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.
98 std::shared_ptr<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(std::shared_ptr<PreprocessorOptions> PPOpts,
654 DiagnosticsEngine &diags, LangOptions &opts, SourceManager &SM,
655 HeaderSearch &Headers, ModuleLoader &TheModuleLoader,
656 IdentifierInfoLookup *IILookup = nullptr,
657 bool OwnsHeaderSearch = false,
658 TranslationUnitKind TUKind = TU_Complete);
662 /// \brief Initialize the preprocessor using information about the target.
664 /// \param Target is owned by the caller and must remain valid for the
665 /// lifetime of the preprocessor.
666 /// \param AuxTarget is owned by the caller and must remain valid for
667 /// the lifetime of the preprocessor.
668 void Initialize(const TargetInfo &Target,
669 const TargetInfo *AuxTarget = nullptr);
671 /// \brief Initialize the preprocessor to parse a model file
673 /// To parse model files the preprocessor of the original source is reused to
674 /// preserver the identifier table. However to avoid some duplicate
675 /// information in the preprocessor some cleanup is needed before it is used
676 /// to parse model files. This method does that cleanup.
677 void InitializeForModelFile();
679 /// \brief Cleanup after model file parsing
680 void FinalizeForModelFile();
682 /// \brief Retrieve the preprocessor options used to initialize this
684 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; }
686 DiagnosticsEngine &getDiagnostics() const { return *Diags; }
687 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; }
689 const LangOptions &getLangOpts() const { return LangOpts; }
690 const TargetInfo &getTargetInfo() const { return *Target; }
691 const TargetInfo *getAuxTargetInfo() const { return AuxTarget; }
692 FileManager &getFileManager() const { return FileMgr; }
693 SourceManager &getSourceManager() const { return SourceMgr; }
694 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; }
696 IdentifierTable &getIdentifierTable() { return Identifiers; }
697 const IdentifierTable &getIdentifierTable() const { return Identifiers; }
698 SelectorTable &getSelectorTable() { return Selectors; }
699 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; }
700 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; }
702 void setPTHManager(PTHManager* pm);
704 PTHManager *getPTHManager() { return PTH.get(); }
706 void setExternalSource(ExternalPreprocessorSource *Source) {
707 ExternalSource = Source;
710 ExternalPreprocessorSource *getExternalSource() const {
711 return ExternalSource;
714 /// \brief Retrieve the module loader associated with this preprocessor.
715 ModuleLoader &getModuleLoader() const { return TheModuleLoader; }
717 bool hadModuleLoaderFatalFailure() const {
718 return TheModuleLoader.HadFatalFailure;
721 /// \brief True if we are currently preprocessing a #if or #elif directive
722 bool isParsingIfOrElifDirective() const {
723 return ParsingIfOrElifDirective;
726 /// \brief Control whether the preprocessor retains comments in output.
727 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) {
728 this->KeepComments = KeepComments | KeepMacroComments;
729 this->KeepMacroComments = KeepMacroComments;
732 bool getCommentRetentionState() const { return KeepComments; }
734 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; }
735 bool getPragmasEnabled() const { return PragmasEnabled; }
737 void SetSuppressIncludeNotFoundError(bool Suppress) {
738 SuppressIncludeNotFoundError = Suppress;
741 bool GetSuppressIncludeNotFoundError() {
742 return SuppressIncludeNotFoundError;
745 /// Sets whether the preprocessor is responsible for producing output or if
746 /// it is producing tokens to be consumed by Parse and Sema.
747 void setPreprocessedOutput(bool IsPreprocessedOutput) {
748 PreprocessedOutput = IsPreprocessedOutput;
751 /// Returns true if the preprocessor is responsible for generating output,
752 /// false if it is producing tokens to be consumed by Parse and Sema.
753 bool isPreprocessedOutput() const { return PreprocessedOutput; }
755 /// \brief Return true if we are lexing directly from the specified lexer.
756 bool isCurrentLexer(const PreprocessorLexer *L) const {
757 return CurPPLexer == L;
760 /// \brief Return the current lexer being lexed from.
762 /// Note that this ignores any potentially active macro expansions and _Pragma
763 /// expansions going on at the time.
764 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; }
766 /// \brief Return the current file lexer being lexed from.
768 /// Note that this ignores any potentially active macro expansions and _Pragma
769 /// expansions going on at the time.
770 PreprocessorLexer *getCurrentFileLexer() const;
772 /// \brief Return the submodule owning the file being lexed.
773 Module *getCurrentSubmodule() const { return CurSubmodule; }
775 /// \brief Returns the FileID for the preprocessor predefines.
776 FileID getPredefinesFileID() const { return PredefinesFileID; }
779 /// \brief Accessors for preprocessor callbacks.
781 /// Note that this class takes ownership of any PPCallbacks object given to
783 PPCallbacks *getPPCallbacks() const { return Callbacks.get(); }
784 void addPPCallbacks(std::unique_ptr<PPCallbacks> C) {
786 C = llvm::make_unique<PPChainedCallbacks>(std::move(C),
787 std::move(Callbacks));
788 Callbacks = std::move(C);
792 bool isMacroDefined(StringRef Id) {
793 return isMacroDefined(&Identifiers.get(Id));
795 bool isMacroDefined(const IdentifierInfo *II) {
796 return II->hasMacroDefinition() &&
797 (!getLangOpts().Modules || (bool)getMacroDefinition(II));
800 /// \brief Determine whether II is defined as a macro within the module M,
801 /// if that is a module that we've already preprocessed. Does not check for
802 /// macros imported into M.
803 bool isMacroDefinedInLocalModule(const IdentifierInfo *II, Module *M) {
804 if (!II->hasMacroDefinition())
806 auto I = Submodules.find(M);
807 if (I == Submodules.end())
809 auto J = I->second.Macros.find(II);
810 if (J == I->second.Macros.end())
812 auto *MD = J->second.getLatest();
813 return MD && MD->isDefined();
816 MacroDefinition getMacroDefinition(const IdentifierInfo *II) {
817 if (!II->hasMacroDefinition())
818 return MacroDefinition();
820 MacroState &S = CurSubmoduleState->Macros[II];
821 auto *MD = S.getLatest();
822 while (MD && isa<VisibilityMacroDirective>(MD))
823 MD = MD->getPrevious();
824 return MacroDefinition(dyn_cast_or_null<DefMacroDirective>(MD),
825 S.getActiveModuleMacros(*this, II),
826 S.isAmbiguous(*this, II));
829 MacroDefinition getMacroDefinitionAtLoc(const IdentifierInfo *II,
830 SourceLocation Loc) {
831 if (!II->hadMacroDefinition())
832 return MacroDefinition();
834 MacroState &S = CurSubmoduleState->Macros[II];
835 MacroDirective::DefInfo DI;
836 if (auto *MD = S.getLatest())
837 DI = MD->findDirectiveAtLoc(Loc, getSourceManager());
838 // FIXME: Compute the set of active module macros at the specified location.
