1 //===--- Preprocessor.h - C Language Family Preprocessor --------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
11 /// \brief Defines the clang::Preprocessor interface.
13 //===----------------------------------------------------------------------===//
15 #ifndef LLVM_CLANG_LEX_PREPROCESSOR_H
16 #define LLVM_CLANG_LEX_PREPROCESSOR_H
18 #include "clang/Basic/Builtins.h"
19 #include "clang/Basic/Diagnostic.h"
20 #include "clang/Basic/IdentifierTable.h"
21 #include "clang/Basic/SourceLocation.h"
22 #include "clang/Lex/Lexer.h"
23 #include "clang/Lex/MacroInfo.h"
24 #include "clang/Lex/ModuleMap.h"
25 #include "clang/Lex/PPCallbacks.h"
26 #include "clang/Lex/PTHLexer.h"
27 #include "clang/Lex/TokenLexer.h"
28 #include "llvm/ADT/ArrayRef.h"
29 #include "llvm/ADT/DenseMap.h"
30 #include "llvm/ADT/IntrusiveRefCntPtr.h"
31 #include "llvm/ADT/SmallPtrSet.h"
32 #include "llvm/ADT/SmallVector.h"
33 #include "llvm/ADT/TinyPtrVector.h"
34 #include "llvm/Support/Allocator.h"
35 #include "llvm/Support/Registry.h"
40 template<unsigned InternalLen> class SmallString;
46 class ExternalPreprocessorSource;
50 class PragmaNamespace;
56 class CodeCompletionHandler;
57 class DirectoryLookup;
58 class PreprocessingRecord;
61 class PreprocessorOptions;
63 /// \brief Stores token information for comparing actual tokens with
64 /// predefined values. Only handles simple tokens and identifiers.
70 TokenValue(tok::TokenKind Kind) : Kind(Kind), II(nullptr) {
71 assert(Kind != tok::raw_identifier && "Raw identifiers are not supported.");
72 assert(Kind != tok::identifier &&
73 "Identifiers should be created by TokenValue(IdentifierInfo *)");
74 assert(!tok::isLiteral(Kind) && "Literals are not supported.");
75 assert(!tok::isAnnotation(Kind) && "Annotations are not supported.");
77 TokenValue(IdentifierInfo *II) : Kind(tok::identifier), II(II) {}
78 bool operator==(const Token &Tok) const {
79 return Tok.getKind() == Kind &&
80 (!II || II == Tok.getIdentifierInfo());
84 /// \brief Context in which macro name is used.
86 MU_Other = 0, // other than #define or #undef
87 MU_Define = 1, // macro name specified in #define
88 MU_Undef = 2 // macro name specified in #undef
91 /// \brief Engages in a tight little dance with the lexer to efficiently
92 /// preprocess tokens.
94 /// Lexers know only about tokens within a single source file, and don't
95 /// know anything about preprocessor-level issues like the \#include stack,
96 /// token expansion, etc.
97 class Preprocessor : public RefCountedBase<Preprocessor> {
98 IntrusiveRefCntPtr<PreprocessorOptions> PPOpts;
99 DiagnosticsEngine *Diags;
100 LangOptions &LangOpts;
101 const TargetInfo *Target;
102 const TargetInfo *AuxTarget;
103 FileManager &FileMgr;
104 SourceManager &SourceMgr;
105 std::unique_ptr<ScratchBuffer> ScratchBuf;
106 HeaderSearch &HeaderInfo;
107 ModuleLoader &TheModuleLoader;
109 /// \brief External source of macros.
110 ExternalPreprocessorSource *ExternalSource;
113 /// An optional PTHManager object used for getting tokens from
114 /// a token cache rather than lexing the original source file.
115 std::unique_ptr<PTHManager> PTH;
117 /// A BumpPtrAllocator object used to quickly allocate and release
118 /// objects internal to the Preprocessor.
119 llvm::BumpPtrAllocator BP;
121 /// Identifiers for builtin macros and other builtins.
122 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__
123 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__
124 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__
125 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__
126 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__
127 IdentifierInfo *Ident__COUNTER__; // __COUNTER__
128 IdentifierInfo *Ident_Pragma, *Ident__pragma; // _Pragma, __pragma
129 IdentifierInfo *Ident__identifier; // __identifier
130 IdentifierInfo *Ident__VA_ARGS__; // __VA_ARGS__
131 IdentifierInfo *Ident__has_feature; // __has_feature
132 IdentifierInfo *Ident__has_extension; // __has_extension
133 IdentifierInfo *Ident__has_builtin; // __has_builtin
134 IdentifierInfo *Ident__has_attribute; // __has_attribute
135 IdentifierInfo *Ident__has_include; // __has_include
136 IdentifierInfo *Ident__has_include_next; // __has_include_next
137 IdentifierInfo *Ident__has_warning; // __has_warning
138 IdentifierInfo *Ident__is_identifier; // __is_identifier
139 IdentifierInfo *Ident__building_module; // __building_module
140 IdentifierInfo *Ident__MODULE__; // __MODULE__
141 IdentifierInfo *Ident__has_cpp_attribute; // __has_cpp_attribute
142 IdentifierInfo *Ident__has_declspec; // __has_declspec_attribute
144 SourceLocation DATELoc, TIMELoc;
145 unsigned CounterValue; // Next __COUNTER__ value.
148 /// \brief Maximum depth of \#includes.
149 MaxAllowedIncludeStackDepth = 200
152 // State that is set before the preprocessor begins.
153 bool KeepComments : 1;
154 bool KeepMacroComments : 1;
155 bool SuppressIncludeNotFoundError : 1;
157 // State that changes while the preprocessor runs:
158 bool InMacroArgs : 1; // True if parsing fn macro invocation args.
160 /// Whether the preprocessor owns the header search object.
161 bool OwnsHeaderSearch : 1;
163 /// True if macro expansion is disabled.
164 bool DisableMacroExpansion : 1;
166 /// Temporarily disables DisableMacroExpansion (i.e. enables expansion)
167 /// when parsing preprocessor directives.
168 bool MacroExpansionInDirectivesOverride : 1;
170 class ResetMacroExpansionHelper;
172 /// \brief Whether we have already loaded macros from the external source.
173 mutable bool ReadMacrosFromExternalSource : 1;
175 /// \brief True if pragmas are enabled.
176 bool PragmasEnabled : 1;
178 /// \brief True if the current build action is a preprocessing action.
179 bool PreprocessedOutput : 1;
181 /// \brief True if we are currently preprocessing a #if or #elif directive
182 bool ParsingIfOrElifDirective;
184 /// \brief True if we are pre-expanding macro arguments.
185 bool InMacroArgPreExpansion;
187 /// \brief Mapping/lookup information for all identifiers in
188 /// the program, including program keywords.
189 mutable IdentifierTable Identifiers;
191 /// \brief This table contains all the selectors in the program.
193 /// Unlike IdentifierTable above, this table *isn't* populated by the
194 /// preprocessor. It is declared/expanded here because its role/lifetime is
195 /// conceptually similar to the IdentifierTable. In addition, the current
196 /// control flow (in clang::ParseAST()), make it convenient to put here.
198 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to
199 /// the lifetime of the preprocessor.
200 SelectorTable Selectors;
202 /// \brief Information about builtins.
203 Builtin::Context BuiltinInfo;
205 /// \brief Tracks all of the pragmas that the client registered
206 /// with this preprocessor.
207 std::unique_ptr<PragmaNamespace> PragmaHandlers;
209 /// \brief Pragma handlers of the original source is stored here during the
210 /// parsing of a model file.
211 std::unique_ptr<PragmaNamespace> PragmaHandlersBackup;
213 /// \brief Tracks all of the comment handlers that the client registered
214 /// with this preprocessor.
215 std::vector<CommentHandler *> CommentHandlers;
217 /// \brief True if we want to ignore EOF token and continue later on (thus
218 /// avoid tearing the Lexer and etc. down).
219 bool IncrementalProcessing;
221 /// The kind of translation unit we are processing.
222 TranslationUnitKind TUKind;
224 /// \brief The code-completion handler.
225 CodeCompletionHandler *CodeComplete;
227 /// \brief The file that we're performing code-completion for, if any.
228 const FileEntry *CodeCompletionFile;
230 /// \brief The offset in file for the code-completion point.
231 unsigned CodeCompletionOffset;
233 /// \brief The location for the code-completion point. This gets instantiated
234 /// when the CodeCompletionFile gets \#include'ed for preprocessing.
235 SourceLocation CodeCompletionLoc;
237 /// \brief The start location for the file of the code-completion point.
239 /// This gets instantiated when the CodeCompletionFile gets \#include'ed
240 /// for preprocessing.
241 SourceLocation CodeCompletionFileLoc;
243 /// \brief The source location of the \c import contextual keyword we just
245 SourceLocation ModuleImportLoc;
247 /// \brief The module import path that we're currently processing.
248 SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> ModuleImportPath;
250 /// \brief Whether the last token we lexed was an '@'.
253 /// \brief Whether the module import expects an identifier next. Otherwise,
254 /// it expects a '.' or ';'.
255 bool ModuleImportExpectsIdentifier;
257 /// \brief The source location of the currently-active
258 /// \#pragma clang arc_cf_code_audited begin.
