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 //===----------------------------------------------------------------------===//
10 // This file defines the Preprocessor interface.
12 //===----------------------------------------------------------------------===//
14 #ifndef LLVM_CLANG_LEX_PREPROCESSOR_H
15 #define LLVM_CLANG_LEX_PREPROCESSOR_H
17 #include "clang/Lex/MacroInfo.h"
18 #include "clang/Lex/Lexer.h"
19 #include "clang/Lex/PTHLexer.h"
20 #include "clang/Lex/PPCallbacks.h"
21 #include "clang/Lex/TokenLexer.h"
22 #include "clang/Lex/PTHManager.h"
23 #include "clang/Basic/Builtins.h"
24 #include "clang/Basic/Diagnostic.h"
25 #include "clang/Basic/IdentifierTable.h"
26 #include "clang/Basic/SourceLocation.h"
27 #include "llvm/ADT/DenseMap.h"
28 #include "llvm/ADT/IntrusiveRefCntPtr.h"
29 #include "llvm/ADT/SmallPtrSet.h"
30 #include "llvm/ADT/OwningPtr.h"
31 #include "llvm/ADT/SmallVector.h"
32 #include "llvm/ADT/ArrayRef.h"
33 #include "llvm/Support/Allocator.h"
37 template<unsigned InternalLen> class SmallString;
43 class ExternalPreprocessorSource;
47 class PragmaNamespace;
53 class CodeCompletionHandler;
54 class DirectoryLookup;
55 class PreprocessingRecord;
58 /// Preprocessor - This object engages in a tight little dance with the lexer to
59 /// efficiently preprocess tokens. Lexers know only about tokens within a
60 /// single source file, and don't know anything about preprocessor-level issues
61 /// like the #include stack, token expansion, etc.
63 class Preprocessor : public RefCountedBase<Preprocessor> {
64 DiagnosticsEngine *Diags;
65 LangOptions &LangOpts;
66 const TargetInfo *Target;
68 SourceManager &SourceMgr;
69 ScratchBuffer *ScratchBuf;
70 HeaderSearch &HeaderInfo;
71 ModuleLoader &TheModuleLoader;
73 /// \brief External source of macros.
74 ExternalPreprocessorSource *ExternalSource;
77 /// PTH - An optional PTHManager object used for getting tokens from
78 /// a token cache rather than lexing the original source file.
79 OwningPtr<PTHManager> PTH;
81 /// BP - A BumpPtrAllocator object used to quickly allocate and release
82 /// objects internal to the Preprocessor.
83 llvm::BumpPtrAllocator BP;
85 /// Identifiers for builtin macros and other builtins.
86 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__
87 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__
88 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__
89 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__
90 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__
91 IdentifierInfo *Ident__COUNTER__; // __COUNTER__
92 IdentifierInfo *Ident_Pragma, *Ident__pragma; // _Pragma, __pragma
93 IdentifierInfo *Ident__VA_ARGS__; // __VA_ARGS__
94 IdentifierInfo *Ident__has_feature; // __has_feature
95 IdentifierInfo *Ident__has_extension; // __has_extension
96 IdentifierInfo *Ident__has_builtin; // __has_builtin
97 IdentifierInfo *Ident__has_attribute; // __has_attribute
98 IdentifierInfo *Ident__has_include; // __has_include
99 IdentifierInfo *Ident__has_include_next; // __has_include_next
100 IdentifierInfo *Ident__has_warning; // __has_warning
102 SourceLocation DATELoc, TIMELoc;
103 unsigned CounterValue; // Next __COUNTER__ value.
106 /// MaxIncludeStackDepth - Maximum depth of #includes.
107 MaxAllowedIncludeStackDepth = 200
110 // State that is set before the preprocessor begins.
111 bool KeepComments : 1;
112 bool KeepMacroComments : 1;
113 bool SuppressIncludeNotFoundError : 1;
115 // State that changes while the preprocessor runs:
116 bool InMacroArgs : 1; // True if parsing fn macro invocation args.
118 /// Whether the preprocessor owns the header search object.
119 bool OwnsHeaderSearch : 1;
121 /// DisableMacroExpansion - True if macro expansion is disabled.
122 bool DisableMacroExpansion : 1;
124 /// \brief Whether we have already loaded macros from the external source.
125 mutable bool ReadMacrosFromExternalSource : 1;
127 /// \brief True if we are pre-expanding macro arguments.
128 bool InMacroArgPreExpansion;
130 /// Identifiers - This is mapping/lookup information for all identifiers in
131 /// the program, including program keywords.
132 mutable IdentifierTable Identifiers;
134 /// Selectors - This table contains all the selectors in the program. Unlike
135 /// IdentifierTable above, this table *isn't* populated by the preprocessor.
136 /// It is declared/expanded here because it's role/lifetime is
137 /// conceptually similar the IdentifierTable. In addition, the current control
138 /// flow (in clang::ParseAST()), make it convenient to put here.
139 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to
140 /// the lifetime of the preprocessor.
141 SelectorTable Selectors;
143 /// BuiltinInfo - Information about builtins.
144 Builtin::Context BuiltinInfo;
146 /// PragmaHandlers - This tracks all of the pragmas that the client registered
147 /// with this preprocessor.
148 PragmaNamespace *PragmaHandlers;
150 /// \brief Tracks all of the comment handlers that the client registered
151 /// with this preprocessor.
152 std::vector<CommentHandler *> CommentHandlers;
154 /// \brief True if we want to ignore EOF token and continue later on (thus
155 /// avoid tearing the Lexer and etc. down).
156 bool IncrementalProcessing;
158 /// \brief The code-completion handler.
159 CodeCompletionHandler *CodeComplete;
161 /// \brief The file that we're performing code-completion for, if any.