839 return MacroDefinition(DI.getDirective(),
840 S.getActiveModuleMacros(*this, II),
841 S.isAmbiguous(*this, II));
844 /// \brief Given an identifier, return its latest non-imported MacroDirective
845 /// if it is \#define'd and not \#undef'd, or null if it isn't \#define'd.
846 MacroDirective *getLocalMacroDirective(const IdentifierInfo *II) const {
847 if (!II->hasMacroDefinition())
850 auto *MD = getLocalMacroDirectiveHistory(II);
851 if (!MD || MD->getDefinition().isUndefined())
857 const MacroInfo *getMacroInfo(const IdentifierInfo *II) const {
858 return const_cast<Preprocessor*>(this)->getMacroInfo(II);
861 MacroInfo *getMacroInfo(const IdentifierInfo *II) {
862 if (!II->hasMacroDefinition())
864 if (auto MD = getMacroDefinition(II))
865 return MD.getMacroInfo();
869 /// \brief Given an identifier, return the latest non-imported macro
870 /// directive for that identifier.
872 /// One can iterate over all previous macro directives from the most recent
874 MacroDirective *getLocalMacroDirectiveHistory(const IdentifierInfo *II) const;
876 /// \brief Add a directive to the macro directive history for this identifier.
877 void appendMacroDirective(IdentifierInfo *II, MacroDirective *MD);
878 DefMacroDirective *appendDefMacroDirective(IdentifierInfo *II, MacroInfo *MI,
879 SourceLocation Loc) {
880 DefMacroDirective *MD = AllocateDefMacroDirective(MI, Loc);
881 appendMacroDirective(II, MD);
884 DefMacroDirective *appendDefMacroDirective(IdentifierInfo *II,
886 return appendDefMacroDirective(II, MI, MI->getDefinitionLoc());
888 /// \brief Set a MacroDirective that was loaded from a PCH file.
889 void setLoadedMacroDirective(IdentifierInfo *II, MacroDirective *ED,
892 /// \brief Register an exported macro for a module and identifier.
893 ModuleMacro *addModuleMacro(Module *Mod, IdentifierInfo *II, MacroInfo *Macro,
894 ArrayRef<ModuleMacro *> Overrides, bool &IsNew);
895 ModuleMacro *getModuleMacro(Module *Mod, IdentifierInfo *II);
897 /// \brief Get the list of leaf (non-overridden) module macros for a name.
898 ArrayRef<ModuleMacro*> getLeafModuleMacros(const IdentifierInfo *II) const {
899 if (II->isOutOfDate())
900 updateOutOfDateIdentifier(const_cast<IdentifierInfo&>(*II));
901 auto I = LeafModuleMacros.find(II);
902 if (I != LeafModuleMacros.end())
908 /// Iterators for the macro history table. Currently defined macros have
909 /// IdentifierInfo::hasMacroDefinition() set and an empty
910 /// MacroInfo::getUndefLoc() at the head of the list.
911 typedef MacroMap::const_iterator macro_iterator;
912 macro_iterator macro_begin(bool IncludeExternalMacros = true) const;
913 macro_iterator macro_end(bool IncludeExternalMacros = true) const;
914 llvm::iterator_range<macro_iterator>
915 macros(bool IncludeExternalMacros = true) const {
916 return llvm::make_range(macro_begin(IncludeExternalMacros),
917 macro_end(IncludeExternalMacros));
921 /// \brief Return the name of the macro defined before \p Loc that has
922 /// spelling \p Tokens. If there are multiple macros with same spelling,
923 /// return the last one defined.
924 StringRef getLastMacroWithSpelling(SourceLocation Loc,
925 ArrayRef<TokenValue> Tokens) const;
927 const std::string &getPredefines() const { return Predefines; }
928 /// \brief Set the predefines for this Preprocessor.
930 /// These predefines are automatically injected when parsing the main file.
931 void setPredefines(const char *P) { Predefines = P; }
932 void setPredefines(StringRef P) { Predefines = P; }
934 /// Return information about the specified preprocessor
935 /// identifier token.
936 IdentifierInfo *getIdentifierInfo(StringRef Name) const {
937 return &Identifiers.get(Name);
940 /// \brief Add the specified pragma handler to this preprocessor.
942 /// If \p Namespace is non-null, then it is a token required to exist on the
943 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC".
944 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler);
945 void AddPragmaHandler(PragmaHandler *Handler) {
946 AddPragmaHandler(StringRef(), Handler);
949 /// \brief Remove the specific pragma handler from this preprocessor.
951 /// If \p Namespace is non-null, then it should be the namespace that
952 /// \p Handler was added to. It is an error to remove a handler that
953 /// has not been registered.
954 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler);
955 void RemovePragmaHandler(PragmaHandler *Handler) {
956 RemovePragmaHandler(StringRef(), Handler);
959 /// Install empty handlers for all pragmas (making them ignored).
960 void IgnorePragmas();
962 /// \brief Add the specified comment handler to the preprocessor.
963 void addCommentHandler(CommentHandler *Handler);
965 /// \brief Remove the specified comment handler.
967 /// It is an error to remove a handler that has not been registered.
968 void removeCommentHandler(CommentHandler *Handler);
970 /// \brief Set the code completion handler to the given object.
971 void setCodeCompletionHandler(CodeCompletionHandler &Handler) {
972 CodeComplete = &Handler;
975 /// \brief Retrieve the current code-completion handler.
976 CodeCompletionHandler *getCodeCompletionHandler() const {
980 /// \brief Clear out the code completion handler.
981 void clearCodeCompletionHandler() {
982 CodeComplete = nullptr;
985 /// \brief Hook used by the lexer to invoke the "natural language" code
986 /// completion point.