259 SourceLocation PragmaARCCFCodeAuditedLoc;
261 /// \brief The source location of the currently-active
262 /// \#pragma clang assume_nonnull begin.
263 SourceLocation PragmaAssumeNonNullLoc;
265 /// \brief True if we hit the code-completion point.
266 bool CodeCompletionReached;
268 /// \brief The code completion token containing the information
269 /// on the stem that is to be code completed.
270 IdentifierInfo *CodeCompletionII;
272 /// \brief The directory that the main file should be considered to occupy,
273 /// if it does not correspond to a real file (as happens when building a
275 const DirectoryEntry *MainFileDir;
277 /// \brief The number of bytes that we will initially skip when entering the
278 /// main file, along with a flag that indicates whether skipping this number
279 /// of bytes will place the lexer at the start of a line.
281 /// This is used when loading a precompiled preamble.
282 std::pair<int, bool> SkipMainFilePreamble;
284 /// \brief The current top of the stack that we're lexing from if
285 /// not expanding a macro and we are lexing directly from source code.
287 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null.
288 std::unique_ptr<Lexer> CurLexer;
290 /// \brief The current top of stack that we're lexing from if
291 /// not expanding from a macro and we are lexing from a PTH cache.
293 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null.
294 std::unique_ptr<PTHLexer> CurPTHLexer;
296 /// \brief The current top of the stack what we're lexing from
297 /// if not expanding a macro.
299 /// This is an alias for either CurLexer or CurPTHLexer.
300 PreprocessorLexer *CurPPLexer;
302 /// \brief Used to find the current FileEntry, if CurLexer is non-null
303 /// and if applicable.
305 /// This allows us to implement \#include_next and find directory-specific
307 const DirectoryLookup *CurDirLookup;
309 /// \brief The current macro we are expanding, if we are expanding a macro.
311 /// One of CurLexer and CurTokenLexer must be null.
312 std::unique_ptr<TokenLexer> CurTokenLexer;
314 /// \brief The kind of lexer we're currently working with.
320 CLK_LexAfterModuleImport
323 /// \brief If the current lexer is for a submodule that is being built, this
324 /// is that submodule.
325 Module *CurSubmodule;
327 /// \brief Keeps track of the stack of files currently
328 /// \#included, and macros currently being expanded from, not counting
329 /// CurLexer/CurTokenLexer.
330 struct IncludeStackInfo {
331 enum CurLexerKind CurLexerKind;
332 Module *TheSubmodule;
333 std::unique_ptr<Lexer> TheLexer;
334 std::unique_ptr<PTHLexer> ThePTHLexer;
335 PreprocessorLexer *ThePPLexer;
336 std::unique_ptr<TokenLexer> TheTokenLexer;
337 const DirectoryLookup *TheDirLookup;
339 // The following constructors are completely useless copies of the default
340 // versions, only needed to pacify MSVC.
341 IncludeStackInfo(enum CurLexerKind CurLexerKind, Module *TheSubmodule,
342 std::unique_ptr<Lexer> &&TheLexer,
343 std::unique_ptr<PTHLexer> &&ThePTHLexer,
344 PreprocessorLexer *ThePPLexer,
345 std::unique_ptr<TokenLexer> &&TheTokenLexer,
346 const DirectoryLookup *TheDirLookup)
347 : CurLexerKind(std::move(CurLexerKind)),
348 TheSubmodule(std::move(TheSubmodule)), TheLexer(std::move(TheLexer)),
349 ThePTHLexer(std::move(ThePTHLexer)),
350 ThePPLexer(std::move(ThePPLexer)),
351 TheTokenLexer(std::move(TheTokenLexer)),
352 TheDirLookup(std::move(TheDirLookup)) {}
354 std::vector<IncludeStackInfo> IncludeMacroStack;
356 /// \brief Actions invoked when some preprocessor activity is
357 /// encountered (e.g. a file is \#included, etc).
358 std::unique_ptr<PPCallbacks> Callbacks;
360 struct MacroExpandsInfo {
364 MacroExpandsInfo(Token Tok, MacroDefinition MD, SourceRange Range)
365 : Tok(Tok), MD(MD), Range(Range) { }
367 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks;
369 /// Information about a name that has been used to define a module macro.
370 struct ModuleMacroInfo {
371 ModuleMacroInfo(MacroDirective *MD)
372 : MD(MD), ActiveModuleMacrosGeneration(0), IsAmbiguous(false) {}
374 /// The most recent macro directive for this identifier.
376 /// The active module macros for this identifier.
377 llvm::TinyPtrVector<ModuleMacro*> ActiveModuleMacros;
378 /// The generation number at which we last updated ActiveModuleMacros.
379 /// \see Preprocessor::VisibleModules.
380 unsigned ActiveModuleMacrosGeneration;
381 /// Whether this macro name is ambiguous.
383 /// The module macros that are overridden by this macro.
384 llvm::TinyPtrVector<ModuleMacro*> OverriddenMacros;
387 /// The state of a macro for an identifier.
389 mutable llvm::PointerUnion<MacroDirective *, ModuleMacroInfo *> State;
391 ModuleMacroInfo *getModuleInfo(Preprocessor &PP,
392 const IdentifierInfo *II) const {
393 if (II->isOutOfDate())
394 PP.updateOutOfDateIdentifier(const_cast<IdentifierInfo&>(*II));
395 // FIXME: Find a spare bit on IdentifierInfo and store a
396 // HasModuleMacros flag.
397 if (!II->hasMacroDefinition() ||
398 (!PP.getLangOpts().Modules &&
399 !PP.getLangOpts().ModulesLocalVisibility) ||
400 !PP.CurSubmoduleState->VisibleModules.getGeneration())
403 auto *Info = State.dyn_cast<ModuleMacroInfo*>();
405 Info = new (PP.getPreprocessorAllocator())
406 ModuleMacroInfo(State.get<MacroDirective *>());
410 if (PP.CurSubmoduleState->VisibleModules.getGeneration() !=
411 Info->ActiveModuleMacrosGeneration)
412 PP.updateModuleMacroInfo(II, *Info);
417 MacroState() : MacroState(nullptr) {}
418 MacroState(MacroDirective *MD) : State(MD) {}
419 MacroState(MacroState &&O) noexcept : State(O.State) {
420 O.State = (MacroDirective *)nullptr;
422 MacroState &operator=(MacroState &&O) noexcept {
424 O.State = (MacroDirective *)nullptr;
429 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
430 Info->~ModuleMacroInfo();
433 MacroDirective *getLatest() const {
434 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
436 return State.get<MacroDirective*>();
438 void setLatest(MacroDirective *MD) {
439 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
445 bool isAmbiguous(Preprocessor &PP, const IdentifierInfo *II) const {
446 auto *Info = getModuleInfo(PP, II);
447 return Info ? Info->IsAmbiguous : false;
449 ArrayRef<ModuleMacro *>
450 getActiveModuleMacros(Preprocessor &PP, const IdentifierInfo *II) const {
451 if (auto *Info = getModuleInfo(PP, II))
452 return Info->ActiveModuleMacros;
456 MacroDirective::DefInfo findDirectiveAtLoc(SourceLocation Loc,
457 SourceManager &SourceMgr) const {
458 // FIXME: Incorporate module macros into the result of this.
459 if (auto *Latest = getLatest())
460 return Latest->findDirectiveAtLoc(Loc, SourceMgr);
461 return MacroDirective::DefInfo();
464 void overrideActiveModuleMacros(Preprocessor &PP, IdentifierInfo *II) {
465 if (auto *Info = getModuleInfo(PP, II)) {
466 Info->OverriddenMacros.insert(Info->OverriddenMacros.end(),
467 Info->ActiveModuleMacros.begin(),
468 Info->ActiveModuleMacros.end());
469 Info->ActiveModuleMacros.clear();
470 Info->IsAmbiguous = false;
473 ArrayRef<ModuleMacro*> getOverriddenMacros() const {
474 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
475 return Info->OverriddenMacros;
478 void setOverriddenMacros(Preprocessor &PP,
479 ArrayRef<ModuleMacro *> Overrides) {
480 auto *Info = State.dyn_cast<ModuleMacroInfo*>();
482 if (Overrides.empty())
484 Info = new (PP.getPreprocessorAllocator())
485 ModuleMacroInfo(State.get<MacroDirective *>());
488 Info->OverriddenMacros.clear();
489 Info->OverriddenMacros.insert(Info->OverriddenMacros.end(),
490 Overrides.begin(), Overrides.end());
491 Info->ActiveModuleMacrosGeneration = 0;
495 /// For each IdentifierInfo that was associated with a macro, we
496 /// keep a mapping to the history of all macro definitions and #undefs in
497 /// the reverse order (the latest one is in the head of the list).
499 /// This mapping lives within the \p CurSubmoduleState.
500 typedef llvm::DenseMap<const IdentifierInfo *, MacroState> MacroMap;
502 friend class ASTReader;
504 struct SubmoduleState;
506 /// \brief Information about a submodule that we're currently building.
507 struct BuildingSubmoduleInfo {
508 BuildingSubmoduleInfo(Module *M, SourceLocation ImportLoc,
509 SubmoduleState *OuterSubmoduleState,
510 unsigned OuterPendingModuleMacroNames)
511 : M(M), ImportLoc(ImportLoc), OuterSubmoduleState(OuterSubmoduleState),
512 OuterPendingModuleMacroNames(OuterPendingModuleMacroNames) {}
514 /// The module that we are building.