162 const FileEntry *CodeCompletionFile;
164 /// \brief The offset in file for the code-completion point.
165 unsigned CodeCompletionOffset;
167 /// \brief The location for the code-completion point. This gets instantiated
168 /// when the CodeCompletionFile gets #include'ed for preprocessing.
169 SourceLocation CodeCompletionLoc;
171 /// \brief The start location for the file of the code-completion point.
172 /// This gets instantiated when the CodeCompletionFile gets #include'ed
173 /// for preprocessing.
174 SourceLocation CodeCompletionFileLoc;
176 /// \brief The source location of the 'import' contextual keyword we just
178 SourceLocation ModuleImportLoc;
180 /// \brief The module import path that we're currently processing.
181 llvm::SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2>
184 /// \brief Whether the module import expectes an identifier next. Otherwise,
185 /// it expects a '.' or ';'.
186 bool ModuleImportExpectsIdentifier;
188 /// \brief The source location of the currently-active
189 /// #pragma clang arc_cf_code_audited begin.
190 SourceLocation PragmaARCCFCodeAuditedLoc;
192 /// \brief True if we hit the code-completion point.
193 bool CodeCompletionReached;
195 /// \brief The number of bytes that we will initially skip when entering the
196 /// main file, which is used when loading a precompiled preamble, along
197 /// with a flag that indicates whether skipping this number of bytes will
198 /// place the lexer at the start of a line.
199 std::pair<unsigned, bool> SkipMainFilePreamble;
201 /// CurLexer - This is the current top of the stack that we're lexing from if
202 /// not expanding a macro and we are lexing directly from source code.
203 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null.
204 OwningPtr<Lexer> CurLexer;
206 /// CurPTHLexer - This is the current top of stack that we're lexing from if
207 /// not expanding from a macro and we are lexing from a PTH cache.
208 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null.
209 OwningPtr<PTHLexer> CurPTHLexer;
211 /// CurPPLexer - This is the current top of the stack what we're lexing from
212 /// if not expanding a macro. This is an alias for either CurLexer or
214 PreprocessorLexer *CurPPLexer;
216 /// CurLookup - The DirectoryLookup structure used to find the current
217 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to
218 /// implement #include_next and find directory-specific properties.
219 const DirectoryLookup *CurDirLookup;
221 /// CurTokenLexer - This is the current macro we are expanding, if we are
222 /// expanding a macro. One of CurLexer and CurTokenLexer must be null.
223 OwningPtr<TokenLexer> CurTokenLexer;
225 /// \brief The kind of lexer we're currently working with.
231 CLK_LexAfterModuleImport
234 /// IncludeMacroStack - This keeps track of the stack of files currently
235 /// #included, and macros currently being expanded from, not counting
236 /// CurLexer/CurTokenLexer.
237 struct IncludeStackInfo {
238 enum CurLexerKind CurLexerKind;
240 PTHLexer *ThePTHLexer;
241 PreprocessorLexer *ThePPLexer;
242 TokenLexer *TheTokenLexer;
243 const DirectoryLookup *TheDirLookup;
245 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P,
246 PreprocessorLexer* PPL,
247 TokenLexer* TL, const DirectoryLookup *D)
248 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL),
249 TheTokenLexer(TL), TheDirLookup(D) {}
251 std::vector<IncludeStackInfo> IncludeMacroStack;
253 /// Callbacks - These are actions invoked when some preprocessor activity is
254 /// encountered (e.g. a file is #included, etc).
255 PPCallbacks *Callbacks;
257 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping
258 /// to the actual definition of the macro.
259 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros;
261 /// \brief Macros that we want to warn because they are not used at the end
262 /// of the translation unit; we store just their SourceLocations instead
263 /// something like MacroInfo*. The benefit of this is that when we are
264 /// deserializing from PCH, we don't need to deserialize identifier & macros
265 /// just so that we can report that they are unused, we just warn using
266 /// the SourceLocations of this set (that will be filled by the ASTReader).
267 /// We are using SmallPtrSet instead of a vector for faster removal.
268 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy;
269 WarnUnusedMacroLocsTy WarnUnusedMacroLocs;
271 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be
272 /// reused for quick allocation.
273 MacroArgs *MacroArgCache;
274 friend class MacroArgs;
276 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma
277 /// push_macro directive, we keep a MacroInfo stack used to restore
278 /// previous macro value.
279 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo;
281 // Various statistics we track for performance analysis.
282 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma;
283 unsigned NumIf, NumElse, NumEndif;
284 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth;
285 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded;
286 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste;
289 /// Predefines - This string is the predefined macros that preprocessor
290 /// should use from the command line etc.
291 std::string Predefines;
293 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic.
294 enum { TokenLexerCacheSize = 8 };
295 unsigned NumCachedTokenLexers;
296 TokenLexer *TokenLexerCache[TokenLexerCacheSize];
298 /// \brief Keeps macro expanded tokens for TokenLexers.
300 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
301 /// going to lex in the cache and when it finishes the tokens are removed
302 /// from the end of the cache.
303 SmallVector<Token, 16> MacroExpandedTokens;
304 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack;
306 /// \brief A record of the macro definitions and expansions that
307 /// occurred during preprocessing.
309 /// This is an optional side structure that can be enabled with
310 /// \c createPreprocessingRecord() prior to preprocessing.
311 PreprocessingRecord *Record;
313 private: // Cached tokens state.
314 typedef SmallVector<Token, 1> CachedTokensTy;
316 /// CachedTokens - Cached tokens are stored here when we do backtracking or
317 /// lookahead. They are "lexed" by the CachingLex() method.
318 CachedTokensTy CachedTokens;
320 /// CachedLexPos - The position of the cached token that CachingLex() should
321 /// "lex" next. If it points beyond the CachedTokens vector, it means that
322 /// a normal Lex() should be invoked.