987 void CodeCompleteNaturalLanguage();
989 /// \brief Set the code completion token for filtering purposes.
990 void setCodeCompletionIdentifierInfo(IdentifierInfo *Filter) {
991 CodeCompletionII = Filter;
994 /// \brief Get the code completion token for filtering purposes.
995 StringRef getCodeCompletionFilter() {
996 if (CodeCompletionII)
997 return CodeCompletionII->getName();
1001 /// \brief Retrieve the preprocessing record, or NULL if there is no
1002 /// preprocessing record.
1003 PreprocessingRecord *getPreprocessingRecord() const { return Record; }
1005 /// \brief Create a new preprocessing record, which will keep track of
1006 /// all macro expansions, macro definitions, etc.
1007 void createPreprocessingRecord();
1009 /// \brief Enter the specified FileID as the main source file,
1010 /// which implicitly adds the builtin defines etc.
1011 void EnterMainSourceFile();
1013 /// \brief Inform the preprocessor callbacks that processing is complete.
1014 void EndSourceFile();
1016 /// \brief Add a source file to the top of the include stack and
1017 /// start lexing tokens from it instead of the current buffer.
1019 /// Emits a diagnostic, doesn't enter the file, and returns true on error.
1020 bool EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir,
1021 SourceLocation Loc);
1023 /// \brief Add a Macro to the top of the include stack and start lexing
1024 /// tokens from it instead of the current buffer.
1026 /// \param Args specifies the tokens input to a function-like macro.
1027 /// \param ILEnd specifies the location of the ')' for a function-like macro
1028 /// or the identifier for an object-like macro.
1029 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro,
1032 /// \brief Add a "macro" context to the top of the include stack,
1033 /// which will cause the lexer to start returning the specified tokens.
1035 /// If \p DisableMacroExpansion is true, tokens lexed from the token stream
1036 /// will not be subject to further macro expansion. Otherwise, these tokens
1037 /// will be re-macro-expanded when/if expansion is enabled.
1039 /// If \p OwnsTokens is false, this method assumes that the specified stream
1040 /// of tokens has a permanent owner somewhere, so they do not need to be
1041 /// copied. If it is true, it assumes the array of tokens is allocated with
1042 /// \c new[] and the Preprocessor will delete[] it.
1044 void EnterTokenStream(const Token *Toks, unsigned NumToks,
1045 bool DisableMacroExpansion, bool OwnsTokens);
1048 void EnterTokenStream(std::unique_ptr<Token[]> Toks, unsigned NumToks,
1049 bool DisableMacroExpansion) {
1050 EnterTokenStream(Toks.release(), NumToks, DisableMacroExpansion, true);
1052 void EnterTokenStream(ArrayRef<Token> Toks, bool DisableMacroExpansion) {
1053 EnterTokenStream(Toks.data(), Toks.size(), DisableMacroExpansion, false);
1056 /// \brief Pop the current lexer/macro exp off the top of the lexer stack.
1058 /// This should only be used in situations where the current state of the
1059 /// top-of-stack lexer is known.
1060 void RemoveTopOfLexerStack();
1062 /// From the point that this method is called, and until
1063 /// CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor
1064 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will
1065 /// make the Preprocessor re-lex the same tokens.
1067 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can
1068 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will
1069 /// be combined with the EnableBacktrackAtThisPos calls in reverse order.
1071 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack
1072 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of
1073 /// tokens will continue indefinitely.
1075 void EnableBacktrackAtThisPos();
1077 /// \brief Disable the last EnableBacktrackAtThisPos call.
1078 void CommitBacktrackedTokens();
1080 /// \brief Make Preprocessor re-lex the tokens that were lexed since
1081 /// EnableBacktrackAtThisPos() was previously called.
1084 /// \brief True if EnableBacktrackAtThisPos() was called and
1085 /// caching of tokens is on.
1086 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); }
1088 /// \brief Lex the next token for this preprocessor.
1089 void Lex(Token &Result);
1091 void LexAfterModuleImport(Token &Result);
1093 void makeModuleVisible(Module *M, SourceLocation Loc);
1095 SourceLocation getModuleImportLoc(Module *M) const {
1096 return CurSubmoduleState->VisibleModules.getImportLoc(M);
1099 /// \brief Lex a string literal, which may be the concatenation of multiple
1100 /// string literals and may even come from macro expansion.
1101 /// \returns true on success, false if a error diagnostic has been generated.
1102 bool LexStringLiteral(Token &Result, std::string &String,
1103 const char *DiagnosticTag, bool AllowMacroExpansion) {
1104 if (AllowMacroExpansion)
1107 LexUnexpandedToken(Result);
1108 return FinishLexStringLiteral(Result, String, DiagnosticTag,
1109 AllowMacroExpansion);
1112 /// \brief Complete the lexing of a string literal where the first token has
1113 /// already been lexed (see LexStringLiteral).
1114 bool FinishLexStringLiteral(Token &Result, std::string &String,
1115 const char *DiagnosticTag,
1116 bool AllowMacroExpansion);
1118 /// \brief Lex a token. If it's a comment, keep lexing until we get
1119 /// something not a comment.
1121 /// This is useful in -E -C mode where comments would foul up preprocessor
1122 /// directive handling.
1123 void LexNonComment(Token &Result) {
1126 while (Result.getKind() == tok::comment);
1129 /// \brief Just like Lex, but disables macro expansion of identifier tokens.
1130 void LexUnexpandedToken(Token &Result) {
1131 // Disable macro expansion.
1132 bool OldVal = DisableMacroExpansion;
1133 DisableMacroExpansion = true;
1138 DisableMacroExpansion = OldVal;
1141 /// \brief Like LexNonComment, but this disables macro expansion of
1142 /// identifier tokens.
1143 void LexUnexpandedNonComment(Token &Result) {
1145 LexUnexpandedToken(Result);
1146 while (Result.getKind() == tok::comment);
1149 /// \brief Parses a simple integer literal to get its numeric value. Floating
1150 /// point literals and user defined literals are rejected. Used primarily to
1151 /// handle pragmas that accept integer arguments.
1152 bool parseSimpleIntegerLiteral(Token &Tok, uint64_t &Value);
1154 /// Disables macro expansion everywhere except for preprocessor directives.
