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/LLVM.h"
22 #include "clang/Basic/LangOptions.h"
23 #include "clang/Basic/Module.h"
24 #include "clang/Basic/SourceLocation.h"
25 #include "clang/Basic/SourceManager.h"
26 #include "clang/Basic/TokenKinds.h"
27 #include "clang/Lex/Lexer.h"
28 #include "clang/Lex/MacroInfo.h"
29 #include "clang/Lex/ModuleLoader.h"
30 #include "clang/Lex/ModuleMap.h"
31 #include "clang/Lex/PPCallbacks.h"
32 #include "clang/Lex/PTHLexer.h"
33 #include "clang/Lex/Token.h"
34 #include "clang/Lex/TokenLexer.h"
35 #include "llvm/ADT/ArrayRef.h"
36 #include "llvm/ADT/DenseMap.h"
37 #include "llvm/ADT/FoldingSet.h"
38 #include "llvm/ADT/None.h"
39 #include "llvm/ADT/Optional.h"
40 #include "llvm/ADT/PointerUnion.h"
41 #include "llvm/ADT/STLExtras.h"
42 #include "llvm/ADT/SmallPtrSet.h"
43 #include "llvm/ADT/SmallVector.h"
44 #include "llvm/ADT/StringRef.h"
45 #include "llvm/ADT/TinyPtrVector.h"
46 #include "llvm/ADT/iterator_range.h"
47 #include "llvm/Support/Allocator.h"
48 #include "llvm/Support/Casting.h"
49 #include "llvm/Support/Registry.h"
61 template<unsigned InternalLen> class SmallString;
67 class CodeCompletionHandler;
70 class DirectoryLookup;
71 class ExternalPreprocessorSource;
76 class MemoryBufferCache;
78 class PragmaNamespace;
79 class PreprocessingRecord;
80 class PreprocessorLexer;
81 class PreprocessorOptions;
86 /// \brief Stores token information for comparing actual tokens with
87 /// predefined values. Only handles simple tokens and identifiers.
93 TokenValue(tok::TokenKind Kind) : Kind(Kind), II(nullptr) {
94 assert(Kind != tok::raw_identifier && "Raw identifiers are not supported.");
95 assert(Kind != tok::identifier &&
96 "Identifiers should be created by TokenValue(IdentifierInfo *)");
97 assert(!tok::isLiteral(Kind) && "Literals are not supported.");
98 assert(!tok::isAnnotation(Kind) && "Annotations are not supported.");
101 TokenValue(IdentifierInfo *II) : Kind(tok::identifier), II(II) {}
103 bool operator==(const Token &Tok) const {
104 return Tok.getKind() == Kind &&
105 (!II || II == Tok.getIdentifierInfo());
109 /// \brief Context in which macro name is used.
111 // other than #define or #undef
114 // macro name specified in #define
117 // macro name specified in #undef
121 /// \brief Engages in a tight little dance with the lexer to efficiently
122 /// preprocess tokens.
124 /// Lexers know only about tokens within a single source file, and don't
125 /// know anything about preprocessor-level issues like the \#include stack,
126 /// token expansion, etc.
128 friend class VAOptDefinitionContext;
129 friend class VariadicMacroScopeGuard;
131 std::shared_ptr<PreprocessorOptions> PPOpts;
132 DiagnosticsEngine *Diags;
133 LangOptions &LangOpts;
134 const TargetInfo *Target = nullptr;
135 const TargetInfo *AuxTarget = nullptr;
136 FileManager &FileMgr;
137 SourceManager &SourceMgr;
138 MemoryBufferCache &PCMCache;
139 std::unique_ptr<ScratchBuffer> ScratchBuf;
140 HeaderSearch &HeaderInfo;
141 ModuleLoader &TheModuleLoader;
143 /// \brief External source of macros.
144 ExternalPreprocessorSource *ExternalSource;
146 /// An optional PTHManager object used for getting tokens from
147 /// a token cache rather than lexing the original source file.
148 std::unique_ptr<PTHManager> PTH;
150 /// A BumpPtrAllocator object used to quickly allocate and release
151 /// objects internal to the Preprocessor.
152 llvm::BumpPtrAllocator BP;
154 /// Identifiers for builtin macros and other builtins.
155 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__
156 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__
157 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__
158 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__
159 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__
160 IdentifierInfo *Ident__COUNTER__; // __COUNTER__
161 IdentifierInfo *Ident_Pragma, *Ident__pragma; // _Pragma, __pragma
162 IdentifierInfo *Ident__identifier; // __identifier
163 IdentifierInfo *Ident__VA_ARGS__; // __VA_ARGS__
164 IdentifierInfo *Ident__VA_OPT__; // __VA_OPT__
165 IdentifierInfo *Ident__has_feature; // __has_feature
166 IdentifierInfo *Ident__has_extension; // __has_extension
167 IdentifierInfo *Ident__has_builtin; // __has_builtin
168 IdentifierInfo *Ident__has_attribute; // __has_attribute
169 IdentifierInfo *Ident__has_include; // __has_include
170 IdentifierInfo *Ident__has_include_next; // __has_include_next
171 IdentifierInfo *Ident__has_warning; // __has_warning
172 IdentifierInfo *Ident__is_identifier; // __is_identifier
173 IdentifierInfo *Ident__building_module; // __building_module
174 IdentifierInfo *Ident__MODULE__; // __MODULE__
175 IdentifierInfo *Ident__has_cpp_attribute; // __has_cpp_attribute
176 IdentifierInfo *Ident__has_c_attribute; // __has_c_attribute
177 IdentifierInfo *Ident__has_declspec; // __has_declspec_attribute
178 IdentifierInfo *Ident__is_target_arch; // __is_target_arch
179 IdentifierInfo *Ident__is_target_vendor; // __is_target_vendor
180 IdentifierInfo *Ident__is_target_os; // __is_target_os
181 IdentifierInfo *Ident__is_target_environment; // __is_target_environment
183 SourceLocation DATELoc, TIMELoc;
185 // Next __COUNTER__ value, starts at 0.
186 unsigned CounterValue = 0;
189 /// \brief Maximum depth of \#includes.
190 MaxAllowedIncludeStackDepth = 200
193 // State that is set before the preprocessor begins.
194 bool KeepComments : 1;
195 bool KeepMacroComments : 1;
196 bool SuppressIncludeNotFoundError : 1;
198 // State that changes while the preprocessor runs:
199 bool InMacroArgs : 1; // True if parsing fn macro invocation args.
201 /// Whether the preprocessor owns the header search object.
202 bool OwnsHeaderSearch : 1;
204 /// True if macro expansion is disabled.
205 bool DisableMacroExpansion : 1;
207 /// Temporarily disables DisableMacroExpansion (i.e. enables expansion)
208 /// when parsing preprocessor directives.
209 bool MacroExpansionInDirectivesOverride : 1;
211 class ResetMacroExpansionHelper;
213 /// \brief Whether we have already loaded macros from the external source.
214 mutable bool ReadMacrosFromExternalSource : 1;
216 /// \brief True if pragmas are enabled.
217 bool PragmasEnabled : 1;
219 /// \brief True if the current build action is a preprocessing action.
220 bool PreprocessedOutput : 1;
222 /// \brief True if we are currently preprocessing a #if or #elif directive
223 bool ParsingIfOrElifDirective;
225 /// \brief True if we are pre-expanding macro arguments.
226 bool InMacroArgPreExpansion;
228 /// \brief Mapping/lookup information for all identifiers in
229 /// the program, including program keywords.
230 mutable IdentifierTable Identifiers;
232 /// \brief This table contains all the selectors in the program.
234 /// Unlike IdentifierTable above, this table *isn't* populated by the
235 /// preprocessor. It is declared/expanded here because its role/lifetime is
236 /// conceptually similar to the IdentifierTable. In addition, the current
237 /// control flow (in clang::ParseAST()), make it convenient to put here.
239 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to
240 /// the lifetime of the preprocessor.
241 SelectorTable Selectors;
243 /// \brief Information about builtins.
244 Builtin::Context BuiltinInfo;
246 /// \brief Tracks all of the pragmas that the client registered
247 /// with this preprocessor.
248 std::unique_ptr<PragmaNamespace> PragmaHandlers;
250 /// \brief Pragma handlers of the original source is stored here during the
251 /// parsing of a model file.
252 std::unique_ptr<PragmaNamespace> PragmaHandlersBackup;
254 /// \brief Tracks all of the comment handlers that the client registered
255 /// with this preprocessor.
256 std::vector<CommentHandler *> CommentHandlers;
258 /// \brief True if we want to ignore EOF token and continue later on (thus
259 /// avoid tearing the Lexer and etc. down).
260 bool IncrementalProcessing = false;
262 /// The kind of translation unit we are processing.
263 TranslationUnitKind TUKind;
265 /// \brief The code-completion handler.
266 CodeCompletionHandler *CodeComplete = nullptr;
268 /// \brief The file that we're performing code-completion for, if any.
269 const FileEntry *CodeCompletionFile = nullptr;
271 /// \brief The offset in file for the code-completion point.
272 unsigned CodeCompletionOffset = 0;
274 /// \brief The location for the code-completion point. This gets instantiated
275 /// when the CodeCompletionFile gets \#include'ed for preprocessing.
276 SourceLocation CodeCompletionLoc;
278 /// \brief The start location for the file of the code-completion point.
280 /// This gets instantiated when the CodeCompletionFile gets \#include'ed
281 /// for preprocessing.
282 SourceLocation CodeCompletionFileLoc;
284 /// \brief The source location of the \c import contextual keyword we just
286 SourceLocation ModuleImportLoc;
288 /// \brief The module import path that we're currently processing.
289 SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> ModuleImportPath;
291 /// \brief Whether the last token we lexed was an '@'.
292 bool LastTokenWasAt = false;
294 /// \brief Whether the module import expects an identifier next. Otherwise,
295 /// it expects a '.' or ';'.
296 bool ModuleImportExpectsIdentifier = false;
298 /// \brief The source location of the currently-active
299 /// \#pragma clang arc_cf_code_audited begin.
300 SourceLocation PragmaARCCFCodeAuditedLoc;
302 /// \brief The source location of the currently-active
303 /// \#pragma clang assume_nonnull begin.