516 /// The location at which the module was included.
517 SourceLocation ImportLoc;
518 /// The previous SubmoduleState.
519 SubmoduleState *OuterSubmoduleState;
520 /// The number of pending module macro names when we started building this.
521 unsigned OuterPendingModuleMacroNames;
523 SmallVector<BuildingSubmoduleInfo, 8> BuildingSubmoduleStack;
525 /// \brief Information about a submodule's preprocessor state.
526 struct SubmoduleState {
527 /// The macros for the submodule.
529 /// The set of modules that are visible within the submodule.
530 VisibleModuleSet VisibleModules;
531 // FIXME: CounterValue?
532 // FIXME: PragmaPushMacroInfo?
534 std::map<Module*, SubmoduleState> Submodules;
536 /// The preprocessor state for preprocessing outside of any submodule.
537 SubmoduleState NullSubmoduleState;
539 /// The current submodule state. Will be \p NullSubmoduleState if we're not
541 SubmoduleState *CurSubmoduleState;
543 /// The set of known macros exported from modules.
544 llvm::FoldingSet<ModuleMacro> ModuleMacros;
546 /// The names of potential module macros that we've not yet processed.
547 llvm::SmallVector<const IdentifierInfo*, 32> PendingModuleMacroNames;
549 /// The list of module macros, for each identifier, that are not overridden by
550 /// any other module macro.
551 llvm::DenseMap<const IdentifierInfo *, llvm::TinyPtrVector<ModuleMacro*>>
554 /// \brief Macros that we want to warn because they are not used at the end
555 /// of the translation unit.
557 /// We store just their SourceLocations instead of
558 /// something like MacroInfo*. The benefit of this is that when we are
559 /// deserializing from PCH, we don't need to deserialize identifier & macros
560 /// just so that we can report that they are unused, we just warn using
561 /// the SourceLocations of this set (that will be filled by the ASTReader).
562 /// We are using SmallPtrSet instead of a vector for faster removal.
563 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy;
564 WarnUnusedMacroLocsTy WarnUnusedMacroLocs;
566 /// \brief A "freelist" of MacroArg objects that can be
567 /// reused for quick allocation.
568 MacroArgs *MacroArgCache;
569 friend class MacroArgs;
571 /// For each IdentifierInfo used in a \#pragma push_macro directive,
572 /// we keep a MacroInfo stack used to restore the previous macro value.
573 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo;
575 // Various statistics we track for performance analysis.
576 unsigned NumDirectives, NumDefined, NumUndefined, NumPragma;
577 unsigned NumIf, NumElse, NumEndif;
578 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth;
579 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded;
580 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste;
583 /// \brief The predefined macros that preprocessor should use from the
584 /// command line etc.
585 std::string Predefines;
587 /// \brief The file ID for the preprocessor predefines.
588 FileID PredefinesFileID;
591 /// \brief Cache of macro expanders to reduce malloc traffic.
592 enum { TokenLexerCacheSize = 8 };
593 unsigned NumCachedTokenLexers;
594 std::unique_ptr<TokenLexer> TokenLexerCache[TokenLexerCacheSize];
597 /// \brief Keeps macro expanded tokens for TokenLexers.
599 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
600 /// going to lex in the cache and when it finishes the tokens are removed
601 /// from the end of the cache.
602 SmallVector<Token, 16> MacroExpandedTokens;
603 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack;
605 /// \brief A record of the macro definitions and expansions that
606 /// occurred during preprocessing.
608 /// This is an optional side structure that can be enabled with
609 /// \c createPreprocessingRecord() prior to preprocessing.
610 PreprocessingRecord *Record;
612 /// Cached tokens state.
613 typedef SmallVector<Token, 1> CachedTokensTy;
615 /// \brief Cached tokens are stored here when we do backtracking or
616 /// lookahead. They are "lexed" by the CachingLex() method.
617 CachedTokensTy CachedTokens;
619 /// \brief The position of the cached token that CachingLex() should
622 /// If it points beyond the CachedTokens vector, it means that a normal
623 /// Lex() should be invoked.
624 CachedTokensTy::size_type CachedLexPos;
626 /// \brief Stack of backtrack positions, allowing nested backtracks.
628 /// The EnableBacktrackAtThisPos() method pushes a position to
629 /// indicate where CachedLexPos should be set when the BackTrack() method is
630 /// invoked (at which point the last position is popped).
631 std::vector<CachedTokensTy::size_type> BacktrackPositions;
633 struct MacroInfoChain {
635 MacroInfoChain *Next;
638 /// MacroInfos are managed as a chain for easy disposal. This is the head
640 MacroInfoChain *MIChainHead;
642 struct DeserializedMacroInfoChain {
644 unsigned OwningModuleID; // MUST be immediately after the MacroInfo object
645 // so it can be accessed by MacroInfo::getOwningModuleID().
646 DeserializedMacroInfoChain *Next;
648 DeserializedMacroInfoChain *DeserialMIChainHead;
650 void updateOutOfDateIdentifier(IdentifierInfo &II) const;
653 Preprocessor(IntrusiveRefCntPtr<PreprocessorOptions> PPOpts,
654 DiagnosticsEngine &diags, LangOptions &opts,
655 SourceManager &SM, HeaderSearch &Headers,
656 ModuleLoader &TheModuleLoader,
657 IdentifierInfoLookup *IILookup = nullptr,
658 bool OwnsHeaderSearch = false,
659 TranslationUnitKind TUKind = TU_Complete);
663 /// \brief Initialize the preprocessor using information about the target.
665 /// \param Target is owned by the caller and must remain valid for the
666 /// lifetime of the preprocessor.
667 /// \param AuxTarget is owned by the caller and must remain valid for
668 /// the lifetime of the preprocessor.
669 void Initialize(const TargetInfo &Target,
670 const TargetInfo *AuxTarget = nullptr);
672 /// \brief Initialize the preprocessor to parse a model file
674 /// To parse model files the preprocessor of the original source is reused to
675 /// preserver the identifier table. However to avoid some duplicate
676 /// information in the preprocessor some cleanup is needed before it is used
677 /// to parse model files. This method does that cleanup.
678 void InitializeForModelFile();
680 /// \brief Cleanup after model file parsing
681 void FinalizeForModelFile();
683 /// \brief Retrieve the preprocessor options used to initialize this
685 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; }
687 DiagnosticsEngine &getDiagnostics() const { return *Diags; }
688 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; }
690 const LangOptions &getLangOpts() const { return LangOpts; }
691 const TargetInfo &getTargetInfo() const { return *Target; }
692 const TargetInfo *getAuxTargetInfo() const { return AuxTarget; }
693 FileManager &getFileManager() const { return FileMgr; }
694 SourceManager &getSourceManager() const { return SourceMgr; }
695 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; }
697 IdentifierTable &getIdentifierTable() { return Identifiers; }
698 const IdentifierTable &getIdentifierTable() const { return Identifiers; }
699 SelectorTable &getSelectorTable() { return Selectors; }
700 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; }
701 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; }
703 void setPTHManager(PTHManager* pm);
705 PTHManager *getPTHManager() { return PTH.get(); }
707 void setExternalSource(ExternalPreprocessorSource *Source) {
708 ExternalSource = Source;
711 ExternalPreprocessorSource *getExternalSource() const {
712 return ExternalSource;
715 /// \brief Retrieve the module loader associated with this preprocessor.
716 ModuleLoader &getModuleLoader() const { return TheModuleLoader; }
718 bool hadModuleLoaderFatalFailure() const {
719 return TheModuleLoader.HadFatalFailure;
722 /// \brief True if we are currently preprocessing a #if or #elif directive
723 bool isParsingIfOrElifDirective() const {
724 return ParsingIfOrElifDirective;
727 /// \brief Control whether the preprocessor retains comments in output.
728 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) {
729 this->KeepComments = KeepComments | KeepMacroComments;
730 this->KeepMacroComments = KeepMacroComments;
733 bool getCommentRetentionState() const { return KeepComments; }
735 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; }
736 bool getPragmasEnabled() const { return PragmasEnabled; }
738 void SetSuppressIncludeNotFoundError(bool Suppress) {
739 SuppressIncludeNotFoundError = Suppress;
742 bool GetSuppressIncludeNotFoundError() {
743 return SuppressIncludeNotFoundError;
746 /// Sets whether the preprocessor is responsible for producing output or if
747 /// it is producing tokens to be consumed by Parse and Sema.
748 void setPreprocessedOutput(bool IsPreprocessedOutput) {
749 PreprocessedOutput = IsPreprocessedOutput;
752 /// Returns true if the preprocessor is responsible for generating output,
753 /// false if it is producing tokens to be consumed by Parse and Sema.
754 bool isPreprocessedOutput() const { return PreprocessedOutput; }
756 /// \brief Return true if we are lexing directly from the specified lexer.
757 bool isCurrentLexer(const PreprocessorLexer *L) const {
758 return CurPPLexer == L;
761 /// \brief Return the current lexer being lexed from.
763 /// Note that this ignores any potentially active macro expansions and _Pragma
764 /// expansions going on at the time.