1155 void SetMacroExpansionOnlyInDirectives() {
1156 DisableMacroExpansion = true;
1157 MacroExpansionInDirectivesOverride = true;
1160 /// \brief Peeks ahead N tokens and returns that token without consuming any
1163 /// LookAhead(0) returns the next token that would be returned by Lex(),
1164 /// LookAhead(1) returns the token after it, etc. This returns normal
1165 /// tokens after phase 5. As such, it is equivalent to using
1166 /// 'Lex', not 'LexUnexpandedToken'.
1167 const Token &LookAhead(unsigned N) {
1168 if (CachedLexPos + N < CachedTokens.size())
1169 return CachedTokens[CachedLexPos+N];
1171 return PeekAhead(N+1);
1174 /// \brief When backtracking is enabled and tokens are cached,
1175 /// this allows to revert a specific number of tokens.
1177 /// Note that the number of tokens being reverted should be up to the last
1178 /// backtrack position, not more.
1179 void RevertCachedTokens(unsigned N) {
1180 assert(isBacktrackEnabled() &&
1181 "Should only be called when tokens are cached for backtracking");
1182 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back())
1183 && "Should revert tokens up to the last backtrack position, not more");
1184 assert(signed(CachedLexPos) - signed(N) >= 0 &&
1185 "Corrupted backtrack positions ?");
1189 /// \brief Enters a token in the token stream to be lexed next.
1191 /// If BackTrack() is called afterwards, the token will remain at the
1192 /// insertion point.
1193 void EnterToken(const Token &Tok) {
1194 EnterCachingLexMode();
1195 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok);
1198 /// We notify the Preprocessor that if it is caching tokens (because
1199 /// backtrack is enabled) it should replace the most recent cached tokens
1200 /// with the given annotation token. This function has no effect if
1201 /// backtracking is not enabled.
1203 /// Note that the use of this function is just for optimization, so that the
1204 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is
1206 void AnnotateCachedTokens(const Token &Tok) {
1207 assert(Tok.isAnnotation() && "Expected annotation token");
1208 if (CachedLexPos != 0 && isBacktrackEnabled())
1209 AnnotatePreviousCachedTokens(Tok);
1212 /// Get the location of the last cached token, suitable for setting the end
1213 /// location of an annotation token.
1214 SourceLocation getLastCachedTokenLocation() const {
1215 assert(CachedLexPos != 0);
1216 return CachedTokens[CachedLexPos-1].getLastLoc();
1219 /// \brief Whether \p Tok is the most recent token (`CachedLexPos - 1`) in
1221 bool IsPreviousCachedToken(const Token &Tok) const;
1223 /// \brief Replace token in `CachedLexPos - 1` in CachedTokens by the tokens
1226 /// Useful when a token needs to be split in smaller ones and CachedTokens
1227 /// most recent token must to be updated to reflect that.
1228 void ReplacePreviousCachedToken(ArrayRef<Token> NewToks);
1230 /// \brief Replace the last token with an annotation token.
1232 /// Like AnnotateCachedTokens(), this routine replaces an
1233 /// already-parsed (and resolved) token with an annotation
1234 /// token. However, this routine only replaces the last token with
1235 /// the annotation token; it does not affect any other cached
1236 /// tokens. This function has no effect if backtracking is not
1238 void ReplaceLastTokenWithAnnotation(const Token &Tok) {
1239 assert(Tok.isAnnotation() && "Expected annotation token");
1240 if (CachedLexPos != 0 && isBacktrackEnabled())
1241 CachedTokens[CachedLexPos-1] = Tok;
1244 /// Update the current token to represent the provided
1245 /// identifier, in order to cache an action performed by typo correction.
1246 void TypoCorrectToken(const Token &Tok) {
1247 assert(Tok.getIdentifierInfo() && "Expected identifier token");
1248 if (CachedLexPos != 0 && isBacktrackEnabled())
1249 CachedTokens[CachedLexPos-1] = Tok;
1252 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/
1253 /// CurTokenLexer pointers.
1254 void recomputeCurLexerKind();
1256 /// \brief Returns true if incremental processing is enabled
1257 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; }
1259 /// \brief Enables the incremental processing
1260 void enableIncrementalProcessing(bool value = true) {
1261 IncrementalProcessing = value;
1264 /// \brief Specify the point at which code-completion will be performed.
1266 /// \param File the file in which code completion should occur. If
1267 /// this file is included multiple times, code-completion will
1268 /// perform completion the first time it is included. If NULL, this
1269 /// function clears out the code-completion point.
1271 /// \param Line the line at which code completion should occur
1274 /// \param Column the column at which code completion should occur
1277 /// \returns true if an error occurred, false otherwise.
1278 bool SetCodeCompletionPoint(const FileEntry *File,
1279 unsigned Line, unsigned Column);
1281 /// \brief Determine if we are performing code completion.
1282 bool isCodeCompletionEnabled() const { return CodeCompletionFile != nullptr; }
1284 /// \brief Returns the location of the code-completion point.
1286 /// Returns an invalid location if code-completion is not enabled or the file
1287 /// containing the code-completion point has not been lexed yet.
1288 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; }
1290 /// \brief Returns the start location of the file of code-completion point.
1292 /// Returns an invalid location if code-completion is not enabled or the file
1293 /// containing the code-completion point has not been lexed yet.
1294 SourceLocation getCodeCompletionFileLoc() const {
1295 return CodeCompletionFileLoc;
1298 /// \brief Returns true if code-completion is enabled and we have hit the
1299 /// code-completion point.
1300 bool isCodeCompletionReached() const { return CodeCompletionReached; }
1302 /// \brief Note that we hit the code-completion point.
1303 void setCodeCompletionReached() {
1304 assert(isCodeCompletionEnabled() && "Code-completion not enabled!");
1305 CodeCompletionReached = true;
1306 // Silence any diagnostics that occur after we hit the code-completion.
1307 getDiagnostics().setSuppressAllDiagnostics(true);
1310 /// \brief The location of the currently-active \#pragma clang
1311 /// arc_cf_code_audited begin.
1313 /// Returns an invalid location if there is no such pragma active.
1314 SourceLocation getPragmaARCCFCodeAuditedLoc() const {
1315 return PragmaARCCFCodeAuditedLoc;
1318 /// \brief Set the location of the currently-active \#pragma clang
1319 /// arc_cf_code_audited begin. An invalid location ends the pragma.