304 SourceLocation PragmaAssumeNonNullLoc;
306 /// \brief True if we hit the code-completion point.
307 bool CodeCompletionReached = false;
309 /// \brief The code completion token containing the information
310 /// on the stem that is to be code completed.
311 IdentifierInfo *CodeCompletionII = nullptr;
313 /// \brief The directory that the main file should be considered to occupy,
314 /// if it does not correspond to a real file (as happens when building a
316 const DirectoryEntry *MainFileDir = nullptr;
318 /// \brief The number of bytes that we will initially skip when entering the
319 /// main file, along with a flag that indicates whether skipping this number
320 /// of bytes will place the lexer at the start of a line.
322 /// This is used when loading a precompiled preamble.
323 std::pair<int, bool> SkipMainFilePreamble;
326 struct PreambleSkipInfo {
327 SourceLocation HashTokenLoc;
328 SourceLocation IfTokenLoc;
329 bool FoundNonSkipPortion;
331 SourceLocation ElseLoc;
333 PreambleSkipInfo(SourceLocation HashTokenLoc, SourceLocation IfTokenLoc,
334 bool FoundNonSkipPortion, bool FoundElse,
335 SourceLocation ElseLoc)
336 : HashTokenLoc(HashTokenLoc), IfTokenLoc(IfTokenLoc),
337 FoundNonSkipPortion(FoundNonSkipPortion), FoundElse(FoundElse),
342 friend class ASTReader;
343 friend class MacroArgs;
345 class PreambleConditionalStackStore {
353 PreambleConditionalStackStore() = default;
355 void startRecording() { ConditionalStackState = Recording; }
356 void startReplaying() { ConditionalStackState = Replaying; }
357 bool isRecording() const { return ConditionalStackState == Recording; }
358 bool isReplaying() const { return ConditionalStackState == Replaying; }
360 ArrayRef<PPConditionalInfo> getStack() const {
361 return ConditionalStack;
364 void doneReplaying() {
365 ConditionalStack.clear();
366 ConditionalStackState = Off;
369 void setStack(ArrayRef<PPConditionalInfo> s) {
370 if (!isRecording() && !isReplaying())
372 ConditionalStack.clear();
373 ConditionalStack.append(s.begin(), s.end());
376 bool hasRecordedPreamble() const { return !ConditionalStack.empty(); }
378 bool reachedEOFWhileSkipping() const { return SkipInfo.hasValue(); }
380 void clearSkipInfo() { SkipInfo.reset(); }
382 llvm::Optional<PreambleSkipInfo> SkipInfo;
385 SmallVector<PPConditionalInfo, 4> ConditionalStack;
386 State ConditionalStackState = Off;
387 } PreambleConditionalStack;
389 /// \brief The current top of the stack that we're lexing from if
390 /// not expanding a macro and we are lexing directly from source code.
392 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null.
393 std::unique_ptr<Lexer> CurLexer;
395 /// \brief The current top of stack that we're lexing from if
396 /// not expanding from a macro and we are lexing from a PTH cache.
398 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null.
399 std::unique_ptr<PTHLexer> CurPTHLexer;
401 /// \brief The current top of the stack what we're lexing from
402 /// if not expanding a macro.
404 /// This is an alias for either CurLexer or CurPTHLexer.
405 PreprocessorLexer *CurPPLexer = nullptr;
407 /// \brief Used to find the current FileEntry, if CurLexer is non-null
408 /// and if applicable.
410 /// This allows us to implement \#include_next and find directory-specific
412 const DirectoryLookup *CurDirLookup = nullptr;
414 /// \brief The current macro we are expanding, if we are expanding a macro.
416 /// One of CurLexer and CurTokenLexer must be null.
417 std::unique_ptr<TokenLexer> CurTokenLexer;
419 /// \brief The kind of lexer we're currently working with.
425 CLK_LexAfterModuleImport
426 } CurLexerKind = CLK_Lexer;
428 /// \brief If the current lexer is for a submodule that is being built, this
429 /// is that submodule.
430 Module *CurLexerSubmodule = nullptr;
432 /// \brief Keeps track of the stack of files currently
433 /// \#included, and macros currently being expanded from, not counting
434 /// CurLexer/CurTokenLexer.
435 struct IncludeStackInfo {
436 enum CurLexerKind CurLexerKind;
437 Module *TheSubmodule;
438 std::unique_ptr<Lexer> TheLexer;
439 std::unique_ptr<PTHLexer> ThePTHLexer;
440 PreprocessorLexer *ThePPLexer;
441 std::unique_ptr<TokenLexer> TheTokenLexer;
442 const DirectoryLookup *TheDirLookup;
444 // The following constructors are completely useless copies of the default
445 // versions, only needed to pacify MSVC.
446 IncludeStackInfo(enum CurLexerKind CurLexerKind, Module *TheSubmodule,
447 std::unique_ptr<Lexer> &&TheLexer,
448 std::unique_ptr<PTHLexer> &&ThePTHLexer,
449 PreprocessorLexer *ThePPLexer,
450 std::unique_ptr<TokenLexer> &&TheTokenLexer,
451 const DirectoryLookup *TheDirLookup)
452 : CurLexerKind(std::move(CurLexerKind)),
453 TheSubmodule(std::move(TheSubmodule)), TheLexer(std::move(TheLexer)),
454 ThePTHLexer(std::move(ThePTHLexer)),
455 ThePPLexer(std::move(ThePPLexer)),
456 TheTokenLexer(std::move(TheTokenLexer)),
457 TheDirLookup(std::move(TheDirLookup)) {}
459 std::vector<IncludeStackInfo> IncludeMacroStack;
461 /// \brief Actions invoked when some preprocessor activity is
462 /// encountered (e.g. a file is \#included, etc).
463 std::unique_ptr<PPCallbacks> Callbacks;
465 struct MacroExpandsInfo {
470 MacroExpandsInfo(Token Tok, MacroDefinition MD, SourceRange Range)
471 : Tok(Tok), MD(MD), Range(Range) {}
473 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks;
475 /// Information about a name that has been used to define a module macro.
476 struct ModuleMacroInfo {
477 /// The most recent macro directive for this identifier.
480 /// The active module macros for this identifier.
481 llvm::TinyPtrVector<ModuleMacro *> ActiveModuleMacros;
483 /// The generation number at which we last updated ActiveModuleMacros.
484 /// \see Preprocessor::VisibleModules.
485 unsigned ActiveModuleMacrosGeneration = 0;
487 /// Whether this macro name is ambiguous.
488 bool IsAmbiguous = false;
490 /// The module macros that are overridden by this macro.
491 llvm::TinyPtrVector<ModuleMacro *> OverriddenMacros;
493 ModuleMacroInfo(MacroDirective *MD) : MD(MD) {}
496 /// The state of a macro for an identifier.
498 mutable llvm::PointerUnion<MacroDirective *, ModuleMacroInfo *> State;
500 ModuleMacroInfo *getModuleInfo(Preprocessor &PP,
501 const IdentifierInfo *II) const {
502 if (II->isOutOfDate())
503 PP.updateOutOfDateIdentifier(const_cast<IdentifierInfo&>(*II));
504 // FIXME: Find a spare bit on IdentifierInfo and store a
505 // HasModuleMacros flag.
506 if (!II->hasMacroDefinition() ||
507 (!PP.getLangOpts().Modules &&
508 !PP.getLangOpts().ModulesLocalVisibility) ||
509 !PP.CurSubmoduleState->VisibleModules.getGeneration())
512 auto *Info = State.dyn_cast<ModuleMacroInfo*>();
514 Info = new (PP.getPreprocessorAllocator())
515 ModuleMacroInfo(State.get<MacroDirective *>());
519 if (PP.CurSubmoduleState->VisibleModules.getGeneration() !=
520 Info->ActiveModuleMacrosGeneration)
521 PP.updateModuleMacroInfo(II, *Info);
526 MacroState() : MacroState(nullptr) {}
527 MacroState(MacroDirective *MD) : State(MD) {}
529 MacroState(MacroState &&O) noexcept : State(O.State) {
530 O.State = (MacroDirective *)nullptr;
533 MacroState &operator=(MacroState &&O) noexcept {
535 O.State = (MacroDirective *)nullptr;
541 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
542 Info->~ModuleMacroInfo();
545 MacroDirective *getLatest() const {
546 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
548 return State.get<MacroDirective*>();
551 void setLatest(MacroDirective *MD) {
552 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
558 bool isAmbiguous(Preprocessor &PP, const IdentifierInfo *II) const {
559 auto *Info = getModuleInfo(PP, II);
560 return Info ? Info->IsAmbiguous : false;
563 ArrayRef<ModuleMacro *>
564 getActiveModuleMacros(Preprocessor &PP, const IdentifierInfo *II) const {
565 if (auto *Info = getModuleInfo(PP, II))
566 return Info->ActiveModuleMacros;
570 MacroDirective::DefInfo findDirectiveAtLoc(SourceLocation Loc,
571 SourceManager &SourceMgr) const {
572 // FIXME: Incorporate module macros into the result of this.
573 if (auto *Latest = getLatest())
574 return Latest->findDirectiveAtLoc(Loc, SourceMgr);
578 void overrideActiveModuleMacros(Preprocessor &PP, IdentifierInfo *II) {
579 if (auto *Info = getModuleInfo(PP, II)) {
580 Info->OverriddenMacros.insert(Info->OverriddenMacros.end(),
581 Info->ActiveModuleMacros.begin(),
582 Info->ActiveModuleMacros.end());
583 Info->ActiveModuleMacros.clear();
584 Info->IsAmbiguous = false;
588 ArrayRef<ModuleMacro*> getOverriddenMacros() const {
589 if (auto *Info = State.dyn_cast<ModuleMacroInfo*>())
590 return Info->OverriddenMacros;
594 void setOverriddenMacros(Preprocessor &PP,
595 ArrayRef<ModuleMacro *> Overrides) {
596 auto *Info = State.dyn_cast<ModuleMacroInfo*>();
598 if (Overrides.empty())
600 Info = new (PP.getPreprocessorAllocator())
601 ModuleMacroInfo(State.get<MacroDirective *>());
604 Info->OverriddenMacros.clear();
605 Info->OverriddenMacros.insert(Info->OverriddenMacros.end(),
606 Overrides.begin(), Overrides.end());
607 Info->ActiveModuleMacrosGeneration = 0;
611 /// For each IdentifierInfo that was associated with a macro, we
612 /// keep a mapping to the history of all macro definitions and #undefs in
613 /// the reverse order (the latest one is in the head of the list).
