::getTotalMemory() const {
275   return BP.getTotalMemory()
276     + llvm::capacity_in_bytes(MacroExpandedTokens)
277     + Predefines.capacity() /* Predefines buffer. */
278     + llvm::capacity_in_bytes(Macros)
279     + llvm::capacity_in_bytes(PragmaPushMacroInfo)
280     + llvm::capacity_in_bytes(PoisonReasons)
281     + llvm::capacity_in_bytes(CommentHandlers);
282 }
283 
284 Preprocessor::macro_iterator
macro_end(bool IncludeExternalMacros) const285 Preprocessor::macro_end(bool IncludeExternalMacros) const {
286   if (IncludeExternalMacros && ExternalSource &&
287       !ReadMacrosFromExternalSource) {
288     ReadMacrosFromExternalSource = true;
289     ExternalSource->ReadDefinedMacros();
290   }
291 
292   return Macros.end();
293 }
294 
295 /// \brief Compares macro tokens with a specified token value sequence.
MacroDefinitionEquals(const MacroInfo * MI,ArrayRef<TokenValue> Tokens)296 static bool MacroDefinitionEquals(const MacroInfo *MI,
297                                   ArrayRef<TokenValue> Tokens) {
298   return Tokens.size() == MI->getNumTokens() &&
299       std::equal(Tokens.begin(), Tokens.end(), MI->tokens_begin());
300 }
301 
getLastMacroWithSpelling(SourceLocation Loc,ArrayRef<TokenValue> Tokens) const302 StringRef Preprocessor::getLastMacroWithSpelling(
303                                     SourceLocation Loc,
304                                     ArrayRef<TokenValue> Tokens) const {
305   SourceLocation BestLocation;
306   StringRef BestSpelling;
307   for (Preprocessor::macro_iterator I = macro_begin(), E = macro_end();
308        I != E; ++I) {
309     if (!I->second->getMacroInfo()->isObjectLike())
310       continue;
311     const MacroDirective::DefInfo
312       Def = I->second->findDirectiveAtLoc(Loc, SourceMgr);
313     if (!Def)
314       continue;
315     if (!MacroDefinitionEquals(Def.getMacroInfo(), Tokens))
316       continue;
317     SourceLocation Location = Def.getLocation();
318     // Choose the macro defined latest.
319     if (BestLocation.isInvalid() ||
320         (Location.isValid() &&
321          SourceMgr.isBeforeInTranslationUnit(BestLocation, Location))) {
322       BestLocation = Location;
323       BestSpelling = I->first->getName();
324     }
325   }
326   return BestSpelling;
327 }
328 
recomputeCurLexerKind()329 void Preprocessor::recomputeCurLexerKind() {
330   if (CurLexer)
331     CurLexerKind = CLK_Lexer;
332   else if (CurPTHLexer)
333     CurLexerKind = CLK_PTHLexer;
334   else if (CurTokenLexer)
335     CurLexerKind = CLK_TokenLexer;
336   else
337     CurLexerKind = CLK_CachingLexer;
338 }
339 
SetCodeCompletionPoint(const FileEntry * File,unsigned CompleteLine,unsigned CompleteColumn)340 bool Preprocessor::SetCodeCompletionPoint(const FileEntry *File,
341                                           unsigned CompleteLine,
342                                           unsigned CompleteColumn) {
343   assert(File);
344   assert(CompleteLine && CompleteColumn && "Starts from 1:1");
345   assert(!CodeCompletionFile && "Already set");
346 
347   using llvm::MemoryBuffer;
348 
349   // Load the actual file's contents.
350   bool Invalid = false;
351   const MemoryBuffer *Buffer = SourceMgr.getMemoryBufferForFile(File, &Invalid);
352   if (Invalid)
353     return true;
354 
355   // Find the byte position of the truncation point.
356   const char *Position = Buffer->getBufferStart();
357   for (unsigned Line = 1; Line < CompleteLine; ++Line) {
358     for (; *Position; ++Position) {
359       if (*Position != '\r' && *Position != '\n')
360         continue;
361 
362       // Eat \r\n or \n\r as a single line.