765 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; }
767 /// \brief Return the current file lexer being lexed from.
769 /// Note that this ignores any potentially active macro expansions and _Pragma
770 /// expansions going on at the time.
771 PreprocessorLexer *getCurrentFileLexer() const;
773 /// \brief Return the submodule owning the file being lexed.
774 Module *getCurrentSubmodule() const { return CurSubmodule; }
776 /// \brief Returns the FileID for the preprocessor predefines.
777 FileID getPredefinesFileID() const { return PredefinesFileID; }
780 /// \brief Accessors for preprocessor callbacks.
782 /// Note that this class takes ownership of any PPCallbacks object given to
784 PPCallbacks *getPPCallbacks() const { return Callbacks.get(); }
785 void addPPCallbacks(std::unique_ptr<PPCallbacks> C) {
787 C = llvm::make_unique<PPChainedCallbacks>(std::move(C),
788 std::move(Callbacks));
789 Callbacks = std::move(C);
793 bool isMacroDefined(StringRef Id) {
794 return isMacroDefined(&Identifiers.get(Id));
796 bool isMacroDefined(const IdentifierInfo *II) {
797 return II->hasMacroDefinition() &&
798 (!getLangOpts().Modules || (bool)getMacroDefinition(II));
801 /// \brief Determine whether II is defined as a macro within the module M,
802 /// if that is a module that we've already preprocessed. Does not check for
803 /// macros imported into M.
804 bool isMacroDefinedInLocalModule(const IdentifierInfo *II, Module *M) {
805 if (!II->hasMacroDefinition())
807 auto I = Submodules.find(M);
808 if (I == Submodules.end())
810 auto J = I->second.Macros.find(II);
811 if (J == I->second.Macros.end())
813 auto *MD = J->second.getLatest();
814 return MD && MD->isDefined();
817 MacroDefinition getMacroDefinition(const IdentifierInfo *II) {
818 if (!II->hasMacroDefinition())
819 return MacroDefinition();
821 MacroState &S = CurSubmoduleState->Macros[II];
822 auto *MD = S.getLatest();
823 while (MD && isa<VisibilityMacroDirective>(MD))
824 MD = MD->getPrevious();
825 return MacroDefinition(dyn_cast_or_null<DefMacroDirective>(MD),
826 S.getActiveModuleMacros(*this, II),
827 S.isAmbiguous(*this, II));
830 MacroDefinition getMacroDefinitionAtLoc(const IdentifierInfo *II,
831 SourceLocation Loc) {
832 if (!II->hadMacroDefinition())
833 return MacroDefinition();
835 MacroState &S = CurSubmoduleState->Macros[II];
836 MacroDirective::DefInfo DI;
837 if (auto *MD = S.getLatest())
838 DI = MD->findDirectiveAtLoc(Loc, getSourceManager());
839 // FIXME: Compute the set of active module macros at the specified location.
840 return MacroDefinition(DI.getDirective(),
841 S.getActiveModuleMacros(*this, II),
842 S.isAmbiguous(*this, II));
845 /// \brief Given an identifier, return its latest non-imported MacroDirective
846 /// if it is \#define'd and not \#undef'd, or null if it isn't \#define'd.
847 MacroDirective *getLocalMacroDirective(const IdentifierInfo *II) const {
848 if (!II->hasMacroDefinition())
851 auto *MD = getLocalMacroDirectiveHistory(II);
852 if (!MD || MD->getDefinition().isUndefined())
858 const MacroInfo *getMacroInfo(const IdentifierInfo *II) const {
859 return const_cast<Preprocessor*>(this)->getMacroInfo(II);
862 MacroInfo *getMacroInfo(const IdentifierInfo *II) {
863 if (!II->hasMacroDefinition())
865 if (auto MD = getMacroDefinition(II))
866 return MD.getMacroInfo();
870 /// \brief Given an identifier, return the latest non-imported macro
871 /// directive for that identifier.
873 /// One can iterate over all previous macro directives from the most recent
875 MacroDirective *getLocalMacroDirectiveHistory(const IdentifierInfo *II) const;
877 /// \brief Add a directive to the macro directive history for this identifier.
878 void appendMacroDirective(IdentifierInfo *II, MacroDirective *MD);
879 DefMacroDirective *appendDefMacroDirective(IdentifierInfo *II, MacroInfo *MI,
880 SourceLocation Loc) {
881 DefMacroDirective *MD = AllocateDefMacroDirective(MI, Loc);
882 appendMacroDirective(II, MD);
885 DefMacroDirective *appendDefMacroDirective(IdentifierInfo *II,
887 return appendDefMacroDirective(II, MI, MI->getDefinitionLoc());
889 /// \brief Set a MacroDirective that was loaded from a PCH file.
890 void setLoadedMacroDirective(IdentifierInfo *II, MacroDirective *ED,
893 /// \brief Register an exported macro for a module and identifier.
894 ModuleMacro *addModuleMacro(Module *Mod, IdentifierInfo *II, MacroInfo *Macro,
895 ArrayRef<ModuleMacro *> Overrides, bool &IsNew);
896 ModuleMacro *getModuleMacro(Module *Mod, IdentifierInfo *II);
898 /// \brief Get the list of leaf (non-overridden) module macros for a name.
899 ArrayRef<ModuleMacro*> getLeafModuleMacros(const IdentifierInfo *II) const {
900 if (II->isOutOfDate())
901 updateOutOfDateIdentifier(const_cast<IdentifierInfo&>(*II));
902 auto I = LeafModuleMacros.find(II);
903 if (I != LeafModuleMacros.end())
909 /// Iterators for the macro history table. Currently defined macros have
910 /// IdentifierInfo::hasMacroDefinition() set and an empty
911 /// MacroInfo::getUndefLoc() at the head of the list.
912 typedef MacroMap::const_iterator macro_iterator;
913 macro_iterator macro_begin(bool IncludeExternalMacros = true) const;
914 macro_iterator macro_end(bool IncludeExternalMacros = true) const;
915 llvm::iterator_range<macro_iterator>
916 macros(bool IncludeExternalMacros = true) const {
917 return llvm::make_range(macro_begin(IncludeExternalMacros),
918 macro_end(IncludeExternalMacros));
922 /// \brief Return the name of the macro defined before \p Loc that has
923 /// spelling \p Tokens. If there are multiple macros with same spelling,
924 /// return the last one defined.
925 StringRef getLastMacroWithSpelling(SourceLocation Loc,
926 ArrayRef<TokenValue> Tokens) const;
928 const std::string &getPredefines() const { return Predefines; }
929 /// \brief Set the predefines for this Preprocessor.
931 /// These predefines are automatically injected when parsing the main file.
932 void setPredefines(const char *P) { Predefines = P; }
933 void setPredefines(StringRef P) { Predefines = P; }
935 /// Return information about the specified preprocessor
936 /// identifier token.
937 IdentifierInfo *getIdentifierInfo(StringRef Name) const {
938 return &Identifiers.get(Name);
941 /// \brief Add the specified pragma handler to this preprocessor.
943 /// If \p Namespace is non-null, then it is a token required to exist on the
944 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC".
945 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler);
946 void AddPragmaHandler(PragmaHandler *Handler) {
947 AddPragmaHandler(StringRef(), Handler);
950 /// \brief Remove the specific pragma handler from this preprocessor.
952 /// If \p Namespace is non-null, then it should be the namespace that
953 /// \p Handler was added to. It is an error to remove a handler that
954 /// has not been registered.
955 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler);
956 void RemovePragmaHandler(PragmaHandler *Handler) {
957 RemovePragmaHandler(StringRef(), Handler);
960 /// Install empty handlers for all pragmas (making them ignored).
961 void IgnorePragmas();
963 /// \brief Add the specified comment handler to the preprocessor.
964 void addCommentHandler(CommentHandler *Handler);
966 /// \brief Remove the specified comment handler.
968 /// It is an error to remove a handler that has not been registered.
969 void removeCommentHandler(CommentHandler *Handler);
971 /// \brief Set the code completion handler to the given object.
972 void setCodeCompletionHandler(CodeCompletionHandler &Handler) {
973 CodeComplete = &Handler;
976 /// \brief Retrieve the current code-completion handler.
977 CodeCompletionHandler *getCodeCompletionHandler() const {
981 /// \brief Clear out the code completion handler.
982 void clearCodeCompletionHandler() {
983 CodeComplete = nullptr;
986 /// \brief Hook used by the lexer to invoke the "natural language" code
987 /// completion point.
988 void CodeCompleteNaturalLanguage();
990 /// \brief Set the code completion token for filtering purposes.
991 void setCodeCompletionIdentifierInfo(IdentifierInfo *Filter) {
992 CodeCompletionII = Filter;
995 /// \brief Get the code completion token for filtering purposes.
996 StringRef getCodeCompletionFilter() {
997 if (CodeCompletionII)
998 return CodeCompletionII->getName();
1002 /// \brief Retrieve the preprocessing record, or NULL if there is no
1003 /// preprocessing record.
1004 PreprocessingRecord *getPreprocessingRecord() const { return Record; }
1006 /// \brief Create a new preprocessing record, which will keep track of
1007 /// all macro expansions, macro definitions, etc.