323 CachedTokensTy::size_type CachedLexPos;
325 /// BacktrackPositions - Stack of backtrack positions, allowing nested
326 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to
327 /// indicate where CachedLexPos should be set when the BackTrack() method is
328 /// invoked (at which point the last position is popped).
329 std::vector<CachedTokensTy::size_type> BacktrackPositions;
331 struct MacroInfoChain {
333 MacroInfoChain *Next;
334 MacroInfoChain *Prev;
337 /// MacroInfos are managed as a chain for easy disposal. This is the head
339 MacroInfoChain *MIChainHead;
341 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick
343 MacroInfoChain *MICache;
345 MacroInfo *getInfoForMacro(IdentifierInfo *II) const;
348 Preprocessor(DiagnosticsEngine &diags, LangOptions &opts,
349 const TargetInfo *target,
350 SourceManager &SM, HeaderSearch &Headers,
351 ModuleLoader &TheModuleLoader,
352 IdentifierInfoLookup *IILookup = 0,
353 bool OwnsHeaderSearch = false,
354 bool DelayInitialization = false,
355 bool IncrProcessing = false);
359 /// \brief Initialize the preprocessor, if the constructor did not already
360 /// perform the initialization.
362 /// \param Target Information about the target.
363 void Initialize(const TargetInfo &Target);
365 DiagnosticsEngine &getDiagnostics() const { return *Diags; }
366 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; }
368 const LangOptions &getLangOpts() const { return LangOpts; }
369 const TargetInfo &getTargetInfo() const { return *Target; }
370 FileManager &getFileManager() const { return FileMgr; }
371 SourceManager &getSourceManager() const { return SourceMgr; }
372 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; }
374 IdentifierTable &getIdentifierTable() { return Identifiers; }
375 SelectorTable &getSelectorTable() { return Selectors; }
376 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; }
377 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; }
379 void setPTHManager(PTHManager* pm);
381 PTHManager *getPTHManager() { return PTH.get(); }
383 void setExternalSource(ExternalPreprocessorSource *Source) {
384 ExternalSource = Source;
387 ExternalPreprocessorSource *getExternalSource() const {
388 return ExternalSource;
391 /// \brief Retrieve the module loader associated with this preprocessor.
392 ModuleLoader &getModuleLoader() const { return TheModuleLoader; }
394 /// SetCommentRetentionState - Control whether or not the preprocessor retains
395 /// comments in output.
396 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) {
397 this->KeepComments = KeepComments | KeepMacroComments;
398 this->KeepMacroComments = KeepMacroComments;
401 bool getCommentRetentionState() const { return KeepComments; }
403 void SetSuppressIncludeNotFoundError(bool Suppress) {
404 SuppressIncludeNotFoundError = Suppress;
407 bool GetSuppressIncludeNotFoundError() {
408 return SuppressIncludeNotFoundError;
411 /// isCurrentLexer - Return true if we are lexing directly from the specified
413 bool isCurrentLexer(const PreprocessorLexer *L) const {
414 return CurPPLexer == L;
417 /// getCurrentLexer - Return the current lexer being lexed from. Note
418 /// that this ignores any potentially active macro expansions and _Pragma
419 /// expansions going on at the time.
420 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; }
422 /// getCurrentFileLexer - Return the current file lexer being lexed from.
423 /// Note that this ignores any potentially active macro expansions and _Pragma
424 /// expansions going on at the time.
425 PreprocessorLexer *getCurrentFileLexer() const;
427 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks.
428 /// Note that this class takes ownership of any PPCallbacks object given to
430 PPCallbacks *getPPCallbacks() const { return Callbacks; }
431 void addPPCallbacks(PPCallbacks *C) {
433 C = new PPChainedCallbacks(C, Callbacks);
437 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to
438 /// or null if it isn't #define'd.
439 MacroInfo *getMacroInfo(IdentifierInfo *II) const {
440 if (!II->hasMacroDefinition())
443 return getInfoForMacro(II);
446 /// setMacroInfo - Specify a macro for this identifier.
448 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI,
449 bool LoadedFromAST = false);
451 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current
452 /// state of the macro table. This visits every currently-defined macro.
453 typedef llvm::DenseMap<IdentifierInfo*,
454 MacroInfo*>::const_iterator macro_iterator;
455 macro_iterator macro_begin(bool IncludeExternalMacros = true) const;
456 macro_iterator macro_end(bool IncludeExternalMacros = true) const;
458 const std::string &getPredefines() const { return Predefines; }
459 /// setPredefines - Set the predefines for this Preprocessor. These
460 /// predefines are automatically injected when parsing the main file.
461 void setPredefines(const char *P) { Predefines = P; }
462 void setPredefines(const std::string &P) { Predefines = P; }
464 /// getIdentifierInfo - Return information about the specified preprocessor
465 /// identifier token. The version of this method that takes two character
466 /// pointers is preferred unless the identifier is already available as a
467 /// string (this avoids allocation and copying of memory to construct an
469 IdentifierInfo *getIdentifierInfo(StringRef Name) const {
470 return &Identifiers.get(Name);
473 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor.
474 /// If 'Namespace' is non-null, then it is a token required to exist on the
475 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC".
476 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler);
477 void AddPragmaHandler(PragmaHandler *Handler) {
478 AddPragmaHandler(StringRef(), Handler);
481 /// RemovePragmaHandler - Remove the specific pragma handler from
482 /// the preprocessor. If \arg Namespace is non-null, then it should
483 /// be the namespace that \arg Handler was added to. It is an error
484 /// to remove a handler that has not been registered.