615 /// This mapping lives within the \p CurSubmoduleState.
616 using MacroMap = llvm::DenseMap<const IdentifierInfo *, MacroState>;
618 struct SubmoduleState;
620 /// \brief Information about a submodule that we're currently building.
621 struct BuildingSubmoduleInfo {
622 /// The module that we are building.
625 /// The location at which the module was included.
626 SourceLocation ImportLoc;
628 /// Whether we entered this submodule via a pragma.
631 /// The previous SubmoduleState.
632 SubmoduleState *OuterSubmoduleState;
634 /// The number of pending module macro names when we started building this.
635 unsigned OuterPendingModuleMacroNames;
637 BuildingSubmoduleInfo(Module *M, SourceLocation ImportLoc, bool IsPragma,
638 SubmoduleState *OuterSubmoduleState,
639 unsigned OuterPendingModuleMacroNames)
640 : M(M), ImportLoc(ImportLoc), IsPragma(IsPragma),
641 OuterSubmoduleState(OuterSubmoduleState),
642 OuterPendingModuleMacroNames(OuterPendingModuleMacroNames) {}
644 SmallVector<BuildingSubmoduleInfo, 8> BuildingSubmoduleStack;
646 /// \brief Information about a submodule's preprocessor state.
647 struct SubmoduleState {
648 /// The macros for the submodule.
651 /// The set of modules that are visible within the submodule.
652 VisibleModuleSet VisibleModules;
654 // FIXME: CounterValue?
655 // FIXME: PragmaPushMacroInfo?
657 std::map<Module *, SubmoduleState> Submodules;
659 /// The preprocessor state for preprocessing outside of any submodule.
660 SubmoduleState NullSubmoduleState;
662 /// The current submodule state. Will be \p NullSubmoduleState if we're not
664 SubmoduleState *CurSubmoduleState;
666 /// The set of known macros exported from modules.
667 llvm::FoldingSet<ModuleMacro> ModuleMacros;
669 /// The names of potential module macros that we've not yet processed.
670 llvm::SmallVector<const IdentifierInfo *, 32> PendingModuleMacroNames;
672 /// The list of module macros, for each identifier, that are not overridden by
673 /// any other module macro.
674 llvm::DenseMap<const IdentifierInfo *, llvm::TinyPtrVector<ModuleMacro *>>
677 /// \brief Macros that we want to warn because they are not used at the end
678 /// of the translation unit.
680 /// We store just their SourceLocations instead of
681 /// something like MacroInfo*. The benefit of this is that when we are
682 /// deserializing from PCH, we don't need to deserialize identifier & macros
683 /// just so that we can report that they are unused, we just warn using
684 /// the SourceLocations of this set (that will be filled by the ASTReader).
685 /// We are using SmallPtrSet instead of a vector for faster removal.
686 using WarnUnusedMacroLocsTy = llvm::SmallPtrSet<SourceLocation, 32>;
687 WarnUnusedMacroLocsTy WarnUnusedMacroLocs;
689 /// \brief A "freelist" of MacroArg objects that can be
690 /// reused for quick allocation.
691 MacroArgs *MacroArgCache = nullptr;
693 /// For each IdentifierInfo used in a \#pragma push_macro directive,
694 /// we keep a MacroInfo stack used to restore the previous macro value.
695 llvm::DenseMap<IdentifierInfo *, std::vector<MacroInfo *>>
698 // Various statistics we track for performance analysis.
699 unsigned NumDirectives = 0;
700 unsigned NumDefined = 0;
701 unsigned NumUndefined = 0;
702 unsigned NumPragma = 0;
704 unsigned NumElse = 0;
705 unsigned NumEndif = 0;
706 unsigned NumEnteredSourceFiles = 0;
707 unsigned MaxIncludeStackDepth = 0;
708 unsigned NumMacroExpanded = 0;
709 unsigned NumFnMacroExpanded = 0;
710 unsigned NumBuiltinMacroExpanded = 0;
711 unsigned NumFastMacroExpanded = 0;
712 unsigned NumTokenPaste = 0;
713 unsigned NumFastTokenPaste = 0;
714 unsigned NumSkipped = 0;
716 /// \brief The predefined macros that preprocessor should use from the
717 /// command line etc.
718 std::string Predefines;
720 /// \brief The file ID for the preprocessor predefines.
721 FileID PredefinesFileID;
724 /// \brief Cache of macro expanders to reduce malloc traffic.
725 enum { TokenLexerCacheSize = 8 };
726 unsigned NumCachedTokenLexers;
727 std::unique_ptr<TokenLexer> TokenLexerCache[TokenLexerCacheSize];
730 /// \brief Keeps macro expanded tokens for TokenLexers.
732 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
733 /// going to lex in the cache and when it finishes the tokens are removed
734 /// from the end of the cache.
735 SmallVector<Token, 16> MacroExpandedTokens;
736 std::vector<std::pair<TokenLexer *, size_t>> MacroExpandingLexersStack;
738 /// \brief A record of the macro definitions and expansions that
739 /// occurred during preprocessing.
741 /// This is an optional side structure that can be enabled with
742 /// \c createPreprocessingRecord() prior to preprocessing.
743 PreprocessingRecord *Record = nullptr;
745 /// Cached tokens state.
746 using CachedTokensTy = SmallVector<Token, 1>;
748 /// \brief Cached tokens are stored here when we do backtracking or
749 /// lookahead. They are "lexed" by the CachingLex() method.
750 CachedTokensTy CachedTokens;
752 /// \brief The position of the cached token that CachingLex() should
755 /// If it points beyond the CachedTokens vector, it means that a normal
756 /// Lex() should be invoked.
757 CachedTokensTy::size_type CachedLexPos = 0;
759 /// \brief Stack of backtrack positions, allowing nested backtracks.
761 /// The EnableBacktrackAtThisPos() method pushes a position to
762 /// indicate where CachedLexPos should be set when the BackTrack() method is
763 /// invoked (at which point the last position is popped).
764 std::vector<CachedTokensTy::size_type> BacktrackPositions;
766 struct MacroInfoChain {
768 MacroInfoChain *Next;
771 /// MacroInfos are managed as a chain for easy disposal. This is the head
773 MacroInfoChain *MIChainHead = nullptr;
775 void updateOutOfDateIdentifier(IdentifierInfo &II) const;
778 Preprocessor(std::shared_ptr<PreprocessorOptions> PPOpts,
779 DiagnosticsEngine &diags, LangOptions &opts, SourceManager &SM,
780 MemoryBufferCache &PCMCache,
781 HeaderSearch &Headers, ModuleLoader &TheModuleLoader,
782 IdentifierInfoLookup *IILookup = nullptr,
783 bool OwnsHeaderSearch = false,
784 TranslationUnitKind TUKind = TU_Complete);
788 /// \brief Initialize the preprocessor using information about the target.
790 /// \param Target is owned by the caller and must remain valid for the
791 /// lifetime of the preprocessor.
792 /// \param AuxTarget is owned by the caller and must remain valid for
793 /// the lifetime of the preprocessor.
794 void Initialize(const TargetInfo &Target,
795 const TargetInfo *AuxTarget = nullptr);
797 /// \brief Initialize the preprocessor to parse a model file
799 /// To parse model files the preprocessor of the original source is reused to
800 /// preserver the identifier table. However to avoid some duplicate
801 /// information in the preprocessor some cleanup is needed before it is used
802 /// to parse model files. This method does that cleanup.
803 void InitializeForModelFile();
805 /// \brief Cleanup after model file parsing
806 void FinalizeForModelFile();
808 /// \brief Retrieve the preprocessor options used to initialize this
810 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; }
812 DiagnosticsEngine &getDiagnostics() const { return *Diags; }
813 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; }
815 const LangOptions &getLangOpts() const { return LangOpts; }
816 const TargetInfo &getTargetInfo() const { return *Target; }
817 const TargetInfo *getAuxTargetInfo() const { return AuxTarget; }
818 FileManager &getFileManager() const { return FileMgr; }
819 SourceManager &getSourceManager() const { return SourceMgr; }
820 MemoryBufferCache &getPCMCache() const { return PCMCache; }
821 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; }
823 IdentifierTable &getIdentifierTable() { return Identifiers; }
824 const IdentifierTable &getIdentifierTable() const { return Identifiers; }
825 SelectorTable &getSelectorTable() { return Selectors; }
826 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; }
827 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; }
829 void setPTHManager(PTHManager* pm);
831 PTHManager *getPTHManager() { return PTH.get(); }
833 void setExternalSource(ExternalPreprocessorSource *Source) {
834 ExternalSource = Source;
837 ExternalPreprocessorSource *getExternalSource() const {
838 return ExternalSource;
841 /// \brief Retrieve the module loader associated with this preprocessor.
842 ModuleLoader &getModuleLoader() const { return TheModuleLoader; }
844 bool hadModuleLoaderFatalFailure() const {
845 return TheModuleLoader.HadFatalFailure;
848 /// \brief True if we are currently preprocessing a #if or #elif directive
849 bool isParsingIfOrElifDirective() const {
850 return ParsingIfOrElifDirective;
853 /// \brief Control whether the preprocessor retains comments in output.
854 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) {
855 this->KeepComments = KeepComments | KeepMacroComments;
856 this->KeepMacroComments = KeepMacroComments;
859 bool getCommentRetentionState() const { return KeepComments; }
861 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; }
862 bool getPragmasEnabled() const { return PragmasEnabled; }
864 void SetSuppressIncludeNotFoundError(bool Suppress) {
865 SuppressIncludeNotFoundError = Suppress;
868 bool GetSuppressIncludeNotFoundError() {
869 return SuppressIncludeNotFoundError;
872 /// Sets whether the preprocessor is responsible for producing output or if
873 /// it is producing tokens to be consumed by Parse and Sema.