363       if ((Position[1] == '\r' || Position[1] == '\n') &&
364           Position[0] != Position[1])
365         ++Position;
366       ++Position;
367       break;
368     }
369   }
370 
371   Position += CompleteColumn - 1;
372 
373   // Insert '\0' at the code-completion point.
374   if (Position < Buffer->getBufferEnd()) {
375     CodeCompletionFile = File;
376     CodeCompletionOffset = Position - Buffer->getBufferStart();
377 
378     MemoryBuffer *NewBuffer =
379         MemoryBuffer::getNewUninitMemBuffer(Buffer->getBufferSize() + 1,
380                                             Buffer->getBufferIdentifier());
381     char *NewBuf = const_cast<char*>(NewBuffer->getBufferStart());
382     char *NewPos = std::copy(Buffer->getBufferStart(), Position, NewBuf);
383     *NewPos = '\0';
384     std::copy(Position, Buffer->getBufferEnd(), NewPos+1);
385     SourceMgr.overrideFileContents(File, NewBuffer);
386   }
387 
388   return false;
389 }
390 
CodeCompleteNaturalLanguage()391 void Preprocessor::CodeCompleteNaturalLanguage() {
392   if (CodeComplete)
393     CodeComplete->CodeCompleteNaturalLanguage();
394   setCodeCompletionReached();
395 }
396 
397 /// getSpelling - This method is used to get the spelling of a token into a
398 /// SmallVector. Note that the returned StringRef may not point to the
399 /// supplied buffer if a copy can be avoided.
getSpelling(const Token & Tok,SmallVectorImpl<char> & Buffer,bool * Invalid) const400 StringRef Preprocessor::getSpelling(const Token &Tok,
401                                           SmallVectorImpl<char> &Buffer,
402                                           bool *Invalid) const {
403   // NOTE: this has to be checked *before* testing for an IdentifierInfo.
404   if (Tok.isNot(tok::raw_identifier) && !Tok.hasUCN()) {
405     // Try the fast path.
406     if (const IdentifierInfo *II = Tok.getIdentifierInfo())
407       return II->getName();
408   }
409 
410   // Resize the buffer if we need to copy into it.
411   if (Tok.needsCleaning())
412     Buffer.resize(Tok.getLength());
413 
414   const char *Ptr = Buffer.data();
415   unsigned Len = getSpelling(Tok, Ptr, Invalid);
416   return StringRef(Ptr, Len);
417 }
418 
419 /// CreateString - Plop the specified string into a scratch buffer and return a
420 /// location for it.  If specified, the source location provides a source
421 /// location for the token.
CreateString(StringRef Str,Token & Tok,SourceLocation ExpansionLocStart,SourceLocation ExpansionLocEnd)422 void Preprocessor::CreateString(StringRef Str, Token &Tok,
423                                 SourceLocation ExpansionLocStart,
424                                 SourceLocation ExpansionLocEnd) {
425   Tok.setLength(Str.size());
426 
427   const char *DestPtr;
428   SourceLocation Loc = ScratchBuf->getToken(Str.data(), Str.size(), DestPtr);
429 
430   if (ExpansionLocStart.isValid())
431     Loc = SourceMgr.createExpansionLoc(Loc, ExpansionLocStart,
432                                        ExpansionLocEnd, Str.size());
433   Tok.setLocation(Loc);
434 
435   // If this is a raw identifier or a literal token, set the pointer data.
436   if (Tok.is(tok::raw_identifier))
437     Tok.setRawIdentifierData(DestPtr);
438   else if (Tok.isLiteral())
439     Tok.setLiteralData(DestPtr);
440 }
441 
getCurrentModule()442 Module *Preprocessor::getCurrentModule() {
443   if (getLangOpts().CurrentModule.empty())
444     return nullptr;
445 
446   return getHeaderSearchInfo().lookupModule(getLangOpts().CurrentModule);
447 }
448 
449 //===----------------------------------------------------------------------===//
450 // Preprocessor Initialization Methods
451 //===----------------------------------------------------------------------===//
452 
453 
454 /// EnterMainSourceFile - Enter the specified FileID as the main source file,
455 /// which implicitly adds the builtin defines etc.