1008 void createPreprocessingRecord();
1010 /// \brief Enter the specified FileID as the main source file,
1011 /// which implicitly adds the builtin defines etc.
1012 void EnterMainSourceFile();
1014 /// \brief Inform the preprocessor callbacks that processing is complete.
1015 void EndSourceFile();
1017 /// \brief Add a source file to the top of the include stack and
1018 /// start lexing tokens from it instead of the current buffer.
1020 /// Emits a diagnostic, doesn't enter the file, and returns true on error.
1021 bool EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir,
1022 SourceLocation Loc);
1024 /// \brief Add a Macro to the top of the include stack and start lexing
1025 /// tokens from it instead of the current buffer.
1027 /// \param Args specifies the tokens input to a function-like macro.
1028 /// \param ILEnd specifies the location of the ')' for a function-like macro
1029 /// or the identifier for an object-like macro.
1030 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro,
1033 /// \brief Add a "macro" context to the top of the include stack,
1034 /// which will cause the lexer to start returning the specified tokens.
1036 /// If \p DisableMacroExpansion is true, tokens lexed from the token stream
1037 /// will not be subject to further macro expansion. Otherwise, these tokens
1038 /// will be re-macro-expanded when/if expansion is enabled.
1040 /// If \p OwnsTokens is false, this method assumes that the specified stream
1041 /// of tokens has a permanent owner somewhere, so they do not need to be
1042 /// copied. If it is true, it assumes the array of tokens is allocated with
1043 /// \c new[] and the Preprocessor will delete[] it.
1045 void EnterTokenStream(const Token *Toks, unsigned NumToks,
1046 bool DisableMacroExpansion, bool OwnsTokens);
1049 void EnterTokenStream(std::unique_ptr<Token[]> Toks, unsigned NumToks,
1050 bool DisableMacroExpansion) {
1051 EnterTokenStream(Toks.release(), NumToks, DisableMacroExpansion, true);
1053 void EnterTokenStream(ArrayRef<Token> Toks, bool DisableMacroExpansion) {
1054 EnterTokenStream(Toks.data(), Toks.size(), DisableMacroExpansion, false);
1057 /// \brief Pop the current lexer/macro exp off the top of the lexer stack.
1059 /// This should only be used in situations where the current state of the
1060 /// top-of-stack lexer is known.
1061 void RemoveTopOfLexerStack();
1063 /// From the point that this method is called, and until
1064 /// CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor
1065 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will
1066 /// make the Preprocessor re-lex the same tokens.
1068 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can
1069 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will
1070 /// be combined with the EnableBacktrackAtThisPos calls in reverse order.
1072 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack
1073 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of
1074 /// tokens will continue indefinitely.
1076 void EnableBacktrackAtThisPos();
1078 /// \brief Disable the last EnableBacktrackAtThisPos call.
1079 void CommitBacktrackedTokens();
1081 /// \brief Make Preprocessor re-lex the tokens that were lexed since
1082 /// EnableBacktrackAtThisPos() was previously called.
1085 /// \brief True if EnableBacktrackAtThisPos() was called and
1086 /// caching of tokens is on.
1087 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); }
1089 /// \brief Lex the next token for this preprocessor.
1090 void Lex(Token &Result);
1092 void LexAfterModuleImport(Token &Result);
1094 void makeModuleVisible(Module *M, SourceLocation Loc);
1096 SourceLocation getModuleImportLoc(Module *M) const {
1097 return CurSubmoduleState->VisibleModules.getImportLoc(M);
1100 /// \brief Lex a string literal, which may be the concatenation of multiple
1101 /// string literals and may even come from macro expansion.
1102 /// \returns true on success, false if a error diagnostic has been generated.
1103 bool LexStringLiteral(Token &Result, std::string &String,
1104 const char *DiagnosticTag, bool AllowMacroExpansion) {
1105 if (AllowMacroExpansion)
1108 LexUnexpandedToken(Result);
1109 return FinishLexStringLiteral(Result, String, DiagnosticTag,
1110 AllowMacroExpansion);
1113 /// \brief Complete the lexing of a string literal where the first token has
1114 /// already been lexed (see LexStringLiteral).
1115 bool FinishLexStringLiteral(Token &Result, std::string &String,
1116 const char *DiagnosticTag,
1117 bool AllowMacroExpansion);
1119 /// \brief Lex a token. If it's a comment, keep lexing until we get
1120 /// something not a comment.
1122 /// This is useful in -E -C mode where comments would foul up preprocessor
1123 /// directive handling.
1124 void LexNonComment(Token &Result) {
1127 while (Result.getKind() == tok::comment);
1130 /// \brief Just like Lex, but disables macro expansion of identifier tokens.
1131 void LexUnexpandedToken(Token &Result) {
1132 // Disable macro expansion.
1133 bool OldVal = DisableMacroExpansion;
1134 DisableMacroExpansion = true;
1139 DisableMacroExpansion = OldVal;
1142 /// \brief Like LexNonComment, but this disables macro expansion of
1143 /// identifier tokens.
1144 void LexUnexpandedNonComment(Token &Result) {
1146 LexUnexpandedToken(Result);
1147 while (Result.getKind() == tok::comment);
1150 /// \brief Parses a simple integer literal to get its numeric value. Floating
1151 /// point literals and user defined literals are rejected. Used primarily to
1152 /// handle pragmas that accept integer arguments.
1153 bool parseSimpleIntegerLiteral(Token &Tok, uint64_t &Value);
1155 /// Disables macro expansion everywhere except for preprocessor directives.
1156 void SetMacroExpansionOnlyInDirectives() {
1157 DisableMacroExpansion = true;
1158 MacroExpansionInDirectivesOverride = true;
1161 /// \brief Peeks ahead N tokens and returns that token without consuming any
1164 /// LookAhead(0) returns the next token that would be returned by Lex(),
1165 /// LookAhead(1) returns the token after it, etc. This returns normal
1166 /// tokens after phase 5. As such, it is equivalent to using
1167 /// 'Lex', not 'LexUnexpandedToken'.
1168 const Token &LookAhead(unsigned N) {
1169 if (CachedLexPos + N < CachedTokens.size())
1170 return CachedTokens[CachedLexPos+N];
1172 return PeekAhead(N+1);
1175 /// \brief When backtracking is enabled and tokens are cached,
1176 /// this allows to revert a specific number of tokens.
1178 /// Note that the number of tokens being reverted should be up to the last
1179 /// backtrack position, not more.
1180 void RevertCachedTokens(unsigned N) {
1181 assert(isBacktrackEnabled() &&
1182 "Should only be called when tokens are cached for backtracking");
1183 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back())
1184 && "Should revert tokens up to the last backtrack position, not more");
1185 assert(signed(CachedLexPos) - signed(N) >= 0 &&
1186 "Corrupted backtrack positions ?");
1190 /// \brief Enters a token in the token stream to be lexed next.
1192 /// If BackTrack() is called afterwards, the token will remain at the
1193 /// insertion point.
1194 void EnterToken(const Token &Tok) {
1195 EnterCachingLexMode();
1196 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok);
1199 /// We notify the Preprocessor that if it is caching tokens (because
1200 /// backtrack is enabled) it should replace the most recent cached tokens
1201 /// with the given annotation token. This function has no effect if
1202 /// backtracking is not enabled.
1204 /// Note that the use of this function is just for optimization, so that the
1205 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is
1207 void AnnotateCachedTokens(const Token &Tok) {
1208 assert(Tok.isAnnotation() && "Expected annotation token");
1209 if (CachedLexPos != 0 && isBacktrackEnabled())
1210 AnnotatePreviousCachedTokens(Tok);
1213 /// Get the location of the last cached token, suitable for setting the end
1214 /// location of an annotation token.
1215 SourceLocation getLastCachedTokenLocation() const {
1216 assert(CachedLexPos != 0);
1217 return CachedTokens[CachedLexPos-1].getLastLoc();
1220 /// \brief Whether \p Tok is the most recent token (`CachedLexPos - 1`) in
1222 bool IsPreviousCachedToken(const Token &Tok) const;
1224 /// \brief Replace token in `CachedLexPos - 1` in CachedTokens by the tokens
1227 /// Useful when a token needs to be split in smaller ones and CachedTokens
1228 /// most recent token must to be updated to reflect that.
1229 void ReplacePreviousCachedToken(ArrayRef<Token> NewToks);
1231 /// \brief Replace the last token with an annotation token.
1233 /// Like AnnotateCachedTokens(), this routine replaces an
1234 /// already-parsed (and resolved) token with an annotation
1235 /// token. However, this routine only replaces the last token with
1236 /// the annotation token; it does not affect any other cached
1237 /// tokens. This function has no effect if backtracking is not
1239 void ReplaceLastTokenWithAnnotation(const Token &Tok) {
1240 assert(Tok.isAnnotation() && "Expected annotation token");
1241 if (CachedLexPos != 0 && isBacktrackEnabled())
1242 CachedTokens[CachedLexPos-1] = Tok;
1245 /// Update the current token to represent the provided
1246 /// identifier, in order to cache an action performed by typo correction.
1247 void TypoCorrectToken(const Token &Tok) {
1248 assert(Tok.getIdentifierInfo() && "Expected identifier token");
1249 if (CachedLexPos != 0 && isBacktrackEnabled())
1250 CachedTokens[CachedLexPos-1] = Tok;
1253 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/
1254 /// CurTokenLexer pointers.