874 void setPreprocessedOutput(bool IsPreprocessedOutput) {
875 PreprocessedOutput = IsPreprocessedOutput;
878 /// Returns true if the preprocessor is responsible for generating output,
879 /// false if it is producing tokens to be consumed by Parse and Sema.
880 bool isPreprocessedOutput() const { return PreprocessedOutput; }
882 /// \brief Return true if we are lexing directly from the specified lexer.
883 bool isCurrentLexer(const PreprocessorLexer *L) const {
884 return CurPPLexer == L;
887 /// \brief Return the current lexer being lexed from.
889 /// Note that this ignores any potentially active macro expansions and _Pragma
890 /// expansions going on at the time.
891 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; }
893 /// \brief Return the current file lexer being lexed from.
895 /// Note that this ignores any potentially active macro expansions and _Pragma
896 /// expansions going on at the time.
897 PreprocessorLexer *getCurrentFileLexer() const;
899 /// \brief Return the submodule owning the file being lexed. This may not be
900 /// the current module if we have changed modules since entering the file.
901 Module *getCurrentLexerSubmodule() const { return CurLexerSubmodule; }
903 /// \brief Returns the FileID for the preprocessor predefines.
904 FileID getPredefinesFileID() const { return PredefinesFileID; }
907 /// \brief Accessors for preprocessor callbacks.
909 /// Note that this class takes ownership of any PPCallbacks object given to
911 PPCallbacks *getPPCallbacks() const { return Callbacks.get(); }
912 void addPPCallbacks(std::unique_ptr<PPCallbacks> C) {
914 C = llvm::make_unique<PPChainedCallbacks>(std::move(C),
915 std::move(Callbacks));
916 Callbacks = std::move(C);
920 bool isMacroDefined(StringRef Id) {
921 return isMacroDefined(&Identifiers.get(Id));
923 bool isMacroDefined(const IdentifierInfo *II) {
924 return II->hasMacroDefinition() &&
925 (!getLangOpts().Modules || (bool)getMacroDefinition(II));
928 /// \brief Determine whether II is defined as a macro within the module M,
929 /// if that is a module that we've already preprocessed. Does not check for
930 /// macros imported into M.
931 bool isMacroDefinedInLocalModule(const IdentifierInfo *II, Module *M) {
932 if (!II->hasMacroDefinition())
934 auto I = Submodules.find(M);
935 if (I == Submodules.end())
937 auto J = I->second.Macros.find(II);
938 if (J == I->second.Macros.end())
940 auto *MD = J->second.getLatest();
941 return MD && MD->isDefined();
944 MacroDefinition getMacroDefinition(const IdentifierInfo *II) {
945 if (!II->hasMacroDefinition())
948 MacroState &S = CurSubmoduleState->Macros[II];
949 auto *MD = S.getLatest();
950 while (MD && isa<VisibilityMacroDirective>(MD))
951 MD = MD->getPrevious();
952 return MacroDefinition(dyn_cast_or_null<DefMacroDirective>(MD),
953 S.getActiveModuleMacros(*this, II),
954 S.isAmbiguous(*this, II));
957 MacroDefinition getMacroDefinitionAtLoc(const IdentifierInfo *II,
958 SourceLocation Loc) {
959 if (!II->hadMacroDefinition())
962 MacroState &S = CurSubmoduleState->Macros[II];
963 MacroDirective::DefInfo DI;
964 if (auto *MD = S.getLatest())
965 DI = MD->findDirectiveAtLoc(Loc, getSourceManager());
966 // FIXME: Compute the set of active module macros at the specified location.
967 return MacroDefinition(DI.getDirective(),
968 S.getActiveModuleMacros(*this, II),
969 S.isAmbiguous(*this, II));
972 /// \brief Given an identifier, return its latest non-imported MacroDirective
973 /// if it is \#define'd and not \#undef'd, or null if it isn't \#define'd.
974 MacroDirective *getLocalMacroDirective(const IdentifierInfo *II) const {
975 if (!II->hasMacroDefinition())
978 auto *MD = getLocalMacroDirectiveHistory(II);
979 if (!MD || MD->getDefinition().isUndefined())
985 const MacroInfo *getMacroInfo(const IdentifierInfo *II) const {
986 return const_cast<Preprocessor*>(this)->getMacroInfo(II);
989 MacroInfo *getMacroInfo(const IdentifierInfo *II) {
990 if (!II->hasMacroDefinition())
992 if (auto MD = getMacroDefinition(II))
993 return MD.getMacroInfo();
997 /// \brief Given an identifier, return the latest non-imported macro
998 /// directive for that identifier.
1000 /// One can iterate over all previous macro directives from the most recent
1002 MacroDirective *getLocalMacroDirectiveHistory(const IdentifierInfo *II) const;
1004 /// \brief Add a directive to the macro directive history for this identifier.
1005 void appendMacroDirective(IdentifierInfo *II, MacroDirective *MD);
1006 DefMacroDirective *appendDefMacroDirective(IdentifierInfo *II, MacroInfo *MI,
1007 SourceLocation Loc) {
1008 DefMacroDirective *MD = AllocateDefMacroDirective(MI, Loc);
1009 appendMacroDirective(II, MD);
1012 DefMacroDirective *appendDefMacroDirective(IdentifierInfo *II,
1014 return appendDefMacroDirective(II, MI, MI->getDefinitionLoc());
1017 /// \brief Set a MacroDirective that was loaded from a PCH file.
1018 void setLoadedMacroDirective(IdentifierInfo *II, MacroDirective *ED,
1019 MacroDirective *MD);
1021 /// \brief Register an exported macro for a module and identifier.
1022 ModuleMacro *addModuleMacro(Module *Mod, IdentifierInfo *II, MacroInfo *Macro,
1023 ArrayRef<ModuleMacro *> Overrides, bool &IsNew);
1024 ModuleMacro *getModuleMacro(Module *Mod, IdentifierInfo *II);
1026 /// \brief Get the list of leaf (non-overridden) module macros for a name.
1027 ArrayRef<ModuleMacro*> getLeafModuleMacros(const IdentifierInfo *II) const {
1028 if (II->isOutOfDate())
1029 updateOutOfDateIdentifier(const_cast<IdentifierInfo&>(*II));
1030 auto I = LeafModuleMacros.find(II);
1031 if (I != LeafModuleMacros.end())
1037 /// Iterators for the macro history table. Currently defined macros have
1038 /// IdentifierInfo::hasMacroDefinition() set and an empty
1039 /// MacroInfo::getUndefLoc() at the head of the list.
1040 using macro_iterator = MacroMap::const_iterator;
1042 macro_iterator macro_begin(bool IncludeExternalMacros = true) const;
1043 macro_iterator macro_end(bool IncludeExternalMacros = true) const;
1044 llvm::iterator_range<macro_iterator>
1046 macros(bool IncludeExternalMacros = true) const {
1047 return llvm::make_range(macro_begin(IncludeExternalMacros),
1048 macro_end(IncludeExternalMacros));
1052 /// \brief Return the name of the macro defined before \p Loc that has
1053 /// spelling \p Tokens. If there are multiple macros with same spelling,
1054 /// return the last one defined.
1055 StringRef getLastMacroWithSpelling(SourceLocation Loc,
1056 ArrayRef<TokenValue> Tokens) const;
1058 const std::string &getPredefines() const { return Predefines; }
1060 /// \brief Set the predefines for this Preprocessor.
1062 /// These predefines are automatically injected when parsing the main file.
1063 void setPredefines(const char *P) { Predefines = P; }
1064 void setPredefines(StringRef P) { Predefines = P; }
1066 /// Return information about the specified preprocessor
1067 /// identifier token.
1068 IdentifierInfo *getIdentifierInfo(StringRef Name) const {
1069 return &Identifiers.get(Name);
1072 /// \brief Add the specified pragma handler to this preprocessor.
1074 /// If \p Namespace is non-null, then it is a token required to exist on the
1075 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC".
1076 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler);
1077 void AddPragmaHandler(PragmaHandler *Handler) {
1078 AddPragmaHandler(StringRef(), Handler);
1081 /// \brief Remove the specific pragma handler from this preprocessor.
1083 /// If \p Namespace is non-null, then it should be the namespace that
1084 /// \p Handler was added to. It is an error to remove a handler that
1085 /// has not been registered.
1086 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler);
1087 void RemovePragmaHandler(PragmaHandler *Handler) {
1088 RemovePragmaHandler(StringRef(), Handler);
1091 /// Install empty handlers for all pragmas (making them ignored).
1092 void IgnorePragmas();
1094 /// \brief Add the specified comment handler to the preprocessor.
1095 void addCommentHandler(CommentHandler *Handler);
1097 /// \brief Remove the specified comment handler.
1099 /// It is an error to remove a handler that has not been registered.
1100 void removeCommentHandler(CommentHandler *Handler);
1102 /// \brief Set the code completion handler to the given object.
1103 void setCodeCompletionHandler(CodeCompletionHandler &Handler) {
1104 CodeComplete = &Handler;
1107 /// \brief Retrieve the current code-completion handler.
1108 CodeCompletionHandler *getCodeCompletionHandler() const {
1109 return CodeComplete;
1112 /// \brief Clear out the code completion handler.
1113 void clearCodeCompletionHandler() {
1114 CodeComplete = nullptr;
1117 /// \brief Hook used by the lexer to invoke the "natural language" code
1118 /// completion point.
1119 void CodeCompleteNaturalLanguage();
1121 /// \brief Set the code completion token for filtering purposes.
1122 void setCodeCompletionIdentifierInfo(IdentifierInfo *Filter) {
1123 CodeCompletionII = Filter;
1126 /// \brief Get the code completion token for filtering purposes.
1127 StringRef getCodeCompletionFilter() {
1128 if (CodeCompletionII)
1129 return CodeCompletionII->getName();
1133 /// \brief Retrieve the preprocessing record, or NULL if there is no
1134 /// preprocessing record.
1135 PreprocessingRecord *getPreprocessingRecord() const { return Record; }
1137 /// \brief Create a new preprocessing record, which will keep track of
1138 /// all macro expansions, macro definitions, etc.