EnterMainSourceFile()456 void Preprocessor::EnterMainSourceFile() {
457   // We do not allow the preprocessor to reenter the main file.  Doing so will
458   // cause FileID's to accumulate information from both runs (e.g. #line
459   // information) and predefined macros aren't guaranteed to be set properly.
460   assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!");
461   FileID MainFileID = SourceMgr.getMainFileID();
462 
463   // If MainFileID is loaded it means we loaded an AST file, no need to enter
464   // a main file.
465   if (!SourceMgr.isLoadedFileID(MainFileID)) {
466     // Enter the main file source buffer.
467     EnterSourceFile(MainFileID, nullptr, SourceLocation());
468 
469     // If we've been asked to skip bytes in the main file (e.g., as part of a
470     // precompiled preamble), do so now.
471     if (SkipMainFilePreamble.first > 0)
472       CurLexer->SkipBytes(SkipMainFilePreamble.first,
473                           SkipMainFilePreamble.second);
474 
475     // Tell the header info that the main file was entered.  If the file is later
476     // #imported, it won't be re-entered.
477     if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID))
478       HeaderInfo.IncrementIncludeCount(FE);
479   }
480 
481   // Preprocess Predefines to populate the initial preprocessor state.
482   llvm::MemoryBuffer *SB =
483     llvm::MemoryBuffer::getMemBufferCopy(Predefines, "<built-in>");
484   assert(SB && "Cannot create predefined source buffer");
485   FileID FID = SourceMgr.createFileID(SB);
486   assert(!FID.isInvalid() && "Could not create FileID for predefines?");
487   setPredefinesFileID(FID);
488 
489   // Start parsing the predefines.
490   EnterSourceFile(FID, nullptr, SourceLocation());
491 }
492 
EndSourceFile()493 void Preprocessor::EndSourceFile() {
494   // Notify the client that we reached the end of the source file.
495   if (Callbacks)
496     Callbacks->EndOfMainFile();
497 }
498 
499 //===----------------------------------------------------------------------===//
500 // Lexer Event Handling.
501 //===----------------------------------------------------------------------===//
502 
503 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the
504 /// identifier information for the token and install it into the token,
505 /// updating the token kind accordingly.
LookUpIdentifierInfo(Token & Identifier) const506 IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier) const {
507   assert(!Identifier.getRawIdentifier().empty() && "No raw identifier data!");
508 
509   // Look up this token, see if it is a macro, or if it is a language keyword.
510   IdentifierInfo *II;
511   if (!Identifier.needsCleaning() && !Identifier.hasUCN()) {
512     // No cleaning needed, just use the characters from the lexed buffer.
513     II = getIdentifierInfo(Identifier.getRawIdentifier());
514   } else {
515     // Cleaning needed, alloca a buffer, clean into it, then use the buffer.
516     SmallString<64> IdentifierBuffer;
517     StringRef CleanedStr = getSpelling(Identifier, IdentifierBuffer);
518 
519     if (Identifier.hasUCN()) {
520       SmallString<64> UCNIdentifierBuffer;
521       expandUCNs(UCNIdentifierBuffer, CleanedStr);
522       II = getIdentifierInfo(UCNIdentifierBuffer);
523     } else {
524       II = getIdentifierInfo(CleanedStr);
525     }
526   }
527 
528   // Update the token info (identifier info and appropriate token kind).