1255 void recomputeCurLexerKind();
1257 /// \brief Returns true if incremental processing is enabled
1258 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; }
1260 /// \brief Enables the incremental processing
1261 void enableIncrementalProcessing(bool value = true) {
1262 IncrementalProcessing = value;
1265 /// \brief Specify the point at which code-completion will be performed.
1267 /// \param File the file in which code completion should occur. If
1268 /// this file is included multiple times, code-completion will
1269 /// perform completion the first time it is included. If NULL, this
1270 /// function clears out the code-completion point.
1272 /// \param Line the line at which code completion should occur
1275 /// \param Column the column at which code completion should occur
1278 /// \returns true if an error occurred, false otherwise.
1279 bool SetCodeCompletionPoint(const FileEntry *File,
1280 unsigned Line, unsigned Column);
1282 /// \brief Determine if we are performing code completion.
1283 bool isCodeCompletionEnabled() const { return CodeCompletionFile != nullptr; }
1285 /// \brief Returns the location of the code-completion point.
1287 /// Returns an invalid location if code-completion is not enabled or the file
1288 /// containing the code-completion point has not been lexed yet.
1289 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; }
1291 /// \brief Returns the start location of the file of code-completion point.
1293 /// Returns an invalid location if code-completion is not enabled or the file
1294 /// containing the code-completion point has not been lexed yet.
1295 SourceLocation getCodeCompletionFileLoc() const {
1296 return CodeCompletionFileLoc;
1299 /// \brief Returns true if code-completion is enabled and we have hit the
1300 /// code-completion point.
1301 bool isCodeCompletionReached() const { return CodeCompletionReached; }
1303 /// \brief Note that we hit the code-completion point.
1304 void setCodeCompletionReached() {
1305 assert(isCodeCompletionEnabled() && "Code-completion not enabled!");
1306 CodeCompletionReached = true;
1307 // Silence any diagnostics that occur after we hit the code-completion.
1308 getDiagnostics().setSuppressAllDiagnostics(true);
1311 /// \brief The location of the currently-active \#pragma clang
1312 /// arc_cf_code_audited begin.
1314 /// Returns an invalid location if there is no such pragma active.
1315 SourceLocation getPragmaARCCFCodeAuditedLoc() const {
1316 return PragmaARCCFCodeAuditedLoc;
1319 /// \brief Set the location of the currently-active \#pragma clang
1320 /// arc_cf_code_audited begin. An invalid location ends the pragma.
1321 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) {
1322 PragmaARCCFCodeAuditedLoc = Loc;
1325 /// \brief The location of the currently-active \#pragma clang
1326 /// assume_nonnull begin.
1328 /// Returns an invalid location if there is no such pragma active.
1329 SourceLocation getPragmaAssumeNonNullLoc() const {
1330 return PragmaAssumeNonNullLoc;
1333 /// \brief Set the location of the currently-active \#pragma clang
1334 /// assume_nonnull begin. An invalid location ends the pragma.
1335 void setPragmaAssumeNonNullLoc(SourceLocation Loc) {
1336 PragmaAssumeNonNullLoc = Loc;
1339 /// \brief Set the directory in which the main file should be considered
1340 /// to have been found, if it is not a real file.
1341 void setMainFileDir(const DirectoryEntry *Dir) {
1345 /// \brief Instruct the preprocessor to skip part of the main source file.
1347 /// \param Bytes The number of bytes in the preamble to skip.
1349 /// \param StartOfLine Whether skipping these bytes puts the lexer at the
1350 /// start of a line.
1351 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) {
1352 SkipMainFilePreamble.first = Bytes;
1353 SkipMainFilePreamble.second = StartOfLine;
1356 /// Forwarding function for diagnostics. This emits a diagnostic at
1357 /// the specified Token's location, translating the token's start
1358 /// position in the current buffer into a SourcePosition object for rendering.
1359 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const {
1360 return Diags->Report(Loc, DiagID);
1363 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const {
1364 return Diags->Report(Tok.getLocation(), DiagID);
1367 /// Return the 'spelling' of the token at the given
1368 /// location; does not go up to the spelling location or down to the
1369 /// expansion location.
1371 /// \param buffer A buffer which will be used only if the token requires
1372 /// "cleaning", e.g. if it contains trigraphs or escaped newlines
1373 /// \param invalid If non-null, will be set \c true if an error occurs.
1374 StringRef getSpelling(SourceLocation loc,
1375 SmallVectorImpl<char> &buffer,
1376 bool *invalid = nullptr) const {
1377 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid);
1380 /// \brief Return the 'spelling' of the Tok token.
1382 /// The spelling of a token is the characters used to represent the token in
1383 /// the source file after trigraph expansion and escaped-newline folding. In
1384 /// particular, this wants to get the true, uncanonicalized, spelling of
1385 /// things like digraphs, UCNs, etc.
1387 /// \param Invalid If non-null, will be set \c true if an error occurs.
1388 std::string getSpelling(const Token &Tok, bool *Invalid = nullptr) const {
1389 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid);
1392 /// \brief Get the spelling of a token into a preallocated buffer, instead
1393 /// of as an std::string.
1395 /// The caller is required to allocate enough space for the token, which is
1396 /// guaranteed to be at least Tok.getLength() bytes long. The length of the
1397 /// actual result is returned.
1399 /// Note that this method may do two possible things: it may either fill in
1400 /// the buffer specified with characters, or it may *change the input pointer*
1401 /// to point to a constant buffer with the data already in it (avoiding a
1402 /// copy). The caller is not allowed to modify the returned buffer pointer
1403 /// if an internal buffer is returned.
1404 unsigned getSpelling(const Token &Tok, const char *&Buffer,
1405 bool *Invalid = nullptr) const {
1406 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid);
1409 /// \brief Get the spelling of a token into a SmallVector.
1411 /// Note that the returned StringRef may not point to the
1412 /// supplied buffer if a copy can be avoided.
1413 StringRef getSpelling(const Token &Tok,
1414 SmallVectorImpl<char> &Buffer,
1415 bool *Invalid = nullptr) const;
1417 /// \brief Relex the token at the specified location.
1418 /// \returns true if there was a failure, false on success.
1419 bool getRawToken(SourceLocation Loc, Token &Result,
1420 bool IgnoreWhiteSpace = false) {
1421 return Lexer::getRawToken(Loc, Result, SourceMgr, LangOpts, IgnoreWhiteSpace);
1424 /// \brief Given a Token \p Tok that is a numeric constant with length 1,
1425 /// return the character.
1427 getSpellingOfSingleCharacterNumericConstant(const Token &Tok,
1428 bool *Invalid = nullptr) const {
1429 assert(Tok.is(tok::numeric_constant) &&
1430 Tok.getLength() == 1 && "Called on unsupported token");
1431 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1");
1433 // If the token is carrying a literal data pointer, just use it.
1434 if (const char *D = Tok.getLiteralData())
1437 // Otherwise, fall back on getCharacterData, which is slower, but always
1439 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid);
1442 /// \brief Retrieve the name of the immediate macro expansion.
1444 /// This routine starts from a source location, and finds the name of the
1445 /// macro responsible for its immediate expansion. It looks through any
1446 /// intervening macro argument expansions to compute this. It returns a
1447 /// StringRef that refers to the SourceManager-owned buffer of the source
1448 /// where that macro name is spelled. Thus, the result shouldn't out-live
1449 /// the SourceManager.
1450 StringRef getImmediateMacroName(SourceLocation Loc) {
1451 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts());
1454 /// \brief Plop the specified string into a scratch buffer and set the
1455 /// specified token's location and length to it.
1457 /// If specified, the source location provides a location of the expansion
1458 /// point of the token.
1459 void CreateString(StringRef Str, Token &Tok,
1460 SourceLocation ExpansionLocStart = SourceLocation(),
1461 SourceLocation ExpansionLocEnd = SourceLocation());
1463 /// \brief Computes the source location just past the end of the
1464 /// token at this source location.
1466 /// This routine can be used to produce a source location that
1467 /// points just past the end of the token referenced by \p Loc, and
1468 /// is generally used when a diagnostic needs to point just after a
1469 /// token where it expected something different that it received. If
1470 /// the returned source location would not be meaningful (e.g., if
1471 /// it points into a macro), this routine returns an invalid
1472 /// source location.
1474 /// \param Offset an offset from the end of the token, where the source
1475 /// location should refer to. The default offset (0) produces a source
1476 /// location pointing just past the end of the token; an offset of 1 produces
1477 /// a source location pointing to the last character in the token, etc.
1478 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) {
1479 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts);
1482 /// \brief Returns true if the given MacroID location points at the first
1483 /// token of the macro expansion.
1485 /// \param MacroBegin If non-null and function returns true, it is set to
1486 /// begin location of the macro.
1487 bool isAtStartOfMacroExpansion(SourceLocation loc,
1488 SourceLocation *MacroBegin = nullptr) const {
1489 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts,
1493 /// \brief Returns true if the given MacroID location points at the last
1494 /// token of the macro expansion.
1496 /// \param MacroEnd If non-null and function returns true, it is set to
1497 /// end location of the macro.