1320 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) {
1321 PragmaARCCFCodeAuditedLoc = Loc;
1324 /// \brief The location of the currently-active \#pragma clang
1325 /// assume_nonnull begin.
1327 /// Returns an invalid location if there is no such pragma active.
1328 SourceLocation getPragmaAssumeNonNullLoc() const {
1329 return PragmaAssumeNonNullLoc;
1332 /// \brief Set the location of the currently-active \#pragma clang
1333 /// assume_nonnull begin. An invalid location ends the pragma.
1334 void setPragmaAssumeNonNullLoc(SourceLocation Loc) {
1335 PragmaAssumeNonNullLoc = Loc;
1338 /// \brief Set the directory in which the main file should be considered
1339 /// to have been found, if it is not a real file.
1340 void setMainFileDir(const DirectoryEntry *Dir) {
1344 /// \brief Instruct the preprocessor to skip part of the main source file.
1346 /// \param Bytes The number of bytes in the preamble to skip.
1348 /// \param StartOfLine Whether skipping these bytes puts the lexer at the
1349 /// start of a line.
1350 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) {
1351 SkipMainFilePreamble.first = Bytes;
1352 SkipMainFilePreamble.second = StartOfLine;
1355 /// Forwarding function for diagnostics. This emits a diagnostic at
1356 /// the specified Token's location, translating the token's start
1357 /// position in the current buffer into a SourcePosition object for rendering.
1358 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const {
1359 return Diags->Report(Loc, DiagID);
1362 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const {
1363 return Diags->Report(Tok.getLocation(), DiagID);
1366 /// Return the 'spelling' of the token at the given
1367 /// location; does not go up to the spelling location or down to the
1368 /// expansion location.
1370 /// \param buffer A buffer which will be used only if the token requires
1371 /// "cleaning", e.g. if it contains trigraphs or escaped newlines
1372 /// \param invalid If non-null, will be set \c true if an error occurs.
1373 StringRef getSpelling(SourceLocation loc,
1374 SmallVectorImpl<char> &buffer,
1375 bool *invalid = nullptr) const {
1376 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid);
1379 /// \brief Return the 'spelling' of the Tok token.
1381 /// The spelling of a token is the characters used to represent the token in
1382 /// the source file after trigraph expansion and escaped-newline folding. In
1383 /// particular, this wants to get the true, uncanonicalized, spelling of
1384 /// things like digraphs, UCNs, etc.
1386 /// \param Invalid If non-null, will be set \c true if an error occurs.
1387 std::string getSpelling(const Token &Tok, bool *Invalid = nullptr) const {
1388 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid);
1391 /// \brief Get the spelling of a token into a preallocated buffer, instead
1392 /// of as an std::string.
1394 /// The caller is required to allocate enough space for the token, which is
1395 /// guaranteed to be at least Tok.getLength() bytes long. The length of the
1396 /// actual result is returned.
1398 /// Note that this method may do two possible things: it may either fill in
1399 /// the buffer specified with characters, or it may *change the input pointer*
1400 /// to point to a constant buffer with the data already in it (avoiding a
1401 /// copy). The caller is not allowed to modify the returned buffer pointer
1402 /// if an internal buffer is returned.
1403 unsigned getSpelling(const Token &Tok, const char *&Buffer,
1404 bool *Invalid = nullptr) const {
1405 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid);
1408 /// \brief Get the spelling of a token into a SmallVector.
1410 /// Note that the returned StringRef may not point to the
1411 /// supplied buffer if a copy can be avoided.
1412 StringRef getSpelling(const Token &Tok,
1413 SmallVectorImpl<char> &Buffer,
1414 bool *Invalid = nullptr) const;
1416 /// \brief Relex the token at the specified location.
1417 /// \returns true if there was a failure, false on success.
1418 bool getRawToken(SourceLocation Loc, Token &Result,
1419 bool IgnoreWhiteSpace = false) {
1420 return Lexer::getRawToken(Loc, Result, SourceMgr, LangOpts, IgnoreWhiteSpace);
1423 /// \brief Given a Token \p Tok that is a numeric constant with length 1,
1424 /// return the character.
1426 getSpellingOfSingleCharacterNumericConstant(const Token &Tok,
1427 bool *Invalid = nullptr) const {
1428 assert(Tok.is(tok::numeric_constant) &&
1429 Tok.getLength() == 1 && "Called on unsupported token");
1430 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1");
1432 // If the token is carrying a literal data pointer, just use it.
1433 if (const char *D = Tok.getLiteralData())
1436 // Otherwise, fall back on getCharacterData, which is slower, but always
1438 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid);
1441 /// \brief Retrieve the name of the immediate macro expansion.
1443 /// This routine starts from a source location, and finds the name of the
1444 /// macro responsible for its immediate expansion. It looks through any
1445 /// intervening macro argument expansions to compute this. It returns a
1446 /// StringRef that refers to the SourceManager-owned buffer of the source
1447 /// where that macro name is spelled. Thus, the result shouldn't out-live
1448 /// the SourceManager.
1449 StringRef getImmediateMacroName(SourceLocation Loc) {
1450 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts());
1453 /// \brief Plop the specified string into a scratch buffer and set the
1454 /// specified token's location and length to it.
1456 /// If specified, the source location provides a location of the expansion
1457 /// point of the token.
1458 void CreateString(StringRef Str, Token &Tok,
1459 SourceLocation ExpansionLocStart = SourceLocation(),
1460 SourceLocation ExpansionLocEnd = SourceLocation());
1462 /// \brief Computes the source location just past the end of the
1463 /// token at this source location.
1465 /// This routine can be used to produce a source location that
1466 /// points just past the end of the token referenced by \p Loc, and
1467 /// is generally used when a diagnostic needs to point just after a
1468 /// token where it expected something different that it received. If
1469 /// the returned source location would not be meaningful (e.g., if
1470 /// it points into a macro), this routine returns an invalid
1471 /// source location.
1473 /// \param Offset an offset from the end of the token, where the source
1474 /// location should refer to. The default offset (0) produces a source
1475 /// location pointing just past the end of the token; an offset of 1 produces
1476 /// a source location pointing to the last character in the token, etc.
1477 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) {
1478 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts);
1481 /// \brief Returns true if the given MacroID location points at the first
1482 /// token of the macro expansion.
1484 /// \param MacroBegin If non-null and function returns true, it is set to
1485 /// begin location of the macro.