1139 void createPreprocessingRecord();
1141 /// \brief Enter the specified FileID as the main source file,
1142 /// which implicitly adds the builtin defines etc.
1143 void EnterMainSourceFile();
1145 /// \brief Inform the preprocessor callbacks that processing is complete.
1146 void EndSourceFile();
1148 /// \brief Add a source file to the top of the include stack and
1149 /// start lexing tokens from it instead of the current buffer.
1151 /// Emits a diagnostic, doesn't enter the file, and returns true on error.
1152 bool EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir,
1153 SourceLocation Loc);
1155 /// \brief Add a Macro to the top of the include stack and start lexing
1156 /// tokens from it instead of the current buffer.
1158 /// \param Args specifies the tokens input to a function-like macro.
1159 /// \param ILEnd specifies the location of the ')' for a function-like macro
1160 /// or the identifier for an object-like macro.
1161 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro,
1164 /// \brief Add a "macro" context to the top of the include stack,
1165 /// which will cause the lexer to start returning the specified tokens.
1167 /// If \p DisableMacroExpansion is true, tokens lexed from the token stream
1168 /// will not be subject to further macro expansion. Otherwise, these tokens
1169 /// will be re-macro-expanded when/if expansion is enabled.
1171 /// If \p OwnsTokens is false, this method assumes that the specified stream
1172 /// of tokens has a permanent owner somewhere, so they do not need to be
1173 /// copied. If it is true, it assumes the array of tokens is allocated with
1174 /// \c new[] and the Preprocessor will delete[] it.
1176 void EnterTokenStream(const Token *Toks, unsigned NumToks,
1177 bool DisableMacroExpansion, bool OwnsTokens);
1180 void EnterTokenStream(std::unique_ptr<Token[]> Toks, unsigned NumToks,
1181 bool DisableMacroExpansion) {
1182 EnterTokenStream(Toks.release(), NumToks, DisableMacroExpansion, true);
1185 void EnterTokenStream(ArrayRef<Token> Toks, bool DisableMacroExpansion) {
1186 EnterTokenStream(Toks.data(), Toks.size(), DisableMacroExpansion, false);
1189 /// \brief Pop the current lexer/macro exp off the top of the lexer stack.
1191 /// This should only be used in situations where the current state of the
1192 /// top-of-stack lexer is known.
1193 void RemoveTopOfLexerStack();
1195 /// From the point that this method is called, and until
1196 /// CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor
1197 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will
1198 /// make the Preprocessor re-lex the same tokens.
1200 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can
1201 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will
1202 /// be combined with the EnableBacktrackAtThisPos calls in reverse order.
1204 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack
1205 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of
1206 /// tokens will continue indefinitely.
1208 void EnableBacktrackAtThisPos();
1210 /// \brief Disable the last EnableBacktrackAtThisPos call.
1211 void CommitBacktrackedTokens();
1213 struct CachedTokensRange {
1214 CachedTokensTy::size_type Begin, End;
1218 /// \brief A range of cached tokens that should be erased after lexing
1219 /// when backtracking requires the erasure of such cached tokens.
1220 Optional<CachedTokensRange> CachedTokenRangeToErase;
1223 /// \brief Returns the range of cached tokens that were lexed since
1224 /// EnableBacktrackAtThisPos() was previously called.
1225 CachedTokensRange LastCachedTokenRange();
1227 /// \brief Erase the range of cached tokens that were lexed since
1228 /// EnableBacktrackAtThisPos() was previously called.
1229 void EraseCachedTokens(CachedTokensRange TokenRange);
1231 /// \brief Make Preprocessor re-lex the tokens that were lexed since
1232 /// EnableBacktrackAtThisPos() was previously called.
1235 /// \brief True if EnableBacktrackAtThisPos() was called and
1236 /// caching of tokens is on.
1237 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); }
1239 /// \brief Lex the next token for this preprocessor.
1240 void Lex(Token &Result);
1242 void LexAfterModuleImport(Token &Result);
1244 void makeModuleVisible(Module *M, SourceLocation Loc);
1246 SourceLocation getModuleImportLoc(Module *M) const {
1247 return CurSubmoduleState->VisibleModules.getImportLoc(M);
1250 /// \brief Lex a string literal, which may be the concatenation of multiple
1251 /// string literals and may even come from macro expansion.
1252 /// \returns true on success, false if a error diagnostic has been generated.
1253 bool LexStringLiteral(Token &Result, std::string &String,
1254 const char *DiagnosticTag, bool AllowMacroExpansion) {
1255 if (AllowMacroExpansion)
1258 LexUnexpandedToken(Result);
1259 return FinishLexStringLiteral(Result, String, DiagnosticTag,
1260 AllowMacroExpansion);
1263 /// \brief Complete the lexing of a string literal where the first token has
1264 /// already been lexed (see LexStringLiteral).
1265 bool FinishLexStringLiteral(Token &Result, std::string &String,
1266 const char *DiagnosticTag,
1267 bool AllowMacroExpansion);
1269 /// \brief Lex a token. If it's a comment, keep lexing until we get
1270 /// something not a comment.
1272 /// This is useful in -E -C mode where comments would foul up preprocessor
1273 /// directive handling.
1274 void LexNonComment(Token &Result) {
1277 while (Result.getKind() == tok::comment);
1280 /// \brief Just like Lex, but disables macro expansion of identifier tokens.
1281 void LexUnexpandedToken(Token &Result) {
1282 // Disable macro expansion.
1283 bool OldVal = DisableMacroExpansion;
1284 DisableMacroExpansion = true;
1289 DisableMacroExpansion = OldVal;
1292 /// \brief Like LexNonComment, but this disables macro expansion of
1293 /// identifier tokens.
1294 void LexUnexpandedNonComment(Token &Result) {
1296 LexUnexpandedToken(Result);
1297 while (Result.getKind() == tok::comment);
1300 /// \brief Parses a simple integer literal to get its numeric value. Floating
1301 /// point literals and user defined literals are rejected. Used primarily to
1302 /// handle pragmas that accept integer arguments.
1303 bool parseSimpleIntegerLiteral(Token &Tok, uint64_t &Value);
1305 /// Disables macro expansion everywhere except for preprocessor directives.
1306 void SetMacroExpansionOnlyInDirectives() {
1307 DisableMacroExpansion = true;
1308 MacroExpansionInDirectivesOverride = true;
1311 /// \brief Peeks ahead N tokens and returns that token without consuming any
1314 /// LookAhead(0) returns the next token that would be returned by Lex(),
1315 /// LookAhead(1) returns the token after it, etc. This returns normal
1316 /// tokens after phase 5. As such, it is equivalent to using
1317 /// 'Lex', not 'LexUnexpandedToken'.
1318 const Token &LookAhead(unsigned N) {
1319 if (CachedLexPos + N < CachedTokens.size())
1320 return CachedTokens[CachedLexPos+N];
1322 return PeekAhead(N+1);
1325 /// \brief When backtracking is enabled and tokens are cached,
1326 /// this allows to revert a specific number of tokens.
1328 /// Note that the number of tokens being reverted should be up to the last
1329 /// backtrack position, not more.
1330 void RevertCachedTokens(unsigned N) {
1331 assert(isBacktrackEnabled() &&
1332 "Should only be called when tokens are cached for backtracking");
1333 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back())
1334 && "Should revert tokens up to the last backtrack position, not more");
1335 assert(signed(CachedLexPos) - signed(N) >= 0 &&
1336 "Corrupted backtrack positions ?");
1340 /// \brief Enters a token in the token stream to be lexed next.
1342 /// If BackTrack() is called afterwards, the token will remain at the
1343 /// insertion point.
1344 void EnterToken(const Token &Tok) {
1345 EnterCachingLexMode();
1346 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok);
1349 /// We notify the Preprocessor that if it is caching tokens (because
1350 /// backtrack is enabled) it should replace the most recent cached tokens
1351 /// with the given annotation token. This function has no effect if
1352 /// backtracking is not enabled.
1354 /// Note that the use of this function is just for optimization, so that the
1355 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is
1357 void AnnotateCachedTokens(const Token &Tok) {
1358 assert(Tok.isAnnotation() && "Expected annotation token");
1359 if (CachedLexPos != 0 && isBacktrackEnabled())
1360 AnnotatePreviousCachedTokens(Tok);
1363 /// Get the location of the last cached token, suitable for setting the end
1364 /// location of an annotation token.
1365 SourceLocation getLastCachedTokenLocation() const {
1366 assert(CachedLexPos != 0);
1367 return CachedTokens[CachedLexPos-1].getLastLoc();
1370 /// \brief Whether \p Tok is the most recent token (`CachedLexPos - 1`) in
1372 bool IsPreviousCachedToken(const Token &Tok) const;
1374 /// \brief Replace token in `CachedLexPos - 1` in CachedTokens by the tokens
1377 /// Useful when a token needs to be split in smaller ones and CachedTokens
1378 /// most recent token must to be updated to reflect that.
1379 void ReplacePreviousCachedToken(ArrayRef<Token> NewToks);
1381 /// \brief Replace the last token with an annotation token.
1383 /// Like AnnotateCachedTokens(), this routine replaces an
1384 /// already-parsed (and resolved) token with an annotation
1385 /// token. However, this routine only replaces the last token with
1386 /// the annotation token; it does not affect any other cached
1387 /// tokens. This function has no effect if backtracking is not
1389 void ReplaceLastTokenWithAnnotation(const Token &Tok) {
1390 assert(Tok.isAnnotation() && "Expected annotation token");
1391 if (CachedLexPos != 0 && isBacktrackEnabled())
1392 CachedTokens[CachedLexPos-1] = Tok;
1395 /// Enter an annotation token into the token stream.
1396 void EnterAnnotationToken(SourceRange Range, tok::TokenKind Kind,
1397 void *AnnotationVal);
1399 /// Update the current token to represent the provided
1400 /// identifier, in order to cache an action performed by typo correction.
1401 void TypoCorrectToken(const Token &Tok) {
1402 assert(Tok.getIdentifierInfo() && "Expected identifier token");
1403 if (CachedLexPos != 0 && isBacktrackEnabled())
1404 CachedTokens[CachedLexPos-1] = Tok;
1407 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/
1408 /// CurTokenLexer pointers.