485 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler);
486 void RemovePragmaHandler(PragmaHandler *Handler) {
487 RemovePragmaHandler(StringRef(), Handler);
490 /// \brief Add the specified comment handler to the preprocessor.
491 void AddCommentHandler(CommentHandler *Handler);
493 /// \brief Remove the specified comment handler.
495 /// It is an error to remove a handler that has not been registered.
496 void RemoveCommentHandler(CommentHandler *Handler);
498 /// \brief Set the code completion handler to the given object.
499 void setCodeCompletionHandler(CodeCompletionHandler &Handler) {
500 CodeComplete = &Handler;
503 /// \brief Retrieve the current code-completion handler.
504 CodeCompletionHandler *getCodeCompletionHandler() const {
508 /// \brief Clear out the code completion handler.
509 void clearCodeCompletionHandler() {
513 /// \brief Hook used by the lexer to invoke the "natural language" code
514 /// completion point.
515 void CodeCompleteNaturalLanguage();
517 /// \brief Retrieve the preprocessing record, or NULL if there is no
518 /// preprocessing record.
519 PreprocessingRecord *getPreprocessingRecord() const { return Record; }
521 /// \brief Create a new preprocessing record, which will keep track of
522 /// all macro expansions, macro definitions, etc.
523 void createPreprocessingRecord(bool RecordConditionalDirectives);
525 /// EnterMainSourceFile - Enter the specified FileID as the main source file,
526 /// which implicitly adds the builtin defines etc.
527 void EnterMainSourceFile();
529 /// EndSourceFile - Inform the preprocessor callbacks that processing is
531 void EndSourceFile();
533 /// EnterSourceFile - Add a source file to the top of the include stack and
534 /// start lexing tokens from it instead of the current buffer. Emit an error
535 /// and don't enter the file on error.
536 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir,
539 /// EnterMacro - Add a Macro to the top of the include stack and start lexing
540 /// tokens from it instead of the current buffer. Args specifies the
541 /// tokens input to a function-like macro.
543 /// ILEnd specifies the location of the ')' for a function-like macro or the
544 /// identifier for an object-like macro.
545 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args);
547 /// EnterTokenStream - Add a "macro" context to the top of the include stack,
548 /// which will cause the lexer to start returning the specified tokens.
550 /// If DisableMacroExpansion is true, tokens lexed from the token stream will
551 /// not be subject to further macro expansion. Otherwise, these tokens will
552 /// be re-macro-expanded when/if expansion is enabled.
554 /// If OwnsTokens is false, this method assumes that the specified stream of
555 /// tokens has a permanent owner somewhere, so they do not need to be copied.
556 /// If it is true, it assumes the array of tokens is allocated with new[] and
559 void EnterTokenStream(const Token *Toks, unsigned NumToks,
560 bool DisableMacroExpansion, bool OwnsTokens);
562 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the
563 /// lexer stack. This should only be used in situations where the current
564 /// state of the top-of-stack lexer is known.
565 void RemoveTopOfLexerStack();
567 /// EnableBacktrackAtThisPos - From the point that this method is called, and
568 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor
569 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will
570 /// make the Preprocessor re-lex the same tokens.
572 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can
573 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will
574 /// be combined with the EnableBacktrackAtThisPos calls in reverse order.
576 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack
577 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of
578 /// tokens will continue indefinitely.
580 void EnableBacktrackAtThisPos();
582 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call.
583 void CommitBacktrackedTokens();
585 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since
586 /// EnableBacktrackAtThisPos() was previously called.
589 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and
590 /// caching of tokens is on.
591 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); }
593 /// Lex - To lex a token from the preprocessor, just pull a token from the
594 /// current lexer or macro object.
595 void Lex(Token &Result) {
596 switch (CurLexerKind) {
597 case CLK_Lexer: CurLexer->Lex(Result); break;
598 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break;
599 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break;
600 case CLK_CachingLexer: CachingLex(Result); break;
601 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break;
605 void LexAfterModuleImport(Token &Result);
607 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get
608 /// something not a comment. This is useful in -E -C mode where comments
609 /// would foul up preprocessor directive handling.
610 void LexNonComment(Token &Result) {
613 while (Result.getKind() == tok::comment);
616 /// LexUnexpandedToken - This is just like Lex, but this disables macro
617 /// expansion of identifier tokens.
618 void LexUnexpandedToken(Token &Result) {
619 // Disable macro expansion.
620 bool OldVal = DisableMacroExpansion;
621 DisableMacroExpansion = true;
626 DisableMacroExpansion = OldVal;
629 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro
630 /// expansion of identifier tokens.
631 void LexUnexpandedNonComment(Token &Result) {
633 LexUnexpandedToken(Result);
634 while (Result.getKind() == tok::comment);
637 /// LookAhead - This peeks ahead N tokens and returns that token without
638 /// consuming any tokens. LookAhead(0) returns the next token that would be
639 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This
640 /// returns normal tokens after phase 5. As such, it is equivalent to using
641 /// 'Lex', not 'LexUnexpandedToken'.
642 const Token &LookAhead(unsigned N) {
643 if (CachedLexPos + N < CachedTokens.size())
644 return CachedTokens[CachedLexPos+N];
646 return PeekAhead(N+1);
649 /// RevertCachedTokens - When backtracking is enabled and tokens are cached,
650 /// this allows to revert a specific number of tokens.
651 /// Note that the number of tokens being reverted should be up to the last
652 /// backtrack position, not more.