1486 bool isAtStartOfMacroExpansion(SourceLocation loc,
1487 SourceLocation *MacroBegin = nullptr) const {
1488 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts,
1492 /// \brief Returns true if the given MacroID location points at the last
1493 /// token of the macro expansion.
1495 /// \param MacroEnd If non-null and function returns true, it is set to
1496 /// end location of the macro.
1497 bool isAtEndOfMacroExpansion(SourceLocation loc,
1498 SourceLocation *MacroEnd = nullptr) const {
1499 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd);
1502 /// \brief Print the token to stderr, used for debugging.
1503 void DumpToken(const Token &Tok, bool DumpFlags = false) const;
1504 void DumpLocation(SourceLocation Loc) const;
1505 void DumpMacro(const MacroInfo &MI) const;
1506 void dumpMacroInfo(const IdentifierInfo *II);
1508 /// \brief Given a location that specifies the start of a
1509 /// token, return a new location that specifies a character within the token.
1510 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
1511 unsigned Char) const {
1512 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts);
1515 /// \brief Increment the counters for the number of token paste operations
1518 /// If fast was specified, this is a 'fast paste' case we handled.
1519 void IncrementPasteCounter(bool isFast) {
1521 ++NumFastTokenPaste;
1528 size_t getTotalMemory() const;
1530 /// When the macro expander pastes together a comment (/##/) in Microsoft
1531 /// mode, this method handles updating the current state, returning the
1532 /// token on the next source line.
1533 void HandleMicrosoftCommentPaste(Token &Tok);
1535 //===--------------------------------------------------------------------===//
1536 // Preprocessor callback methods. These are invoked by a lexer as various
1537 // directives and events are found.
1539 /// Given a tok::raw_identifier token, look up the
1540 /// identifier information for the token and install it into the token,
1541 /// updating the token kind accordingly.
1542 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const;
1545 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons;
1549 /// \brief Specifies the reason for poisoning an identifier.
1551 /// If that identifier is accessed while poisoned, then this reason will be
1552 /// used instead of the default "poisoned" diagnostic.
1553 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID);
1555 /// \brief Display reason for poisoned identifier.
1556 void HandlePoisonedIdentifier(Token & Tok);
1558 void MaybeHandlePoisonedIdentifier(Token & Identifier) {
1559 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) {
1560 if(II->isPoisoned()) {
1561 HandlePoisonedIdentifier(Identifier);
1567 /// Identifiers used for SEH handling in Borland. These are only
1568 /// allowed in particular circumstances
1570 IdentifierInfo *Ident__exception_code,
1571 *Ident___exception_code,
1572 *Ident_GetExceptionCode;
1573 // __except filter expression
1574 IdentifierInfo *Ident__exception_info,
1575 *Ident___exception_info,
1576 *Ident_GetExceptionInfo;
1578 IdentifierInfo *Ident__abnormal_termination,
1579 *Ident___abnormal_termination,
1580 *Ident_AbnormalTermination;
1582 const char *getCurLexerEndPos();
1585 void PoisonSEHIdentifiers(bool Poison = true); // Borland
1587 /// \brief Callback invoked when the lexer reads an identifier and has
1588 /// filled in the tokens IdentifierInfo member.
1590 /// This callback potentially macro expands it or turns it into a named
1591 /// token (like 'for').
1593 /// \returns true if we actually computed a token, false if we need to
1595 bool HandleIdentifier(Token &Identifier);
1598 /// \brief Callback invoked when the lexer hits the end of the current file.
1600 /// This either returns the EOF token and returns true, or
1601 /// pops a level off the include stack and returns false, at which point the
1602 /// client should call lex again.
1603 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false);
1605 /// \brief Callback invoked when the current TokenLexer hits the end of its
1607 bool HandleEndOfTokenLexer(Token &Result);
1609 /// \brief Callback invoked when the lexer sees a # token at the start of a
1612 /// This consumes the directive, modifies the lexer/preprocessor state, and
1613 /// advances the lexer(s) so that the next token read is the correct one.
1614 void HandleDirective(Token &Result);
1616 /// \brief Ensure that the next token is a tok::eod token.
1618 /// If not, emit a diagnostic and consume up until the eod.
1619 /// If \p EnableMacros is true, then we consider macros that expand to zero
1620 /// tokens as being ok.
1621 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false);
1623 /// \brief Read and discard all tokens remaining on the current line until
1624 /// the tok::eod token is found.
1625 void DiscardUntilEndOfDirective();
1627 /// \brief Returns true if the preprocessor has seen a use of
1628 /// __DATE__ or __TIME__ in the file so far.
1629 bool SawDateOrTime() const {
1630 return DATELoc != SourceLocation() || TIMELoc != SourceLocation();
1632 unsigned getCounterValue() const { return CounterValue; }
1633 void setCounterValue(unsigned V) { CounterValue = V; }
1635 /// \brief Retrieves the module that we're currently building, if any.
1636 Module *getCurrentModule();
1638 /// \brief Allocate a new MacroInfo object with the provided SourceLocation.
1639 MacroInfo *AllocateMacroInfo(SourceLocation L);
1641 /// \brief Allocate a new MacroInfo object loaded from an AST file.
1642 MacroInfo *AllocateDeserializedMacroInfo(SourceLocation L,
1643 unsigned SubModuleID);
1645 /// \brief Turn the specified lexer token into a fully checked and spelled
1646 /// filename, e.g. as an operand of \#include.
1648 /// The caller is expected to provide a buffer that is large enough to hold
1649 /// the spelling of the filename, but is also expected to handle the case
1650 /// when this method decides to use a different buffer.
1652 /// \returns true if the input filename was in <>'s or false if it was
1654 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename);
1656 /// \brief Given a "foo" or \<foo> reference, look up the indicated file.
1658 /// Returns null on failure. \p isAngled indicates whether the file
1659 /// reference is for system \#include's or not (i.e. using <> instead of "").
1660 const FileEntry *LookupFile(SourceLocation FilenameLoc, StringRef Filename,
1661 bool isAngled, const DirectoryLookup *FromDir,
1662 const FileEntry *FromFile,
1663 const DirectoryLookup *&CurDir,
1664 SmallVectorImpl<char> *SearchPath,
1665 SmallVectorImpl<char> *RelativePath,
1666 ModuleMap::KnownHeader *SuggestedModule,
1667 bool SkipCache = false);
1669 /// \brief Get the DirectoryLookup structure used to find the current
1670 /// FileEntry, if CurLexer is non-null and if applicable.