1409 void recomputeCurLexerKind();
1411 /// \brief Returns true if incremental processing is enabled
1412 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; }
1414 /// \brief Enables the incremental processing
1415 void enableIncrementalProcessing(bool value = true) {
1416 IncrementalProcessing = value;
1419 /// \brief Specify the point at which code-completion will be performed.
1421 /// \param File the file in which code completion should occur. If
1422 /// this file is included multiple times, code-completion will
1423 /// perform completion the first time it is included. If NULL, this
1424 /// function clears out the code-completion point.
1426 /// \param Line the line at which code completion should occur
1429 /// \param Column the column at which code completion should occur
1432 /// \returns true if an error occurred, false otherwise.
1433 bool SetCodeCompletionPoint(const FileEntry *File,
1434 unsigned Line, unsigned Column);
1436 /// \brief Determine if we are performing code completion.
1437 bool isCodeCompletionEnabled() const { return CodeCompletionFile != nullptr; }
1439 /// \brief Returns the location of the code-completion point.
1441 /// Returns an invalid location if code-completion is not enabled or the file
1442 /// containing the code-completion point has not been lexed yet.
1443 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; }
1445 /// \brief Returns the start location of the file of code-completion point.
1447 /// Returns an invalid location if code-completion is not enabled or the file
1448 /// containing the code-completion point has not been lexed yet.
1449 SourceLocation getCodeCompletionFileLoc() const {
1450 return CodeCompletionFileLoc;
1453 /// \brief Returns true if code-completion is enabled and we have hit the
1454 /// code-completion point.
1455 bool isCodeCompletionReached() const { return CodeCompletionReached; }
1457 /// \brief Note that we hit the code-completion point.
1458 void setCodeCompletionReached() {
1459 assert(isCodeCompletionEnabled() && "Code-completion not enabled!");
1460 CodeCompletionReached = true;
1461 // Silence any diagnostics that occur after we hit the code-completion.
1462 getDiagnostics().setSuppressAllDiagnostics(true);
1465 /// \brief The location of the currently-active \#pragma clang
1466 /// arc_cf_code_audited begin.
1468 /// Returns an invalid location if there is no such pragma active.
1469 SourceLocation getPragmaARCCFCodeAuditedLoc() const {
1470 return PragmaARCCFCodeAuditedLoc;
1473 /// \brief Set the location of the currently-active \#pragma clang
1474 /// arc_cf_code_audited begin. An invalid location ends the pragma.
1475 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) {
1476 PragmaARCCFCodeAuditedLoc = Loc;
1479 /// \brief The location of the currently-active \#pragma clang
1480 /// assume_nonnull begin.
1482 /// Returns an invalid location if there is no such pragma active.
1483 SourceLocation getPragmaAssumeNonNullLoc() const {
1484 return PragmaAssumeNonNullLoc;
1487 /// \brief Set the location of the currently-active \#pragma clang
1488 /// assume_nonnull begin. An invalid location ends the pragma.
1489 void setPragmaAssumeNonNullLoc(SourceLocation Loc) {
1490 PragmaAssumeNonNullLoc = Loc;
1493 /// \brief Set the directory in which the main file should be considered
1494 /// to have been found, if it is not a real file.
1495 void setMainFileDir(const DirectoryEntry *Dir) {
1499 /// \brief Instruct the preprocessor to skip part of the main source file.
1501 /// \param Bytes The number of bytes in the preamble to skip.
1503 /// \param StartOfLine Whether skipping these bytes puts the lexer at the
1504 /// start of a line.
1505 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) {
1506 SkipMainFilePreamble.first = Bytes;
1507 SkipMainFilePreamble.second = StartOfLine;
1510 /// Forwarding function for diagnostics. This emits a diagnostic at
1511 /// the specified Token's location, translating the token's start
1512 /// position in the current buffer into a SourcePosition object for rendering.
1513 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const {
1514 return Diags->Report(Loc, DiagID);
1517 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const {
1518 return Diags->Report(Tok.getLocation(), DiagID);
1521 /// Return the 'spelling' of the token at the given
1522 /// location; does not go up to the spelling location or down to the
1523 /// expansion location.
1525 /// \param buffer A buffer which will be used only if the token requires
1526 /// "cleaning", e.g. if it contains trigraphs or escaped newlines
1527 /// \param invalid If non-null, will be set \c true if an error occurs.
1528 StringRef getSpelling(SourceLocation loc,
1529 SmallVectorImpl<char> &buffer,
1530 bool *invalid = nullptr) const {
1531 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid);
1534 /// \brief Return the 'spelling' of the Tok token.
1536 /// The spelling of a token is the characters used to represent the token in
1537 /// the source file after trigraph expansion and escaped-newline folding. In
1538 /// particular, this wants to get the true, uncanonicalized, spelling of
1539 /// things like digraphs, UCNs, etc.
1541 /// \param Invalid If non-null, will be set \c true if an error occurs.
1542 std::string getSpelling(const Token &Tok, bool *Invalid = nullptr) const {
1543 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid);
1546 /// \brief Get the spelling of a token into a preallocated buffer, instead
1547 /// of as an std::string.
1549 /// The caller is required to allocate enough space for the token, which is
1550 /// guaranteed to be at least Tok.getLength() bytes long. The length of the
1551 /// actual result is returned.
1553 /// Note that this method may do two possible things: it may either fill in
1554 /// the buffer specified with characters, or it may *change the input pointer*
1555 /// to point to a constant buffer with the data already in it (avoiding a
1556 /// copy). The caller is not allowed to modify the returned buffer pointer
1557 /// if an internal buffer is returned.
1558 unsigned getSpelling(const Token &Tok, const char *&Buffer,
1559 bool *Invalid = nullptr) const {
1560 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid);
1563 /// \brief Get the spelling of a token into a SmallVector.
1565 /// Note that the returned StringRef may not point to the
1566 /// supplied buffer if a copy can be avoided.
1567 StringRef getSpelling(const Token &Tok,
1568 SmallVectorImpl<char> &Buffer,
1569 bool *Invalid = nullptr) const;
1571 /// \brief Relex the token at the specified location.
1572 /// \returns true if there was a failure, false on success.
1573 bool getRawToken(SourceLocation Loc, Token &Result,
1574 bool IgnoreWhiteSpace = false) {
1575 return Lexer::getRawToken(Loc, Result, SourceMgr, LangOpts, IgnoreWhiteSpace);
1578 /// \brief Given a Token \p Tok that is a numeric constant with length 1,
1579 /// return the character.
1581 getSpellingOfSingleCharacterNumericConstant(const Token &Tok,
1582 bool *Invalid = nullptr) const {
1583 assert(Tok.is(tok::numeric_constant) &&
1584 Tok.getLength() == 1 && "Called on unsupported token");
1585 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1");
1587 // If the token is carrying a literal data pointer, just use it.
1588 if (const char *D = Tok.getLiteralData())
1591 // Otherwise, fall back on getCharacterData, which is slower, but always
1593 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid);
1596 /// \brief Retrieve the name of the immediate macro expansion.
1598 /// This routine starts from a source location, and finds the name of the
1599 /// macro responsible for its immediate expansion. It looks through any
1600 /// intervening macro argument expansions to compute this. It returns a
1601 /// StringRef that refers to the SourceManager-owned buffer of the source
1602 /// where that macro name is spelled. Thus, the result shouldn't out-live
1603 /// the SourceManager.
1604 StringRef getImmediateMacroName(SourceLocation Loc) {
1605 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts());
1608 /// \brief Plop the specified string into a scratch buffer and set the
1609 /// specified token's location and length to it.
1611 /// If specified, the source location provides a location of the expansion
1612 /// point of the token.
1613 void CreateString(StringRef Str, Token &Tok,
1614 SourceLocation ExpansionLocStart = SourceLocation(),
1615 SourceLocation ExpansionLocEnd = SourceLocation());
1617 /// \brief Computes the source location just past the end of the
1618 /// token at this source location.
1620 /// This routine can be used to produce a source location that
1621 /// points just past the end of the token referenced by \p Loc, and
1622 /// is generally used when a diagnostic needs to point just after a
1623 /// token where it expected something different that it received. If
1624 /// the returned source location would not be meaningful (e.g., if
1625 /// it points into a macro), this routine returns an invalid
1626 /// source location.
1628 /// \param Offset an offset from the end of the token, where the source
1629 /// location should refer to. The default offset (0) produces a source
1630 /// location pointing just past the end of the token; an offset of 1 produces
1631 /// a source location pointing to the last character in the token, etc.
1632 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) {
1633 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts);
1636 /// \brief Returns true if the given MacroID location points at the first
1637 /// token of the macro expansion.
1639 /// \param MacroBegin If non-null and function returns true, it is set to
1640 /// begin location of the macro.
1641 bool isAtStartOfMacroExpansion(SourceLocation loc,
1642 SourceLocation *MacroBegin = nullptr) const {
1643 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts,
1647 /// \brief Returns true if the given MacroID location points at the last
1648 /// token of the macro expansion.
1650 /// \param MacroEnd If non-null and function returns true, it is set to
1651 /// end location of the macro.
1652 bool isAtEndOfMacroExpansion(SourceLocation loc,
1653 SourceLocation *MacroEnd = nullptr) const {
1654 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd);
1657 /// \brief Print the token to stderr, used for debugging.
1658 void DumpToken(const Token &Tok, bool DumpFlags = false) const;
1659 void DumpLocation(SourceLocation Loc) const;
1660 void DumpMacro(const MacroInfo &MI) const;
1661 void dumpMacroInfo(const IdentifierInfo *II);
1663 /// \brief Given a location that specifies the start of a
1664 /// token, return a new location that specifies a character within the token.
1665 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
1666 unsigned Char) const {
1667 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts);
1670 /// \brief Increment the counters for the number of token paste operations
1673 /// If fast was specified, this is a 'fast paste' case we handled.
1674 void IncrementPasteCounter(bool isFast) {
1676 ++NumFastTokenPaste;
1683 size_t getTotalMemory() const;
1685 /// When the macro expander pastes together a comment (/##/) in Microsoft
1686 /// mode, this method handles updating the current state, returning the
1687 /// token on the next source line.