529   Identifier.setIdentifierInfo(II);
530   Identifier.setKind(II->getTokenID());
531 
532   return II;
533 }
534 
SetPoisonReason(IdentifierInfo * II,unsigned DiagID)535 void Preprocessor::SetPoisonReason(IdentifierInfo *II, unsigned DiagID) {
536   PoisonReasons[II] = DiagID;
537 }
538 
PoisonSEHIdentifiers(bool Poison)539 void Preprocessor::PoisonSEHIdentifiers(bool Poison) {
540   assert(Ident__exception_code && Ident__exception_info);
541   assert(Ident___exception_code && Ident___exception_info);
542   Ident__exception_code->setIsPoisoned(Poison);
543   Ident___exception_code->setIsPoisoned(Poison);
544   Ident_GetExceptionCode->setIsPoisoned(Poison);
545   Ident__exception_info->setIsPoisoned(Poison);
546   Ident___exception_info->setIsPoisoned(Poison);
547   Ident_GetExceptionInfo->setIsPoisoned(Poison);
548   Ident__abnormal_termination->setIsPoisoned(Poison);
549   Ident___abnormal_termination->setIsPoisoned(Poison);
550   Ident_AbnormalTermination->setIsPoisoned(Poison);
551 }
552 
HandlePoisonedIdentifier(Token & Identifier)553 void Preprocessor::HandlePoisonedIdentifier(Token & Identifier) {
554   assert(Identifier.getIdentifierInfo() &&
555          "Can't handle identifiers without identifier info!");
556   llvm::DenseMap<IdentifierInfo*,unsigned>::const_iterator it =
557     PoisonReasons.find(Identifier.getIdentifierInfo());
558   if(it == PoisonReasons.end())
559     Diag(Identifier, diag::err_pp_used_poisoned_id);
560   else
561     Diag(Identifier,it->second) << Identifier.getIdentifierInfo();
562 }
563 
564 /// HandleIdentifier - This callback is invoked when the lexer reads an
565 /// identifier.  This callback looks up the identifier in the map and/or
566 /// potentially macro expands it or turns it into a named token (like 'for').
567 ///
568 /// Note that callers of this method are guarded by checking the
569 /// IdentifierInfo's 'isHandleIdentifierCase' bit.  If this method changes, the
570 /// IdentifierInfo methods that compute these properties will need to change to
571 /// match.
HandleIdentifier(Token & Identifier)572 bool Preprocessor::HandleIdentifier(Token &Identifier) {
573   assert(Identifier.getIdentifierInfo() &&
574          "Can't handle identifiers without identifier info!");
575 
576   IdentifierInfo &II = *Identifier.getIdentifierInfo();
577 
578   // If the information about this identifier is out of date, update it from
579   // the external source.
580   // We have to treat __VA_ARGS__ in a special way, since it gets
581   // serialized with isPoisoned = true, but our preprocessor may have
582   // unpoisoned it if we're defining a C99 macro.
583   if (II.isOutOfDate()) {
584     bool CurrentIsPoisoned = false;
585     if (&II == Ident__VA_ARGS__)
586       CurrentIsPoisoned = Ident__VA_ARGS__->isPoisoned();
587 
588     ExternalSource->updateOutOfDateIdentifier(II);
589     Identifier.setKind(II.getTokenID());
590 
591     if (&II == Ident__VA_ARGS__)
592       II.setIsPoisoned(CurrentIsPoisoned);
593   }
594 
595   // If this identifier was poisoned, and if it was not produced from a macro
596   // expansion, emit an error.
597   if (II.isPoisoned() && CurPPLexer) {
598     HandlePoisonedIdentifier(Identifier);
599   }
600 
601   // If this is a macro to be expanded, do it.
602   if (MacroDirective *MD = getMacroDirective(&II)) {
603     MacroInfo *MI = MD->getMacroInfo();
604     if (!DisableMacroExpansion) {
605       if (!Identifier.isExpandDisabled() && MI->isEnabled()) {
606         // C99 6.10.3p10: If the preprocessing token immediately after the
607         // macro name isn't a '(', this macro should not be expanded.
608         if (!MI->isFunctionLike() || isNextPPTokenLParen())
609           return HandleMacroExpandedIdentifier(Identifier, MD);
610       } else {
611         // C99 6.10.3.4p2 says that a disabled macro may never again be
612         // expanded, even if it's in a context where it could be expanded in the
613         // future.
614         Identifier.setFlag(Token::DisableExpand);
615         if (MI->isObjectLike() || isNextPPTokenLParen())
616           Diag(Identifier, diag::pp_disabled_macro_expansion);
617       }
618     }
619   }
620 
621   // If this identifier is a keyword in C++11, produce a warning. Don't warn if
622   // we're not considering macro expansion, since this identifier might be the
623   // name of a macro.
624   // FIXME: This warning is disabled in cases where it shouldn't be, like
625   //   "#define constexpr constexpr", "int constexpr;"
626   if (II.isCXX11CompatKeyword() && !DisableMacroExpansion) {
627     Diag(Identifier, diag::warn_cxx11_keyword) << II.getName();
628     // Don't diagnose this keyword again in this translation unit.