1672 /// This allows us to implement \#include_next and find directory-specific
1674 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; }
1676 /// \brief Return true if we're in the top-level file, not in a \#include.
1677 bool isInPrimaryFile() const;
1679 /// \brief Handle cases where the \#include name is expanded
1680 /// from a macro as multiple tokens, which need to be glued together.
1682 /// This occurs for code like:
1684 /// \#define FOO <x/y.h>
1687 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one.
1689 /// This code concatenates and consumes tokens up to the '>' token. It
1690 /// returns false if the > was found, otherwise it returns true if it finds
1691 /// and consumes the EOD marker.
1692 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer,
1693 SourceLocation &End);
1695 /// \brief Lex an on-off-switch (C99 6.10.6p2) and verify that it is
1696 /// followed by EOD. Return true if the token is not a valid on-off-switch.
1697 bool LexOnOffSwitch(tok::OnOffSwitch &OOS);
1699 bool CheckMacroName(Token &MacroNameTok, MacroUse isDefineUndef,
1700 bool *ShadowFlag = nullptr);
1704 void PushIncludeMacroStack() {
1705 assert(CurLexerKind != CLK_CachingLexer && "cannot push a caching lexer");
1706 IncludeMacroStack.emplace_back(
1707 CurLexerKind, CurSubmodule, std::move(CurLexer), std::move(CurPTHLexer),
1708 CurPPLexer, std::move(CurTokenLexer), CurDirLookup);
1709 CurPPLexer = nullptr;
1712 void PopIncludeMacroStack() {
1713 CurLexer = std::move(IncludeMacroStack.back().TheLexer);
1714 CurPTHLexer = std::move(IncludeMacroStack.back().ThePTHLexer);
1715 CurPPLexer = IncludeMacroStack.back().ThePPLexer;
1716 CurTokenLexer = std::move(IncludeMacroStack.back().TheTokenLexer);
1717 CurDirLookup = IncludeMacroStack.back().TheDirLookup;
1718 CurSubmodule = IncludeMacroStack.back().TheSubmodule;
1719 CurLexerKind = IncludeMacroStack.back().CurLexerKind;
1720 IncludeMacroStack.pop_back();
1723 void PropagateLineStartLeadingSpaceInfo(Token &Result);
1725 void EnterSubmodule(Module *M, SourceLocation ImportLoc);
1726 void LeaveSubmodule();
1728 /// Determine whether we need to create module macros for #defines in the
1729 /// current context.
1730 bool needModuleMacros() const;
1732 /// Update the set of active module macros and ambiguity flag for a module
1734 void updateModuleMacroInfo(const IdentifierInfo *II, ModuleMacroInfo &Info);
1736 /// \brief Allocate a new MacroInfo object.
1737 MacroInfo *AllocateMacroInfo();
1739 DefMacroDirective *AllocateDefMacroDirective(MacroInfo *MI,
1740 SourceLocation Loc);
1741 UndefMacroDirective *AllocateUndefMacroDirective(SourceLocation UndefLoc);
1742 VisibilityMacroDirective *AllocateVisibilityMacroDirective(SourceLocation Loc,
1745 /// \brief Lex and validate a macro name, which occurs after a
1746 /// \#define or \#undef.
1748 /// \param MacroNameTok Token that represents the name defined or undefined.
1749 /// \param IsDefineUndef Kind if preprocessor directive.
1750 /// \param ShadowFlag Points to flag that is set if macro name shadows
1753 /// This emits a diagnostic, sets the token kind to eod,
1754 /// and discards the rest of the macro line if the macro name is invalid.
1755 void ReadMacroName(Token &MacroNameTok, MacroUse IsDefineUndef = MU_Other,
1756 bool *ShadowFlag = nullptr);
1758 /// The ( starting an argument list of a macro definition has just been read.
1759 /// Lex the rest of the arguments and the closing ), updating \p MI with
1760 /// what we learn and saving in \p LastTok the last token read.
1761 /// Return true if an error occurs parsing the arg list.
1762 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok);
1764 /// We just read a \#if or related directive and decided that the
1765 /// subsequent tokens are in the \#if'd out portion of the
1766 /// file. Lex the rest of the file, until we see an \#endif. If \p
1767 /// FoundNonSkipPortion is true, then we have already emitted code for part of
1768 /// this \#if directive, so \#else/\#elif blocks should never be entered. If
1769 /// \p FoundElse is false, then \#else directives are ok, if not, then we have
1770 /// already seen one so a \#else directive is a duplicate. When this returns,
1771 /// the caller can lex the first valid token.
1772 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc,
1773 bool FoundNonSkipPortion, bool FoundElse,
1774 SourceLocation ElseLoc = SourceLocation());
1776 /// \brief A fast PTH version of SkipExcludedConditionalBlock.
1777 void PTHSkipExcludedConditionalBlock();
1779 /// \brief Evaluate an integer constant expression that may occur after a
1780 /// \#if or \#elif directive and return it as a bool.
1782 /// If the expression is equivalent to "!defined(X)" return X in IfNDefMacro.
1783 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro);
1785 /// \brief Install the standard preprocessor pragmas:
1786 /// \#pragma GCC poison/system_header/dependency and \#pragma once.
1787 void RegisterBuiltinPragmas();
1789 /// \brief Register builtin macros such as __LINE__ with the identifier table.
1790 void RegisterBuiltinMacros();
1792 /// If an identifier token is read that is to be expanded as a macro, handle
1793 /// it and return the next token as 'Tok'. If we lexed a token, return true;
1794 /// otherwise the caller should lex again.
1795 bool HandleMacroExpandedIdentifier(Token &Tok, const MacroDefinition &MD);
1797 /// \brief Cache macro expanded tokens for TokenLexers.
1799 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
1800 /// going to lex in the cache and when it finishes the tokens are removed
1801 /// from the end of the cache.
1802 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer,
1803 ArrayRef<Token> tokens);
1804 void removeCachedMacroExpandedTokensOfLastLexer();
1805 friend void TokenLexer::ExpandFunctionArguments();
1807 /// Determine whether the next preprocessor token to be
1808 /// lexed is a '('. If so, consume the token and return true, if not, this
1809 /// method should have no observable side-effect on the lexed tokens.