653 void RevertCachedTokens(unsigned N) {
654 assert(isBacktrackEnabled() &&
655 "Should only be called when tokens are cached for backtracking");
656 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back())
657 && "Should revert tokens up to the last backtrack position, not more");
658 assert(signed(CachedLexPos) - signed(N) >= 0 &&
659 "Corrupted backtrack positions ?");
663 /// EnterToken - Enters a token in the token stream to be lexed next. If
664 /// BackTrack() is called afterwards, the token will remain at the insertion
666 void EnterToken(const Token &Tok) {
667 EnterCachingLexMode();
668 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok);
671 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching
672 /// tokens (because backtrack is enabled) it should replace the most recent
673 /// cached tokens with the given annotation token. This function has no effect
674 /// if backtracking is not enabled.
676 /// Note that the use of this function is just for optimization; so that the
677 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is
679 void AnnotateCachedTokens(const Token &Tok) {
680 assert(Tok.isAnnotation() && "Expected annotation token");
681 if (CachedLexPos != 0 && isBacktrackEnabled())
682 AnnotatePreviousCachedTokens(Tok);
685 /// \brief Replace the last token with an annotation token.
687 /// Like AnnotateCachedTokens(), this routine replaces an
688 /// already-parsed (and resolved) token with an annotation
689 /// token. However, this routine only replaces the last token with
690 /// the annotation token; it does not affect any other cached
691 /// tokens. This function has no effect if backtracking is not
693 void ReplaceLastTokenWithAnnotation(const Token &Tok) {
694 assert(Tok.isAnnotation() && "Expected annotation token");
695 if (CachedLexPos != 0 && isBacktrackEnabled())
696 CachedTokens[CachedLexPos-1] = Tok;
699 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/
700 /// CurTokenLexer pointers.
701 void recomputeCurLexerKind();
703 /// \brief Returns true if incremental processing is enabled
704 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; }
706 /// \brief Enables the incremental processing
707 void enableIncrementalProcessing(bool value = true) {
708 IncrementalProcessing = value;
711 /// \brief Specify the point at which code-completion will be performed.
713 /// \param File the file in which code completion should occur. If
714 /// this file is included multiple times, code-completion will
715 /// perform completion the first time it is included. If NULL, this
716 /// function clears out the code-completion point.
718 /// \param Line the line at which code completion should occur
721 /// \param Column the column at which code completion should occur
724 /// \returns true if an error occurred, false otherwise.
725 bool SetCodeCompletionPoint(const FileEntry *File,
726 unsigned Line, unsigned Column);
728 /// \brief Determine if we are performing code completion.
729 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; }
731 /// \brief Returns the location of the code-completion point.
732 /// Returns an invalid location if code-completion is not enabled or the file
733 /// containing the code-completion point has not been lexed yet.
734 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; }
736 /// \brief Returns the start location of the file of code-completion point.
737 /// Returns an invalid location if code-completion is not enabled or the file
738 /// containing the code-completion point has not been lexed yet.
739 SourceLocation getCodeCompletionFileLoc() const {
740 return CodeCompletionFileLoc;
743 /// \brief Returns true if code-completion is enabled and we have hit the
744 /// code-completion point.
745 bool isCodeCompletionReached() const { return CodeCompletionReached; }
747 /// \brief Note that we hit the code-completion point.
748 void setCodeCompletionReached() {
749 assert(isCodeCompletionEnabled() && "Code-completion not enabled!");
750 CodeCompletionReached = true;
751 // Silence any diagnostics that occur after we hit the code-completion.
752 getDiagnostics().setSuppressAllDiagnostics(true);
755 /// \brief The location of the currently-active #pragma clang
756 /// arc_cf_code_audited begin. Returns an invalid location if there
757 /// is no such pragma active.
758 SourceLocation getPragmaARCCFCodeAuditedLoc() const {
759 return PragmaARCCFCodeAuditedLoc;
762 /// \brief Set the location of the currently-active #pragma clang
763 /// arc_cf_code_audited begin. An invalid location ends the pragma.
764 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) {
765 PragmaARCCFCodeAuditedLoc = Loc;
768 /// \brief Instruct the preprocessor to skip part of the main
769 /// the main source file.
771 /// \brief Bytes The number of bytes in the preamble to skip.
773 /// \brief StartOfLine Whether skipping these bytes puts the lexer at the
775 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) {
776 SkipMainFilePreamble.first = Bytes;
777 SkipMainFilePreamble.second = StartOfLine;
780 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at
781 /// the specified Token's location, translating the token's start
782 /// position in the current buffer into a SourcePosition object for rendering.
783 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const {
784 return Diags->Report(Loc, DiagID);
787 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const {
788 return Diags->Report(Tok.getLocation(), DiagID);
791 /// getSpelling() - Return the 'spelling' of the token at the given
792 /// location; does not go up to the spelling location or down to the
793 /// expansion location.
795 /// \param buffer A buffer which will be used only if the token requires
796 /// "cleaning", e.g. if it contains trigraphs or escaped newlines
797 /// \param invalid If non-null, will be set \c true if an error occurs.
798 StringRef getSpelling(SourceLocation loc,
799 SmallVectorImpl<char> &buffer,
800 bool *invalid = 0) const {
801 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid);
804 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a
805 /// token is the characters used to represent the token in the source file
806 /// after trigraph expansion and escaped-newline folding. In particular, this
807 /// wants to get the true, uncanonicalized, spelling of things like digraphs
810 /// \param Invalid If non-null, will be set \c true if an error occurs.
811 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const {
812 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid);
815 /// getSpelling - This method is used to get the spelling of a token into a
816 /// preallocated buffer, instead of as an std::string. The caller is required
817 /// to allocate enough space for the token, which is guaranteed to be at least
818 /// Tok.getLength() bytes long. The length of the actual result is returned.
820 /// Note that this method may do two possible things: it may either fill in
821 /// the buffer specified with characters, or it may *change the input pointer*
822 /// to point to a constant buffer with the data already in it (avoiding a
823 /// copy). The caller is not allowed to modify the returned buffer pointer
824 /// if an internal buffer is returned.