629     II.setIsCXX11CompatKeyword(false);
630   }
631 
632   // C++ 2.11p2: If this is an alternative representation of a C++ operator,
633   // then we act as if it is the actual operator and not the textual
634   // representation of it.
635   if (II.isCPlusPlusOperatorKeyword())
636     Identifier.setIdentifierInfo(nullptr);
637 
638   // If this is an extension token, diagnose its use.
639   // We avoid diagnosing tokens that originate from macro definitions.
640   // FIXME: This warning is disabled in cases where it shouldn't be,
641   // like "#define TY typeof", "TY(1) x".
642   if (II.isExtensionToken() && !DisableMacroExpansion)
643     Diag(Identifier, diag::ext_token_used);
644 
645   // If this is the 'import' contextual keyword following an '@', note
646   // that the next token indicates a module name.
647   //
648   // Note that we do not treat 'import' as a contextual
649   // keyword when we're in a caching lexer, because caching lexers only get
650   // used in contexts where import declarations are disallowed.
651   if (LastTokenWasAt && II.isModulesImport() && !InMacroArgs &&
652       !DisableMacroExpansion && getLangOpts().Modules &&
653       CurLexerKind != CLK_CachingLexer) {
654     ModuleImportLoc = Identifier.getLocation();
655     ModuleImportPath.clear();
656     ModuleImportExpectsIdentifier = true;
657     CurLexerKind = CLK_LexAfterModuleImport;
658   }
659   return true;
660 }
661 
Lex(Token & Result)662 void Preprocessor::Lex(Token &Result) {
663   // We loop here until a lex function retuns a token; this avoids recursion.
664   bool ReturnedToken;
665   do {
666     switch (CurLexerKind) {
667     case CLK_Lexer:
668       ReturnedToken = CurLexer->Lex(Result);
669       break;
670     case CLK_PTHLexer:
671       ReturnedToken = CurPTHLexer->Lex(Result);
672       break;
673     case CLK_TokenLexer:
674       ReturnedToken = CurTokenLexer->Lex(Result);
675       break;
676     case CLK_CachingLexer:
677       CachingLex(Result);
678       ReturnedToken = true;
679       break;
680     case CLK_LexAfterModuleImport:
681       LexAfterModuleImport(Result);
682       ReturnedToken = true;
683       break;
684     }
685   } while (!ReturnedToken);
686 
687   LastTokenWasAt = Result.is(tok::at);
688 }
689 
690 
691 /// \brief Lex a token following the 'import' contextual keyword.
692 ///
LexAfterModuleImport(Token & Result)693 void Preprocessor::LexAfterModuleImport(Token &Result) {
694   // Figure out what kind of lexer we actually have.
695   recomputeCurLexerKind();
696 
697   // Lex the next token.
698   Lex(Result);
699 
700   // The token sequence
701   //
702   //   import identifier (. identifier)*
703   //
704   // indicates a module import directive. We already saw the 'import'
705   // contextual keyword, so now we're looking for the identifiers.
706   if (ModuleImportExpectsIdentifier && Result.getKind() == tok::identifier) {
707     // We expected to see an identifier here, and we did; continue handling
708     // identifiers.
709     ModuleImportPath.push_back(std::make_pair(Result.getIdentifierInfo(),
710                                               Result.getLocation()));
711     ModuleImportExpectsIdentifier = false;
712     CurLexerKind = CLK_LexAfterModuleImport;
713     return;
714   }
715 
716   // If we're expecting a '.' or a ';', and we got a '.', then wait until we
717   // see the next identifier.
718   if (!ModuleImportExpectsIdentifier && Result.getKind() == tok::period) {
719     ModuleImportExpectsIdentifier = true;
720     CurLexerKind = CLK_LexAfterModuleImport;
721     return;
722   }
723 
724   // If we have a non-empty module path, load the named module.