1498 bool isAtEndOfMacroExpansion(SourceLocation loc,
1499 SourceLocation *MacroEnd = nullptr) const {
1500 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd);
1503 /// \brief Print the token to stderr, used for debugging.
1504 void DumpToken(const Token &Tok, bool DumpFlags = false) const;
1505 void DumpLocation(SourceLocation Loc) const;
1506 void DumpMacro(const MacroInfo &MI) const;
1507 void dumpMacroInfo(const IdentifierInfo *II);
1509 /// \brief Given a location that specifies the start of a
1510 /// token, return a new location that specifies a character within the token.
1511 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
1512 unsigned Char) const {
1513 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts);
1516 /// \brief Increment the counters for the number of token paste operations
1519 /// If fast was specified, this is a 'fast paste' case we handled.
1520 void IncrementPasteCounter(bool isFast) {
1522 ++NumFastTokenPaste;
1529 size_t getTotalMemory() const;
1531 /// When the macro expander pastes together a comment (/##/) in Microsoft
1532 /// mode, this method handles updating the current state, returning the
1533 /// token on the next source line.
1534 void HandleMicrosoftCommentPaste(Token &Tok);
1536 //===--------------------------------------------------------------------===//
1537 // Preprocessor callback methods. These are invoked by a lexer as various
1538 // directives and events are found.
1540 /// Given a tok::raw_identifier token, look up the
1541 /// identifier information for the token and install it into the token,
1542 /// updating the token kind accordingly.
1543 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const;
1546 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons;
1550 /// \brief Specifies the reason for poisoning an identifier.
1552 /// If that identifier is accessed while poisoned, then this reason will be
1553 /// used instead of the default "poisoned" diagnostic.
1554 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID);
1556 /// \brief Display reason for poisoned identifier.
1557 void HandlePoisonedIdentifier(Token & Tok);
1559 void MaybeHandlePoisonedIdentifier(Token & Identifier) {
1560 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) {
1561 if(II->isPoisoned()) {
1562 HandlePoisonedIdentifier(Identifier);
1568 /// Identifiers used for SEH handling in Borland. These are only
1569 /// allowed in particular circumstances
1571 IdentifierInfo *Ident__exception_code,
1572 *Ident___exception_code,
1573 *Ident_GetExceptionCode;
1574 // __except filter expression
1575 IdentifierInfo *Ident__exception_info,
1576 *Ident___exception_info,
1577 *Ident_GetExceptionInfo;
1579 IdentifierInfo *Ident__abnormal_termination,
1580 *Ident___abnormal_termination,
1581 *Ident_AbnormalTermination;
1583 const char *getCurLexerEndPos();
1586 void PoisonSEHIdentifiers(bool Poison = true); // Borland
1588 /// \brief Callback invoked when the lexer reads an identifier and has
1589 /// filled in the tokens IdentifierInfo member.
1591 /// This callback potentially macro expands it or turns it into a named
1592 /// token (like 'for').
1594 /// \returns true if we actually computed a token, false if we need to
1596 bool HandleIdentifier(Token &Identifier);
1599 /// \brief Callback invoked when the lexer hits the end of the current file.
1601 /// This either returns the EOF token and returns true, or
1602 /// pops a level off the include stack and returns false, at which point the
1603 /// client should call lex again.
1604 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false);
1606 /// \brief Callback invoked when the current TokenLexer hits the end of its
1608 bool HandleEndOfTokenLexer(Token &Result);
1610 /// \brief Callback invoked when the lexer sees a # token at the start of a
1613 /// This consumes the directive, modifies the lexer/preprocessor state, and
1614 /// advances the lexer(s) so that the next token read is the correct one.
1615 void HandleDirective(Token &Result);
1617 /// \brief Ensure that the next token is a tok::eod token.
1619 /// If not, emit a diagnostic and consume up until the eod.
1620 /// If \p EnableMacros is true, then we consider macros that expand to zero
1621 /// tokens as being ok.
1622 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false);
1624 /// \brief Read and discard all tokens remaining on the current line until
1625 /// the tok::eod token is found.
1626 void DiscardUntilEndOfDirective();
1628 /// \brief Returns true if the preprocessor has seen a use of
1629 /// __DATE__ or __TIME__ in the file so far.
1630 bool SawDateOrTime() const {
1631 return DATELoc != SourceLocation() || TIMELoc != SourceLocation();
1633 unsigned getCounterValue() const { return CounterValue; }
1634 void setCounterValue(unsigned V) { CounterValue = V; }
1636 /// \brief Retrieves the module that we're currently building, if any.
1637 Module *getCurrentModule();
1639 /// \brief Allocate a new MacroInfo object with the provided SourceLocation.
1640 MacroInfo *AllocateMacroInfo(SourceLocation L);
1642 /// \brief Allocate a new MacroInfo object loaded from an AST file.
1643 MacroInfo *AllocateDeserializedMacroInfo(SourceLocation L,
1644 unsigned SubModuleID);
1646 /// \brief Turn the specified lexer token into a fully checked and spelled
1647 /// filename, e.g. as an operand of \#include.
1649 /// The caller is expected to provide a buffer that is large enough to hold
1650 /// the spelling of the filename, but is also expected to handle the case
1651 /// when this method decides to use a different buffer.
1653 /// \returns true if the input filename was in <>'s or false if it was
1655 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename);
1657 /// \brief Given a "foo" or \<foo> reference, look up the indicated file.
1659 /// Returns null on failure. \p isAngled indicates whether the file
1660 /// reference is for system \#include's or not (i.e. using <> instead of "").
1661 const FileEntry *LookupFile(SourceLocation FilenameLoc, StringRef Filename,
1662 bool isAngled, const DirectoryLookup *FromDir,
1663 const FileEntry *FromFile,
1664 const DirectoryLookup *&CurDir,
1665 SmallVectorImpl<char> *SearchPath,
1666 SmallVectorImpl<char> *RelativePath,
1667 ModuleMap::KnownHeader *SuggestedModule,
1668 bool SkipCache = false);
1670 /// \brief Get the DirectoryLookup structure used to find the current
1671 /// FileEntry, if CurLexer is non-null and if applicable.
1673 /// This allows us to implement \#include_next and find directory-specific
1675 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; }
1677 /// \brief Return true if we're in the top-level file, not in a \#include.
1678 bool isInPrimaryFile() const;
1680 /// \brief Handle cases where the \#include name is expanded
1681 /// from a macro as multiple tokens, which need to be glued together.
1683 /// This occurs for code like:
1685 /// \#define FOO <x/y.h>
1688 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one.
1690 /// This code concatenates and consumes tokens up to the '>' token. It
1691 /// returns false if the > was found, otherwise it returns true if it finds
1692 /// and consumes the EOD marker.
1693 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer,
1694 SourceLocation &End);
1696 /// \brief Lex an on-off-switch (C99 6.10.6p2) and verify that it is
1697 /// followed by EOD. Return true if the token is not a valid on-off-switch.
1698 bool LexOnOffSwitch(tok::OnOffSwitch &OOS);
1700 bool CheckMacroName(Token &MacroNameTok, MacroUse isDefineUndef,
1701 bool *ShadowFlag = nullptr);
1705 void PushIncludeMacroStack() {
1706 assert(CurLexerKind != CLK_CachingLexer && "cannot push a caching lexer");
1707 IncludeMacroStack.emplace_back(
1708 CurLexerKind, CurSubmodule, std::move(CurLexer), std::move(CurPTHLexer),
1709 CurPPLexer, std::move(CurTokenLexer), CurDirLookup);
1710 CurPPLexer = nullptr;
1713 void PopIncludeMacroStack() {
1714 CurLexer = std::move(IncludeMacroStack.back().TheLexer);
1715 CurPTHLexer = std::move(IncludeMacroStack.back().ThePTHLexer);
1716 CurPPLexer = IncludeMacroStack.back().ThePPLexer;
1717 CurTokenLexer = std::move(IncludeMacroStack.back().TheTokenLexer);
1718 CurDirLookup = IncludeMacroStack.back().TheDirLookup;
1719 CurSubmodule = IncludeMacroStack.back().TheSubmodule;
1720 CurLexerKind = IncludeMacroStack.back().CurLexerKind;
1721 IncludeMacroStack.pop_back();
1724 void PropagateLineStartLeadingSpaceInfo(Token &Result);
1726 void EnterSubmodule(Module *M, SourceLocation ImportLoc);
1727 void LeaveSubmodule();
1729 /// Determine whether we need to create module macros for #defines in the
1730 /// current context.
1731 bool needModuleMacros() const;
1733 /// Update the set of active module macros and ambiguity flag for a module
1735 void updateModuleMacroInfo(const IdentifierInfo *II, ModuleMacroInfo &Info);
1737 /// \brief Allocate a new MacroInfo object.
1738 MacroInfo *AllocateMacroInfo();
1740 DefMacroDirective *AllocateDefMacroDirective(MacroInfo *MI,
1741 SourceLocation Loc);
1742 UndefMacroDirective *AllocateUndefMacroDirective(SourceLocation UndefLoc);
1743 VisibilityMacroDirective *AllocateVisibilityMacroDirective(SourceLocation Loc,
1746 /// \brief Lex and validate a macro name, which occurs after a
1747 /// \#define or \#undef.
1749 /// \param MacroNameTok Token that represents the name defined or undefined.