825 unsigned getSpelling(const Token &Tok, const char *&Buffer,
826 bool *Invalid = 0) const {
827 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid);
830 /// getSpelling - This method is used to get the spelling of a token into a
831 /// SmallVector. Note that the returned StringRef may not point to the
832 /// supplied buffer if a copy can be avoided.
833 StringRef getSpelling(const Token &Tok,
834 SmallVectorImpl<char> &Buffer,
835 bool *Invalid = 0) const;
837 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant
838 /// with length 1, return the character.
839 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok,
840 bool *Invalid = 0) const {
841 assert(Tok.is(tok::numeric_constant) &&
842 Tok.getLength() == 1 && "Called on unsupported token");
843 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1");
845 // If the token is carrying a literal data pointer, just use it.
846 if (const char *D = Tok.getLiteralData())
849 // Otherwise, fall back on getCharacterData, which is slower, but always
851 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid);
854 /// \brief Retrieve the name of the immediate macro expansion.
856 /// This routine starts from a source location, and finds the name of the macro
857 /// responsible for its immediate expansion. It looks through any intervening
858 /// macro argument expansions to compute this. It returns a StringRef which
859 /// refers to the SourceManager-owned buffer of the source where that macro
860 /// name is spelled. Thus, the result shouldn't out-live the SourceManager.
861 StringRef getImmediateMacroName(SourceLocation Loc) {
862 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts());
865 /// CreateString - Plop the specified string into a scratch buffer and set the
866 /// specified token's location and length to it. If specified, the source
867 /// location provides a location of the expansion point of the token.
868 void CreateString(const char *Buf, unsigned Len, Token &Tok,
869 SourceLocation ExpansionLocStart = SourceLocation(),
870 SourceLocation ExpansionLocEnd = SourceLocation());
872 /// \brief Computes the source location just past the end of the
873 /// token at this source location.
875 /// This routine can be used to produce a source location that
876 /// points just past the end of the token referenced by \p Loc, and
877 /// is generally used when a diagnostic needs to point just after a
878 /// token where it expected something different that it received. If
879 /// the returned source location would not be meaningful (e.g., if
880 /// it points into a macro), this routine returns an invalid
883 /// \param Offset an offset from the end of the token, where the source
884 /// location should refer to. The default offset (0) produces a source
885 /// location pointing just past the end of the token; an offset of 1 produces
886 /// a source location pointing to the last character in the token, etc.
887 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) {
888 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts);
891 /// \brief Returns true if the given MacroID location points at the first
892 /// token of the macro expansion.
894 /// \param MacroBegin If non-null and function returns true, it is set to
895 /// begin location of the macro.
896 bool isAtStartOfMacroExpansion(SourceLocation loc,
897 SourceLocation *MacroBegin = 0) const {
898 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts,
902 /// \brief Returns true if the given MacroID location points at the last
903 /// token of the macro expansion.
905 /// \param MacroBegin If non-null and function returns true, it is set to
906 /// end location of the macro.
907 bool isAtEndOfMacroExpansion(SourceLocation loc,
908 SourceLocation *MacroEnd = 0) const {
909 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd);
912 /// DumpToken - Print the token to stderr, used for debugging.
914 void DumpToken(const Token &Tok, bool DumpFlags = false) const;
915 void DumpLocation(SourceLocation Loc) const;
916 void DumpMacro(const MacroInfo &MI) const;
918 /// AdvanceToTokenCharacter - Given a location that specifies the start of a
919 /// token, return a new location that specifies a character within the token.
920 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
921 unsigned Char) const {
922 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts);
925 /// IncrementPasteCounter - Increment the counters for the number of token
926 /// paste operations performed. If fast was specified, this is a 'fast paste'
929 void IncrementPasteCounter(bool isFast) {
938 size_t getTotalMemory() const;
940 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a
941 /// comment (/##/) in microsoft mode, this method handles updating the current
942 /// state, returning the token on the next source line.
943 void HandleMicrosoftCommentPaste(Token &Tok);
945 //===--------------------------------------------------------------------===//
946 // Preprocessor callback methods. These are invoked by a lexer as various
947 // directives and events are found.
949 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the
950 /// identifier information for the token and install it into the token,
951 /// updating the token kind accordingly.
952 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const;
955 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons;
959 // SetPoisonReason - Call this function to indicate the reason for
960 // poisoning an identifier. If that identifier is accessed while
961 // poisoned, then this reason will be used instead of the default
962 // "poisoned" diagnostic.
963 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID);
965 // HandlePoisonedIdentifier - Display reason for poisoned
967 void HandlePoisonedIdentifier(Token & Tok);
969 void MaybeHandlePoisonedIdentifier(Token & Identifier) {
970 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) {
971 if(II->isPoisoned()) {
972 HandlePoisonedIdentifier(Identifier);
978 /// Identifiers used for SEH handling in Borland. These are only
979 /// allowed in particular circumstances
981 IdentifierInfo *Ident__exception_code,
982 *Ident___exception_code,
983 *Ident_GetExceptionCode;
984 // __except filter expression
985 IdentifierInfo *Ident__exception_info,
986 *Ident___exception_info,
987 *Ident_GetExceptionInfo;
989 IdentifierInfo *Ident__abnormal_termination,
990 *Ident___abnormal_termination,
991 *Ident_AbnormalTermination;
993 void PoisonSEHIdentifiers(bool Poison = true); // Borland
995 /// HandleIdentifier - This callback is invoked when the lexer reads an
996 /// identifier and has filled in the tokens IdentifierInfo member. This
997 /// callback potentially macro expands it or turns it into a named token (like
999 void HandleIdentifier(Token &Identifier);
1002 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of
1003 /// the current file. This either returns the EOF token and returns true, or
1004 /// pops a level off the include stack and returns false, at which point the
1005 /// client should call lex again.