725   if (!ModuleImportPath.empty() && getLangOpts().Modules) {
726     Module *Imported = TheModuleLoader.loadModule(ModuleImportLoc,
727                                                   ModuleImportPath,
728                                                   Module::MacrosVisible,
729                                                   /*IsIncludeDirective=*/false);
730     if (Callbacks)
731       Callbacks->moduleImport(ModuleImportLoc, ModuleImportPath, Imported);
732   }
733 }
734 
FinishLexStringLiteral(Token & Result,std::string & String,const char * DiagnosticTag,bool AllowMacroExpansion)735 bool Preprocessor::FinishLexStringLiteral(Token &Result, std::string &String,
736                                           const char *DiagnosticTag,
737                                           bool AllowMacroExpansion) {
738   // We need at least one string literal.
739   if (Result.isNot(tok::string_literal)) {
740     Diag(Result, diag::err_expected_string_literal)
741       << /*Source='in...'*/0 << DiagnosticTag;
742     return false;
743   }
744 
745   // Lex string literal tokens, optionally with macro expansion.
746   SmallVector<Token, 4> StrToks;
747   do {
748     StrToks.push_back(Result);
749 
750     if (Result.hasUDSuffix())
751       Diag(Result, diag::err_invalid_string_udl);
752 
753     if (AllowMacroExpansion)
754       Lex(Result);
755     else
756       LexUnexpandedToken(Result);
757   } while (Result.is(tok::string_literal));
758 
759   // Concatenate and parse the strings.
760   StringLiteralParser Literal(StrToks, *this);
761   assert(Literal.isAscii() && "Didn't allow wide strings in");
762 
763   if (Literal.hadError)
764     return false;
765 
766   if (Literal.Pascal) {
767     Diag(StrToks[0].getLocation(), diag::err_expected_string_literal)
768       << /*Source='in...'*/0 << DiagnosticTag;
769     return false;
770   }
771 
772   String = Literal.GetString();
773   return true;
774 }
775 
parseSimpleIntegerLiteral(Token & Tok,uint64_t & Value)776 bool Preprocessor::parseSimpleIntegerLiteral(Token &Tok, uint64_t &Value) {
777   assert(Tok.is(tok::numeric_constant));
778   SmallString<8> IntegerBuffer;
779   bool NumberInvalid = false;
780   StringRef Spelling = getSpelling(Tok, IntegerBuffer, &NumberInvalid);
781   if (NumberInvalid)
782     return false;
783   NumericLiteralParser Literal(Spelling, Tok.getLocation(), *this);
784   if (Literal.hadError || !Literal.isIntegerLiteral() || Literal.hasUDSuffix())
785     return false;
786   llvm::APInt APVal(64, 0);
787   if (Literal.GetIntegerValue(APVal))
788     return false;
789   Lex(Tok);
790   Value = APVal.getLimitedValue();
791   return true;
792 }
793 
addCommentHandler(CommentHandler * Handler)794 void Preprocessor::addCommentHandler(CommentHandler *Handler) {
795   assert(Handler && "NULL comment handler");
796   assert(std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler) ==
797          CommentHandlers.end() && "Comment handler already registered");
798   CommentHandlers.push_back(Handler);
799 }
800 
removeCommentHandler(CommentHandler * Handler)801 void Preprocessor::removeCommentHandler(CommentHandler *Handler) {
802   std::vector<CommentHandler *>::iterator Pos
803   = std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler);
804   assert(Pos != CommentHandlers.end() && "Comment handler not registered");
805   CommentHandlers.erase(Pos);
806 }
807 
HandleComment(Token & result,SourceRange Comment)808 bool Preprocessor::HandleComment(Token &result, SourceRange Comment) {
809   bool AnyPendingTokens = false;
810   for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(),
811        HEnd = CommentHandlers.end();
812        H != HEnd; ++H) {
813     if ((*H)->HandleComment(*this, Comment))
814       AnyPendingTokens = true;
815   }
816   if (!AnyPendingTokens || getCommentRetentionState())
817     return false;
818   Lex(result);
819   return true;
820 }
821 
~ModuleLoader()822 ModuleLoader::~ModuleLoader() { }
823 
~CommentHandler()824 CommentHandler::~CommentHandler() { }
825 
~CodeCompletionHandler()826 CodeCompletionHandler::~CodeCompletionHandler() { }
827 
createPreprocessingRecord()828 void Preprocessor::createPreprocessingRecord() {
829   if (Record)
830     return;
831 
832   Record = new PreprocessingRecord(getSourceManager());
833   addPPCallbacks(Record);
834 }
835