1810 bool isNextPPTokenLParen();
1812 /// After reading "MACRO(", this method is invoked to read all of the formal
1813 /// arguments specified for the macro invocation. Returns null on error.
1814 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI,
1815 SourceLocation &ExpansionEnd);
1817 /// \brief If an identifier token is read that is to be expanded
1818 /// as a builtin macro, handle it and return the next token as 'Tok'.
1819 void ExpandBuiltinMacro(Token &Tok);
1821 /// \brief Read a \c _Pragma directive, slice it up, process it, then
1822 /// return the first token after the directive.
1823 /// This assumes that the \c _Pragma token has just been read into \p Tok.
1824 void Handle_Pragma(Token &Tok);
1826 /// \brief Like Handle_Pragma except the pragma text is not enclosed within
1827 /// a string literal.
1828 void HandleMicrosoft__pragma(Token &Tok);
1830 /// \brief Add a lexer to the top of the include stack and
1831 /// start lexing tokens from it instead of the current buffer.
1832 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir);
1834 /// \brief Add a lexer to the top of the include stack and
1835 /// start getting tokens from it using the PTH cache.
1836 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir);
1838 /// \brief Set the FileID for the preprocessor predefines.
1839 void setPredefinesFileID(FileID FID) {
1840 assert(PredefinesFileID.isInvalid() && "PredefinesFileID already set!");
1841 PredefinesFileID = FID;
1844 /// \brief Returns true if we are lexing from a file and not a
1845 /// pragma or a macro.
1846 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) {
1847 return L ? !L->isPragmaLexer() : P != nullptr;
1850 static bool IsFileLexer(const IncludeStackInfo& I) {
1851 return IsFileLexer(I.TheLexer.get(), I.ThePPLexer);
1854 bool IsFileLexer() const {
1855 return IsFileLexer(CurLexer.get(), CurPPLexer);
1858 //===--------------------------------------------------------------------===//
1860 void CachingLex(Token &Result);
1861 bool InCachingLexMode() const {
1862 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means
1863 // that we are past EOF, not that we are in CachingLex mode.
1864 return !CurPPLexer && !CurTokenLexer && !CurPTHLexer &&
1865 !IncludeMacroStack.empty();
1867 void EnterCachingLexMode();
1868 void ExitCachingLexMode() {
1869 if (InCachingLexMode())
1870 RemoveTopOfLexerStack();
1872 const Token &PeekAhead(unsigned N);
1873 void AnnotatePreviousCachedTokens(const Token &Tok);
1875 //===--------------------------------------------------------------------===//
1876 /// Handle*Directive - implement the various preprocessor directives. These
1877 /// should side-effect the current preprocessor object so that the next call
1878 /// to Lex() will return the appropriate token next.
1879 void HandleLineDirective();
1880 void HandleDigitDirective(Token &Tok);
1881 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning);
1882 void HandleIdentSCCSDirective(Token &Tok);
1883 void HandleMacroPublicDirective(Token &Tok);
1884 void HandleMacroPrivateDirective();
1887 void HandleIncludeDirective(SourceLocation HashLoc,
1889 const DirectoryLookup *LookupFrom = nullptr,
1890 const FileEntry *LookupFromFile = nullptr,
1891 bool isImport = false);
1892 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok);
1893 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok);
1894 void HandleImportDirective(SourceLocation HashLoc, Token &Tok);
1895 void HandleMicrosoftImportDirective(Token &Tok);
1898 // Module inclusion testing.
1899 /// \brief Find the module that owns the source or header file that
1900 /// \p Loc points to. If the location is in a file that was included
1901 /// into a module, or is outside any module, returns nullptr.
1902 Module *getModuleForLocation(SourceLocation Loc);
1904 /// \brief Find the module that contains the specified location, either
1905 /// directly or indirectly.
1906 Module *getModuleContainingLocation(SourceLocation Loc);
1908 /// \brief We want to produce a diagnostic at location IncLoc concerning a
1909 /// missing module import.
1911 /// \param IncLoc The location at which the missing import was detected.
1912 /// \param MLoc A location within the desired module at which some desired
1913 /// effect occurred (eg, where a desired entity was declared).
1915 /// \return A file that can be #included to import a module containing MLoc.
1916 /// Null if no such file could be determined or if a #include is not
1918 const FileEntry *getModuleHeaderToIncludeForDiagnostics(SourceLocation IncLoc,
1919 SourceLocation MLoc);
1923 void HandleDefineDirective(Token &Tok, bool ImmediatelyAfterTopLevelIfndef);
1924 void HandleUndefDirective();
1926 // Conditional Inclusion.
1927 void HandleIfdefDirective(Token &Tok, bool isIfndef,
1928 bool ReadAnyTokensBeforeDirective);
1929 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective);
1930 void HandleEndifDirective(Token &Tok);
1931 void HandleElseDirective(Token &Tok);
1932 void HandleElifDirective(Token &Tok);
1935 void HandlePragmaDirective(SourceLocation IntroducerLoc,
1936 PragmaIntroducerKind Introducer);
1938 void HandlePragmaOnce(Token &OnceTok);
1939 void HandlePragmaMark();
1940 void HandlePragmaPoison();
1941 void HandlePragmaSystemHeader(Token &SysHeaderTok);
1942 void HandlePragmaDependency(Token &DependencyTok);
1943 void HandlePragmaPushMacro(Token &Tok);
1944 void HandlePragmaPopMacro(Token &Tok);
1945 void HandlePragmaIncludeAlias(Token &Tok);
1946 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok);
1948 // Return true and store the first token only if any CommentHandler
1949 // has inserted some tokens and getCommentRetentionState() is false.
1950 bool HandleComment(Token &Token, SourceRange Comment);
1952 /// \brief A macro is used, update information about macros that need unused
1954 void markMacroAsUsed(MacroInfo *MI);
1957 /// \brief Abstract base class that describes a handler that will receive
1958 /// source ranges for each of the comments encountered in the source file.
1959 class CommentHandler {
1961 virtual ~CommentHandler();
1963 // The handler shall return true if it has pushed any tokens
1964 // to be read using e.g. EnterToken or EnterTokenStream.
1965 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0;
1968 /// \brief Registry of pragma handlers added by plugins
1969 typedef llvm::Registry<PragmaHandler> PragmaHandlerRegistry;
1971 } // end namespace clang