1006 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false);
1008 /// HandleEndOfTokenLexer - This callback is invoked when the current
1009 /// TokenLexer hits the end of its token stream.
1010 bool HandleEndOfTokenLexer(Token &Result);
1012 /// HandleDirective - This callback is invoked when the lexer sees a # token
1013 /// at the start of a line. This consumes the directive, modifies the
1014 /// lexer/preprocessor state, and advances the lexer(s) so that the next token
1015 /// read is the correct one.
1016 void HandleDirective(Token &Result);
1018 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If
1019 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is
1020 /// true, then we consider macros that expand to zero tokens as being ok.
1021 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false);
1023 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the
1024 /// current line until the tok::eod token is found.
1025 void DiscardUntilEndOfDirective();
1027 /// SawDateOrTime - This returns true if the preprocessor has seen a use of
1028 /// __DATE__ or __TIME__ in the file so far.
1029 bool SawDateOrTime() const {
1030 return DATELoc != SourceLocation() || TIMELoc != SourceLocation();
1032 unsigned getCounterValue() const { return CounterValue; }
1033 void setCounterValue(unsigned V) { CounterValue = V; }
1035 /// \brief Retrieves the module that we're currently building, if any.
1036 Module *getCurrentModule();
1038 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide
1040 MacroInfo *AllocateMacroInfo(SourceLocation L);
1042 /// CloneMacroInfo - Allocate a new MacroInfo object which is clone of MI.
1043 MacroInfo *CloneMacroInfo(const MacroInfo &MI);
1045 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully
1046 /// checked and spelled filename, e.g. as an operand of #include. This returns
1047 /// true if the input filename was in <>'s or false if it were in ""'s. The
1048 /// caller is expected to provide a buffer that is large enough to hold the
1049 /// spelling of the filename, but is also expected to handle the case when
1050 /// this method decides to use a different buffer.
1051 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename);
1053 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file,
1054 /// return null on failure. isAngled indicates whether the file reference is
1055 /// for system #include's or not (i.e. using <> instead of "").
1056 const FileEntry *LookupFile(StringRef Filename,
1057 bool isAngled, const DirectoryLookup *FromDir,
1058 const DirectoryLookup *&CurDir,
1059 SmallVectorImpl<char> *SearchPath,
1060 SmallVectorImpl<char> *RelativePath,
1061 Module **SuggestedModule,
1062 bool SkipCache = false);
1064 /// GetCurLookup - The DirectoryLookup structure used to find the current
1065 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to
1066 /// implement #include_next and find directory-specific properties.
1067 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; }
1069 /// isInPrimaryFile - Return true if we're in the top-level file, not in a
1071 bool isInPrimaryFile() const;
1073 /// ConcatenateIncludeName - Handle cases where the #include name is expanded
1074 /// from a macro as multiple tokens, which need to be glued together. This
1075 /// occurs for code like:
1076 /// #define FOO <a/b.h>
1078 /// because in this case, "<a/b.h>" is returned as 7 tokens, not one.
1080 /// This code concatenates and consumes tokens up to the '>' token. It
1081 /// returns false if the > was found, otherwise it returns true if it finds
1082 /// and consumes the EOD marker.
1083 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer,
1084 SourceLocation &End);
1086 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is
1087 /// followed by EOD. Return true if the token is not a valid on-off-switch.
1088 bool LexOnOffSwitch(tok::OnOffSwitch &OOS);
1092 void PushIncludeMacroStack() {
1093 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind,
1097 CurTokenLexer.take(),
1102 void PopIncludeMacroStack() {
1103 CurLexer.reset(IncludeMacroStack.back().TheLexer);
1104 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer);
1105 CurPPLexer = IncludeMacroStack.back().ThePPLexer;
1106 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer);
1107 CurDirLookup = IncludeMacroStack.back().TheDirLookup;
1108 CurLexerKind = IncludeMacroStack.back().CurLexerKind;
1109 IncludeMacroStack.pop_back();
1112 /// AllocateMacroInfo - Allocate a new MacroInfo object.
1113 MacroInfo *AllocateMacroInfo();
1115 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will
1116 /// be reused for allocating new MacroInfo objects.
1117 void ReleaseMacroInfo(MacroInfo* MI);
1119 /// ReadMacroName - Lex and validate a macro name, which occurs after a
1120 /// #define or #undef. This emits a diagnostic, sets the token kind to eod,
1121 /// and discards the rest of the macro line if the macro name is invalid.
1122 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0);
1124 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro
1125 /// definition has just been read. Lex the rest of the arguments and the
1126 /// closing ), updating MI with what we learn and saving in LastTok the
1127 /// last token read.
1128 /// Return true if an error occurs parsing the arg list.
1129 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok);
1131 /// SkipExcludedConditionalBlock - We just read a #if or related directive and
1132 /// decided that the subsequent tokens are in the #if'd out portion of the
1133 /// file. Lex the rest of the file, until we see an #endif. If
1134 /// FoundNonSkipPortion is true, then we have already emitted code for part of
1135 /// this #if directive, so #else/#elif blocks should never be entered. If
1136 /// FoundElse is false, then #else directives are ok, if not, then we have
1137 /// already seen one so a #else directive is a duplicate. When this returns,
1138 /// the caller can lex the first valid token.
1139 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc,
1140 bool FoundNonSkipPortion, bool FoundElse,
1141 SourceLocation ElseLoc = SourceLocation());
1143 /// PTHSkipExcludedConditionalBlock - A fast PTH version of
1144 /// SkipExcludedConditionalBlock.