1750 /// \param IsDefineUndef Kind if preprocessor directive.
1751 /// \param ShadowFlag Points to flag that is set if macro name shadows
1754 /// This emits a diagnostic, sets the token kind to eod,
1755 /// and discards the rest of the macro line if the macro name is invalid.
1756 void ReadMacroName(Token &MacroNameTok, MacroUse IsDefineUndef = MU_Other,
1757 bool *ShadowFlag = nullptr);
1759 /// The ( starting an argument list of a macro definition has just been read.
1760 /// Lex the rest of the arguments and the closing ), updating \p MI with
1761 /// what we learn and saving in \p LastTok the last token read.
1762 /// Return true if an error occurs parsing the arg list.
1763 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok);
1765 /// We just read a \#if or related directive and decided that the
1766 /// subsequent tokens are in the \#if'd out portion of the
1767 /// file. Lex the rest of the file, until we see an \#endif. If \p
1768 /// FoundNonSkipPortion is true, then we have already emitted code for part of
1769 /// this \#if directive, so \#else/\#elif blocks should never be entered. If
1770 /// \p FoundElse is false, then \#else directives are ok, if not, then we have
1771 /// already seen one so a \#else directive is a duplicate. When this returns,
1772 /// the caller can lex the first valid token.
1773 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc,
1774 bool FoundNonSkipPortion, bool FoundElse,
1775 SourceLocation ElseLoc = SourceLocation());
1777 /// \brief A fast PTH version of SkipExcludedConditionalBlock.
1778 void PTHSkipExcludedConditionalBlock();
1780 /// \brief Evaluate an integer constant expression that may occur after a
1781 /// \#if or \#elif directive and return it as a bool.
1783 /// If the expression is equivalent to "!defined(X)" return X in IfNDefMacro.
1784 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro);
1786 /// \brief Install the standard preprocessor pragmas:
1787 /// \#pragma GCC poison/system_header/dependency and \#pragma once.
1788 void RegisterBuiltinPragmas();
1790 /// \brief Register builtin macros such as __LINE__ with the identifier table.
1791 void RegisterBuiltinMacros();
1793 /// If an identifier token is read that is to be expanded as a macro, handle
1794 /// it and return the next token as 'Tok'. If we lexed a token, return true;
1795 /// otherwise the caller should lex again.
1796 bool HandleMacroExpandedIdentifier(Token &Tok, const MacroDefinition &MD);
1798 /// \brief Cache macro expanded tokens for TokenLexers.
1800 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
1801 /// going to lex in the cache and when it finishes the tokens are removed
1802 /// from the end of the cache.
1803 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer,
1804 ArrayRef<Token> tokens);
1805 void removeCachedMacroExpandedTokensOfLastLexer();
1806 friend void TokenLexer::ExpandFunctionArguments();
1808 /// Determine whether the next preprocessor token to be
1809 /// lexed is a '('. If so, consume the token and return true, if not, this
1810 /// method should have no observable side-effect on the lexed tokens.
1811 bool isNextPPTokenLParen();
1813 /// After reading "MACRO(", this method is invoked to read all of the formal
1814 /// arguments specified for the macro invocation. Returns null on error.
1815 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI,
1816 SourceLocation &ExpansionEnd);
1818 /// \brief If an identifier token is read that is to be expanded
1819 /// as a builtin macro, handle it and return the next token as 'Tok'.
1820 void ExpandBuiltinMacro(Token &Tok);
1822 /// \brief Read a \c _Pragma directive, slice it up, process it, then
1823 /// return the first token after the directive.
1824 /// This assumes that the \c _Pragma token has just been read into \p Tok.
1825 void Handle_Pragma(Token &Tok);
1827 /// \brief Like Handle_Pragma except the pragma text is not enclosed within
1828 /// a string literal.
1829 void HandleMicrosoft__pragma(Token &Tok);
1831 /// \brief Add a lexer to the top of the include stack and
1832 /// start lexing tokens from it instead of the current buffer.
1833 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir);
1835 /// \brief Add a lexer to the top of the include stack and
1836 /// start getting tokens from it using the PTH cache.
1837 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir);
1839 /// \brief Set the FileID for the preprocessor predefines.
1840 void setPredefinesFileID(FileID FID) {
1841 assert(PredefinesFileID.isInvalid() && "PredefinesFileID already set!");
1842 PredefinesFileID = FID;
1845 /// \brief Returns true if we are lexing from a file and not a
1846 /// pragma or a macro.
1847 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) {
1848 return L ? !L->isPragmaLexer() : P != nullptr;
1851 static bool IsFileLexer(const IncludeStackInfo& I) {
1852 return IsFileLexer(I.TheLexer.get(), I.ThePPLexer);
1855 bool IsFileLexer() const {
1856 return IsFileLexer(CurLexer.get(), CurPPLexer);
1859 //===--------------------------------------------------------------------===//
1861 void CachingLex(Token &Result);
1862 bool InCachingLexMode() const {
1863 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means
1864 // that we are past EOF, not that we are in CachingLex mode.
1865 return !CurPPLexer && !CurTokenLexer && !CurPTHLexer &&
1866 !IncludeMacroStack.empty();
1868 void EnterCachingLexMode();
1869 void ExitCachingLexMode() {
1870 if (InCachingLexMode())
1871 RemoveTopOfLexerStack();
1873 const Token &PeekAhead(unsigned N);
1874 void AnnotatePreviousCachedTokens(const Token &Tok);
1876 //===--------------------------------------------------------------------===//
1877 /// Handle*Directive - implement the various preprocessor directives. These
1878 /// should side-effect the current preprocessor object so that the next call
1879 /// to Lex() will return the appropriate token next.
1880 void HandleLineDirective();
1881 void HandleDigitDirective(Token &Tok);
1882 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning);
1883 void HandleIdentSCCSDirective(Token &Tok);
1884 void HandleMacroPublicDirective(Token &Tok);
1885 void HandleMacroPrivateDirective();
1888 void HandleIncludeDirective(SourceLocation HashLoc,
1890 const DirectoryLookup *LookupFrom = nullptr,
1891 const FileEntry *LookupFromFile = nullptr,
1892 bool isImport = false);
1893 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok);
1894 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok);
1895 void HandleImportDirective(SourceLocation HashLoc, Token &Tok);
1896 void HandleMicrosoftImportDirective(Token &Tok);
1899 // Module inclusion testing.
1900 /// \brief Find the module that owns the source or header file that
1901 /// \p Loc points to. If the location is in a file that was included
1902 /// into a module, or is outside any module, returns nullptr.
1903 Module *getModuleForLocation(SourceLocation Loc);
1905 /// \brief Find the module that contains the specified location, either
1906 /// directly or indirectly.
1907 Module *getModuleContainingLocation(SourceLocation Loc);
1909 /// \brief We want to produce a diagnostic at location IncLoc concerning a
1910 /// missing module import.
1912 /// \param IncLoc The location at which the missing import was detected.
1913 /// \param MLoc A location within the desired module at which some desired
1914 /// effect occurred (eg, where a desired entity was declared).
1916 /// \return A file that can be #included to import a module containing MLoc.
1917 /// Null if no such file could be determined or if a #include is not
1919 const FileEntry *getModuleHeaderToIncludeForDiagnostics(SourceLocation IncLoc,
1920 SourceLocation MLoc);
1924 void HandleDefineDirective(Token &Tok, bool ImmediatelyAfterTopLevelIfndef);
1925 void HandleUndefDirective();
1927 // Conditional Inclusion.
1928 void HandleIfdefDirective(Token &Tok, bool isIfndef,
1929 bool ReadAnyTokensBeforeDirective);
1930 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective);
1931 void HandleEndifDirective(Token &Tok);
1932 void HandleElseDirective(Token &Tok);
1933 void HandleElifDirective(Token &Tok);
1936 void HandlePragmaDirective(SourceLocation IntroducerLoc,
1937 PragmaIntroducerKind Introducer);
1939 void HandlePragmaOnce(Token &OnceTok);
1940 void HandlePragmaMark();
1941 void HandlePragmaPoison();
1942 void HandlePragmaSystemHeader(Token &SysHeaderTok);
1943 void HandlePragmaDependency(Token &DependencyTok);
1944 void HandlePragmaPushMacro(Token &Tok);
1945 void HandlePragmaPopMacro(Token &Tok);
1946 void HandlePragmaIncludeAlias(Token &Tok);
1947 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok);
1949 // Return true and store the first token only if any CommentHandler
1950 // has inserted some tokens and getCommentRetentionState() is false.
1951 bool HandleComment(Token &Token, SourceRange Comment);
1953 /// \brief A macro is used, update information about macros that need unused
1955 void markMacroAsUsed(MacroInfo *MI);
1958 /// \brief Abstract base class that describes a handler that will receive
1959 /// source ranges for each of the comments encountered in the source file.
1960 class CommentHandler {
1962 virtual ~CommentHandler();
1964 // The handler shall return true if it has pushed any tokens
1965 // to be read using e.g. EnterToken or EnterTokenStream.
1966 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0;
1969 /// \brief Registry of pragma handlers added by plugins
1970 typedef llvm::Registry<PragmaHandler> PragmaHandlerRegistry;
1972 } // end namespace clang