1688 void HandleMicrosoftCommentPaste(Token &Tok);
1690 //===--------------------------------------------------------------------===//
1691 // Preprocessor callback methods. These are invoked by a lexer as various
1692 // directives and events are found.
1694 /// Given a tok::raw_identifier token, look up the
1695 /// identifier information for the token and install it into the token,
1696 /// updating the token kind accordingly.
1697 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const;
1700 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons;
1703 /// \brief Specifies the reason for poisoning an identifier.
1705 /// If that identifier is accessed while poisoned, then this reason will be
1706 /// used instead of the default "poisoned" diagnostic.
1707 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID);
1709 /// \brief Display reason for poisoned identifier.
1710 void HandlePoisonedIdentifier(Token & Tok);
1712 void MaybeHandlePoisonedIdentifier(Token & Identifier) {
1713 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) {
1714 if(II->isPoisoned()) {
1715 HandlePoisonedIdentifier(Identifier);
1721 /// Identifiers used for SEH handling in Borland. These are only
1722 /// allowed in particular circumstances
1724 IdentifierInfo *Ident__exception_code,
1725 *Ident___exception_code,
1726 *Ident_GetExceptionCode;
1727 // __except filter expression
1728 IdentifierInfo *Ident__exception_info,
1729 *Ident___exception_info,
1730 *Ident_GetExceptionInfo;
1732 IdentifierInfo *Ident__abnormal_termination,
1733 *Ident___abnormal_termination,
1734 *Ident_AbnormalTermination;
1736 const char *getCurLexerEndPos();
1737 void diagnoseMissingHeaderInUmbrellaDir(const Module &Mod);
1740 void PoisonSEHIdentifiers(bool Poison = true); // Borland
1742 /// \brief Callback invoked when the lexer reads an identifier and has
1743 /// filled in the tokens IdentifierInfo member.
1745 /// This callback potentially macro expands it or turns it into a named
1746 /// token (like 'for').
1748 /// \returns true if we actually computed a token, false if we need to
1750 bool HandleIdentifier(Token &Identifier);
1752 /// \brief Callback invoked when the lexer hits the end of the current file.
1754 /// This either returns the EOF token and returns true, or
1755 /// pops a level off the include stack and returns false, at which point the
1756 /// client should call lex again.
1757 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false);
1759 /// \brief Callback invoked when the current TokenLexer hits the end of its
1761 bool HandleEndOfTokenLexer(Token &Result);
1763 /// \brief Callback invoked when the lexer sees a # token at the start of a
1766 /// This consumes the directive, modifies the lexer/preprocessor state, and
1767 /// advances the lexer(s) so that the next token read is the correct one.
1768 void HandleDirective(Token &Result);
1770 /// \brief Ensure that the next token is a tok::eod token.
1772 /// If not, emit a diagnostic and consume up until the eod.
1773 /// If \p EnableMacros is true, then we consider macros that expand to zero
1774 /// tokens as being ok.
1775 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false);
1777 /// \brief Read and discard all tokens remaining on the current line until
1778 /// the tok::eod token is found.
1779 void DiscardUntilEndOfDirective();
1781 /// \brief Returns true if the preprocessor has seen a use of
1782 /// __DATE__ or __TIME__ in the file so far.
1783 bool SawDateOrTime() const {
1784 return DATELoc != SourceLocation() || TIMELoc != SourceLocation();
1786 unsigned getCounterValue() const { return CounterValue; }
1787 void setCounterValue(unsigned V) { CounterValue = V; }
1789 /// \brief Retrieves the module that we're currently building, if any.
1790 Module *getCurrentModule();
1792 /// \brief Allocate a new MacroInfo object with the provided SourceLocation.
1793 MacroInfo *AllocateMacroInfo(SourceLocation L);
1795 /// \brief Turn the specified lexer token into a fully checked and spelled
1796 /// filename, e.g. as an operand of \#include.
1798 /// The caller is expected to provide a buffer that is large enough to hold
1799 /// the spelling of the filename, but is also expected to handle the case
1800 /// when this method decides to use a different buffer.
1802 /// \returns true if the input filename was in <>'s or false if it was
1804 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename);
1806 /// \brief Given a "foo" or \<foo> reference, look up the indicated file.
1808 /// Returns null on failure. \p isAngled indicates whether the file
1809 /// reference is for system \#include's or not (i.e. using <> instead of "").
1810 const FileEntry *LookupFile(SourceLocation FilenameLoc, StringRef Filename,
1811 bool isAngled, const DirectoryLookup *FromDir,
1812 const FileEntry *FromFile,
1813 const DirectoryLookup *&CurDir,
1814 SmallVectorImpl<char> *SearchPath,
1815 SmallVectorImpl<char> *RelativePath,
1816 ModuleMap::KnownHeader *SuggestedModule,
1817 bool *IsMapped, bool SkipCache = false);
1819 /// \brief Get the DirectoryLookup structure used to find the current
1820 /// FileEntry, if CurLexer is non-null and if applicable.
1822 /// This allows us to implement \#include_next and find directory-specific
1824 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; }
1826 /// \brief Return true if we're in the top-level file, not in a \#include.
1827 bool isInPrimaryFile() const;
1829 /// \brief Handle cases where the \#include name is expanded
1830 /// from a macro as multiple tokens, which need to be glued together.
1832 /// This occurs for code like:
1834 /// \#define FOO <x/y.h>
1837 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one.
1839 /// This code concatenates and consumes tokens up to the '>' token. It
1840 /// returns false if the > was found, otherwise it returns true if it finds
1841 /// and consumes the EOD marker.
1842 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer,
1843 SourceLocation &End);
1845 /// \brief Lex an on-off-switch (C99 6.10.6p2) and verify that it is
1846 /// followed by EOD. Return true if the token is not a valid on-off-switch.
1847 bool LexOnOffSwitch(tok::OnOffSwitch &OOS);
1849 bool CheckMacroName(Token &MacroNameTok, MacroUse isDefineUndef,
1850 bool *ShadowFlag = nullptr);
1852 void EnterSubmodule(Module *M, SourceLocation ImportLoc, bool ForPragma);
1853 Module *LeaveSubmodule(bool ForPragma);
1856 friend void TokenLexer::ExpandFunctionArguments();
1858 void PushIncludeMacroStack() {
1859 assert(CurLexerKind != CLK_CachingLexer && "cannot push a caching lexer");
1860 IncludeMacroStack.emplace_back(CurLexerKind, CurLexerSubmodule,
1861 std::move(CurLexer), std::move(CurPTHLexer),
1862 CurPPLexer, std::move(CurTokenLexer),
1864 CurPPLexer = nullptr;
1867 void PopIncludeMacroStack() {
1868 CurLexer = std::move(IncludeMacroStack.back().TheLexer);
1869 CurPTHLexer = std::move(IncludeMacroStack.back().ThePTHLexer);
1870 CurPPLexer = IncludeMacroStack.back().ThePPLexer;
1871 CurTokenLexer = std::move(IncludeMacroStack.back().TheTokenLexer);
1872 CurDirLookup = IncludeMacroStack.back().TheDirLookup;
1873 CurLexerSubmodule = IncludeMacroStack.back().TheSubmodule;
1874 CurLexerKind = IncludeMacroStack.back().CurLexerKind;
1875 IncludeMacroStack.pop_back();
1878 void PropagateLineStartLeadingSpaceInfo(Token &Result);
1880 /// Determine whether we need to create module macros for #defines in the
1881 /// current context.
1882 bool needModuleMacros() const;
1884 /// Update the set of active module macros and ambiguity flag for a module
1886 void updateModuleMacroInfo(const IdentifierInfo *II, ModuleMacroInfo &Info);
1888 DefMacroDirective *AllocateDefMacroDirective(MacroInfo *MI,
1889 SourceLocation Loc);
1890 UndefMacroDirective *AllocateUndefMacroDirective(SourceLocation UndefLoc);
1891 VisibilityMacroDirective *AllocateVisibilityMacroDirective(SourceLocation Loc,
1894 /// \brief Lex and validate a macro name, which occurs after a
1895 /// \#define or \#undef.
1897 /// \param MacroNameTok Token that represents the name defined or undefined.
1898 /// \param IsDefineUndef Kind if preprocessor directive.
1899 /// \param ShadowFlag Points to flag that is set if macro name shadows
1902 /// This emits a diagnostic, sets the token kind to eod,
1903 /// and discards the rest of the macro line if the macro name is invalid.
1904 void ReadMacroName(Token &MacroNameTok, MacroUse IsDefineUndef = MU_Other,
1905 bool *ShadowFlag = nullptr);
1907 /// ReadOptionalMacroParameterListAndBody - This consumes all (i.e. the
1908 /// entire line) of the macro's tokens and adds them to MacroInfo, and while
1909 /// doing so performs certain validity checks including (but not limited to):
1910 /// - # (stringization) is followed by a macro parameter
1911 /// \param MacroNameTok - Token that represents the macro name
1912 /// \param ImmediatelyAfterHeaderGuard - Macro follows an #ifdef header guard
1914 /// Either returns a pointer to a MacroInfo object OR emits a diagnostic and
1915 /// returns a nullptr if an invalid sequence of tokens is encountered.
1916 MacroInfo *ReadOptionalMacroParameterListAndBody(
1917 const Token &MacroNameTok, bool ImmediatelyAfterHeaderGuard);
1919 /// The ( starting an argument list of a macro definition has just been read.
1920 /// Lex the rest of the parameters and the closing ), updating \p MI with
1921 /// what we learn and saving in \p LastTok the last token read.
1922 /// Return true if an error occurs parsing the arg list.
1923 bool ReadMacroParameterList(MacroInfo *MI, Token& LastTok);
1925 /// We just read a \#if or related directive and decided that the
1926 /// subsequent tokens are in the \#if'd out portion of the
1927 /// file. Lex the rest of the file, until we see an \#endif. If \p
1928 /// FoundNonSkipPortion is true, then we have already emitted code for part of
1929 /// this \#if directive, so \#else/\#elif blocks should never be entered. If
1930 /// \p FoundElse is false, then \#else directives are ok, if not, then we have
1931 /// already seen one so a \#else directive is a duplicate. When this returns,
1932 /// the caller can lex the first valid token.