1145 void PTHSkipExcludedConditionalBlock();
1147 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that
1148 /// may occur after a #if or #elif directive and return it as a bool. If the
1149 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro.
1150 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro);
1152 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas:
1153 /// #pragma GCC poison/system_header/dependency and #pragma once.
1154 void RegisterBuiltinPragmas();
1156 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the
1157 /// identifier table.
1158 void RegisterBuiltinMacros();
1160 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to
1161 /// be expanded as a macro, handle it and return the next token as 'Tok'. If
1162 /// the macro should not be expanded return true, otherwise return false.
1163 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI);
1165 /// \brief Cache macro expanded tokens for TokenLexers.
1167 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is
1168 /// going to lex in the cache and when it finishes the tokens are removed
1169 /// from the end of the cache.
1170 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer,
1171 ArrayRef<Token> tokens);
1172 void removeCachedMacroExpandedTokensOfLastLexer();
1173 friend void TokenLexer::ExpandFunctionArguments();
1175 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be
1176 /// lexed is a '('. If so, consume the token and return true, if not, this
1177 /// method should have no observable side-effect on the lexed tokens.
1178 bool isNextPPTokenLParen();
1180 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is
1181 /// invoked to read all of the formal arguments specified for the macro
1182 /// invocation. This returns null on error.
1183 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI,
1184 SourceLocation &ExpansionEnd);
1186 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded
1187 /// as a builtin macro, handle it and return the next token as 'Tok'.
1188 void ExpandBuiltinMacro(Token &Tok);
1190 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then
1191 /// return the first token after the directive. The _Pragma token has just
1192 /// been read into 'Tok'.
1193 void Handle_Pragma(Token &Tok);
1195 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text
1196 /// is not enclosed within a string literal.
1197 void HandleMicrosoft__pragma(Token &Tok);
1199 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and
1200 /// start lexing tokens from it instead of the current buffer.
1201 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir);
1203 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and
1204 /// start getting tokens from it using the PTH cache.
1205 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir);
1207 /// IsFileLexer - Returns true if we are lexing from a file and not a
1208 /// pragma or a macro.
1209 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) {
1210 return L ? !L->isPragmaLexer() : P != 0;
1213 static bool IsFileLexer(const IncludeStackInfo& I) {
1214 return IsFileLexer(I.TheLexer, I.ThePPLexer);
1217 bool IsFileLexer() const {
1218 return IsFileLexer(CurLexer.get(), CurPPLexer);
1221 //===--------------------------------------------------------------------===//
1223 void CachingLex(Token &Result);
1224 bool InCachingLexMode() const {
1225 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means
1226 // that we are past EOF, not that we are in CachingLex mode.
1227 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 &&
1228 !IncludeMacroStack.empty();
1230 void EnterCachingLexMode();
1231 void ExitCachingLexMode() {
1232 if (InCachingLexMode())
1233 RemoveTopOfLexerStack();
1235 const Token &PeekAhead(unsigned N);
1236 void AnnotatePreviousCachedTokens(const Token &Tok);
1238 //===--------------------------------------------------------------------===//
1239 /// Handle*Directive - implement the various preprocessor directives. These
1240 /// should side-effect the current preprocessor object so that the next call
1241 /// to Lex() will return the appropriate token next.
1242 void HandleLineDirective(Token &Tok);
1243 void HandleDigitDirective(Token &Tok);
1244 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning);
1245 void HandleIdentSCCSDirective(Token &Tok);
1246 void HandleMacroPublicDirective(Token &Tok);
1247 void HandleMacroPrivateDirective(Token &Tok);
1250 void HandleIncludeDirective(SourceLocation HashLoc,
1252 const DirectoryLookup *LookupFrom = 0,
1253 bool isImport = false);
1254 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok);
1255 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok);
1256 void HandleImportDirective(SourceLocation HashLoc, Token &Tok);
1257 void HandleMicrosoftImportDirective(Token &Tok);
1260 void HandleDefineDirective(Token &Tok);
1261 void HandleUndefDirective(Token &Tok);
1263 // Conditional Inclusion.
1264 void HandleIfdefDirective(Token &Tok, bool isIfndef,
1265 bool ReadAnyTokensBeforeDirective);
1266 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective);
1267 void HandleEndifDirective(Token &Tok);
1268 void HandleElseDirective(Token &Tok);
1269 void HandleElifDirective(Token &Tok);
1272 void HandlePragmaDirective(unsigned Introducer);
1274 void HandlePragmaOnce(Token &OnceTok);
1275 void HandlePragmaMark();
1276 void HandlePragmaPoison(Token &PoisonTok);
1277 void HandlePragmaSystemHeader(Token &SysHeaderTok);
1278 void HandlePragmaDependency(Token &DependencyTok);
1279 void HandlePragmaComment(Token &CommentTok);
1280 void HandlePragmaMessage(Token &MessageTok);
1281 void HandlePragmaPushMacro(Token &Tok);
1282 void HandlePragmaPopMacro(Token &Tok);
1283 void HandlePragmaIncludeAlias(Token &Tok);
1284 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok);
1286 // Return true and store the first token only if any CommentHandler
1287 // has inserted some tokens and getCommentRetentionState() is false.
1288 bool HandleComment(Token &Token, SourceRange Comment);
1290 /// \brief A macro is used, update information about macros that need unused
1292 void markMacroAsUsed(MacroInfo *MI);
1295 /// \brief Abstract base class that describes a handler that will receive
1296 /// source ranges for each of the comments encountered in the source file.
1297 class CommentHandler {
1299 virtual ~CommentHandler();
1301 // The handler shall return true if it has pushed any tokens
1302 // to be read using e.g. EnterToken or EnterTokenStream.
1303 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0;
1306 } // end namespace clang