1933 void SkipExcludedConditionalBlock(SourceLocation HashTokenLoc,
1934 SourceLocation IfTokenLoc,
1935 bool FoundNonSkipPortion, bool FoundElse,
1936 SourceLocation ElseLoc = SourceLocation());
1938 /// \brief A fast PTH version of SkipExcludedConditionalBlock.
1939 void PTHSkipExcludedConditionalBlock();
1941 /// Information about the result for evaluating an expression for a
1942 /// preprocessor directive.
1943 struct DirectiveEvalResult {
1944 /// Whether the expression was evaluated as true or not.
1947 /// True if the expression contained identifiers that were undefined.
1948 bool IncludedUndefinedIds;
1951 /// \brief Evaluate an integer constant expression that may occur after a
1952 /// \#if or \#elif directive and return a \p DirectiveEvalResult object.
1954 /// If the expression is equivalent to "!defined(X)" return X in IfNDefMacro.
1955 DirectiveEvalResult EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro);
1957 /// \brief Install the standard preprocessor pragmas:
1958 /// \#pragma GCC poison/system_header/dependency and \#pragma once.
1959 void RegisterBuiltinPragmas();
1961 /// \brief Register builtin macros such as __LINE__ with the identifier table.
1962 void RegisterBuiltinMacros();
1964 /// If an identifier token is read that is to be expanded as a macro, handle
1965 /// it and return the next token as 'Tok'. If we lexed a token, return true;
1966 /// otherwise the caller should lex again.
1967 bool HandleMacroExpandedIdentifier(Token &Tok, const MacroDefinition &MD);
1969 /// \brief Cache macro expanded tokens for TokenLexers.
1971 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
1972 /// going to lex in the cache and when it finishes the tokens are removed
1973 /// from the end of the cache.
1974 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer,
1975 ArrayRef<Token> tokens);
1977 void removeCachedMacroExpandedTokensOfLastLexer();
1979 /// Determine whether the next preprocessor token to be
1980 /// lexed is a '('. If so, consume the token and return true, if not, this
1981 /// method should have no observable side-effect on the lexed tokens.
1982 bool isNextPPTokenLParen();
1984 /// After reading "MACRO(", this method is invoked to read all of the formal
1985 /// arguments specified for the macro invocation. Returns null on error.
1986 MacroArgs *ReadMacroCallArgumentList(Token &MacroName, MacroInfo *MI,
1987 SourceLocation &ExpansionEnd);
1989 /// \brief If an identifier token is read that is to be expanded
1990 /// as a builtin macro, handle it and return the next token as 'Tok'.
1991 void ExpandBuiltinMacro(Token &Tok);
1993 /// \brief Read a \c _Pragma directive, slice it up, process it, then
1994 /// return the first token after the directive.
1995 /// This assumes that the \c _Pragma token has just been read into \p Tok.
1996 void Handle_Pragma(Token &Tok);
1998 /// \brief Like Handle_Pragma except the pragma text is not enclosed within
1999 /// a string literal.
2000 void HandleMicrosoft__pragma(Token &Tok);
2002 /// \brief Add a lexer to the top of the include stack and
2003 /// start lexing tokens from it instead of the current buffer.
2004 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir);
2006 /// \brief Add a lexer to the top of the include stack and
2007 /// start getting tokens from it using the PTH cache.
2008 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir);
2010 /// \brief Set the FileID for the preprocessor predefines.
2011 void setPredefinesFileID(FileID FID) {
2012 assert(PredefinesFileID.isInvalid() && "PredefinesFileID already set!");
2013 PredefinesFileID = FID;
2016 /// \brief Returns true if we are lexing from a file and not a
2017 /// pragma or a macro.
2018 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) {
2019 return L ? !L->isPragmaLexer() : P != nullptr;
2022 static bool IsFileLexer(const IncludeStackInfo& I) {
2023 return IsFileLexer(I.TheLexer.get(), I.ThePPLexer);
2026 bool IsFileLexer() const {
2027 return IsFileLexer(CurLexer.get(), CurPPLexer);
2030 //===--------------------------------------------------------------------===//
2032 void CachingLex(Token &Result);
2034 bool InCachingLexMode() const {
2035 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means
2036 // that we are past EOF, not that we are in CachingLex mode.
2037 return !CurPPLexer && !CurTokenLexer && !CurPTHLexer &&
2038 !IncludeMacroStack.empty();
2041 void EnterCachingLexMode();
2043 void ExitCachingLexMode() {
2044 if (InCachingLexMode())
2045 RemoveTopOfLexerStack();
2048 const Token &PeekAhead(unsigned N);
2049 void AnnotatePreviousCachedTokens(const Token &Tok);
2051 //===--------------------------------------------------------------------===//
2052 /// Handle*Directive - implement the various preprocessor directives. These
2053 /// should side-effect the current preprocessor object so that the next call
2054 /// to Lex() will return the appropriate token next.
2055 void HandleLineDirective();
2056 void HandleDigitDirective(Token &Tok);
2057 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning);
2058 void HandleIdentSCCSDirective(Token &Tok);
2059 void HandleMacroPublicDirective(Token &Tok);
2060 void HandleMacroPrivateDirective();
2063 void HandleIncludeDirective(SourceLocation HashLoc,
2065 const DirectoryLookup *LookupFrom = nullptr,
2066 const FileEntry *LookupFromFile = nullptr,
2067 bool isImport = false);
2068 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok);
2069 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok);
2070 void HandleImportDirective(SourceLocation HashLoc, Token &Tok);
2071 void HandleMicrosoftImportDirective(Token &Tok);
2074 /// Check that the given module is available, producing a diagnostic if not.
2075 /// \return \c true if the check failed (because the module is not available).
2076 /// \c false if the module appears to be usable.
2077 static bool checkModuleIsAvailable(const LangOptions &LangOpts,
2078 const TargetInfo &TargetInfo,
2079 DiagnosticsEngine &Diags, Module *M);
2081 // Module inclusion testing.
2082 /// \brief Find the module that owns the source or header file that
2083 /// \p Loc points to. If the location is in a file that was included
2084 /// into a module, or is outside any module, returns nullptr.
2085 Module *getModuleForLocation(SourceLocation Loc);
2087 /// \brief We want to produce a diagnostic at location IncLoc concerning a
2088 /// missing module import.
2090 /// \param IncLoc The location at which the missing import was detected.
2091 /// \param M The desired module.
2092 /// \param MLoc A location within the desired module at which some desired
2093 /// effect occurred (eg, where a desired entity was declared).
2095 /// \return A file that can be #included to import a module containing MLoc.
2096 /// Null if no such file could be determined or if a #include is not
2098 const FileEntry *getModuleHeaderToIncludeForDiagnostics(SourceLocation IncLoc,
2100 SourceLocation MLoc);
2102 bool isRecordingPreamble() const {
2103 return PreambleConditionalStack.isRecording();
2106 bool hasRecordedPreamble() const {
2107 return PreambleConditionalStack.hasRecordedPreamble();
2110 ArrayRef<PPConditionalInfo> getPreambleConditionalStack() const {
2111 return PreambleConditionalStack.getStack();
2114 void setRecordedPreambleConditionalStack(ArrayRef<PPConditionalInfo> s) {
2115 PreambleConditionalStack.setStack(s);
2118 void setReplayablePreambleConditionalStack(ArrayRef<PPConditionalInfo> s,
2119 llvm::Optional<PreambleSkipInfo> SkipInfo) {
2120 PreambleConditionalStack.startReplaying();
2121 PreambleConditionalStack.setStack(s);
2122 PreambleConditionalStack.SkipInfo = SkipInfo;
2125 llvm::Optional<PreambleSkipInfo> getPreambleSkipInfo() const {
2126 return PreambleConditionalStack.SkipInfo;
2130 /// \brief After processing predefined file, initialize the conditional stack from
2132 void replayPreambleConditionalStack();
2135 void HandleDefineDirective(Token &Tok, bool ImmediatelyAfterTopLevelIfndef);
2136 void HandleUndefDirective();
2138 // Conditional Inclusion.
2139 void HandleIfdefDirective(Token &Tok, const Token &HashToken,
2140 bool isIfndef, bool ReadAnyTokensBeforeDirective);
2141 void HandleIfDirective(Token &Tok, const Token &HashToken,
2142 bool ReadAnyTokensBeforeDirective);
2143 void HandleEndifDirective(Token &Tok);
2144 void HandleElseDirective(Token &Tok, const Token &HashToken);
2145 void HandleElifDirective(Token &Tok, const Token &HashToken);
2148 void HandlePragmaDirective(SourceLocation IntroducerLoc,
2149 PragmaIntroducerKind Introducer);
2152 void HandlePragmaOnce(Token &OnceTok);
2153 void HandlePragmaMark();
2154 void HandlePragmaPoison();
2155 void HandlePragmaSystemHeader(Token &SysHeaderTok);
2156 void HandlePragmaDependency(Token &DependencyTok);
2157 void HandlePragmaPushMacro(Token &Tok);
2158 void HandlePragmaPopMacro(Token &Tok);
2159 void HandlePragmaIncludeAlias(Token &Tok);
2160 void HandlePragmaModuleBuild(Token &Tok);
2161 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok);
2163 // Return true and store the first token only if any CommentHandler
2164 // has inserted some tokens and getCommentRetentionState() is false.
2165 bool HandleComment(Token &Token, SourceRange Comment);
2167 /// \brief A macro is used, update information about macros that need unused
2169 void markMacroAsUsed(MacroInfo *MI);
2172 /// \brief Abstract base class that describes a handler that will receive
2173 /// source ranges for each of the comments encountered in the source file.
2174 class CommentHandler {
2176 virtual ~CommentHandler();
2178 // The handler shall return true if it has pushed any tokens
2179 // to be read using e.g. EnterToken or EnterTokenStream.
2180 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0;
2183 /// \brief Registry of pragma handlers added by plugins
2184 using PragmaHandlerRegistry = llvm::Registry<PragmaHandler>;
2186 } // namespace clang
2188 #endif // LLVM_CLANG_LEX_PREPROCESSOR_H