Preprocessor.h revision 6fe6a49c4058211ff4489023c78615ec0266c5ff
1//===--- Preprocessor.h - C Language Family Preprocessor --------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file is distributed under the University of Illinois Open Source 6// License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// This file defines the Preprocessor interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_LEX_PREPROCESSOR_H 15#define LLVM_CLANG_LEX_PREPROCESSOR_H 16 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" 34#include <vector> 35 36namespace llvm { 37 template<unsigned InternalLen> class SmallString; 38} 39 40namespace clang { 41 42class SourceManager; 43class ExternalPreprocessorSource; 44class FileManager; 45class FileEntry; 46class HeaderSearch; 47class PragmaNamespace; 48class PragmaHandler; 49class CommentHandler; 50class ScratchBuffer; 51class TargetInfo; 52class PPCallbacks; 53class CodeCompletionHandler; 54class DirectoryLookup; 55class PreprocessingRecord; 56class ModuleLoader; 57 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. 62/// 63class Preprocessor : public RefCountedBase<Preprocessor> { 64 DiagnosticsEngine *Diags; 65 LangOptions &LangOpts; 66 const TargetInfo *Target; 67 FileManager &FileMgr; 68 SourceManager &SourceMgr; 69 ScratchBuffer *ScratchBuf; 70 HeaderSearch &HeaderInfo; 71 ModuleLoader &TheModuleLoader; 72 73 /// \brief External source of macros. 74 ExternalPreprocessorSource *ExternalSource; 75 76 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; 80 81 /// BP - A BumpPtrAllocator object used to quickly allocate and release 82 /// objects internal to the Preprocessor. 83 llvm::BumpPtrAllocator BP; 84 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 101 102 SourceLocation DATELoc, TIMELoc; 103 unsigned CounterValue; // Next __COUNTER__ value. 104 105 enum { 106 /// MaxIncludeStackDepth - Maximum depth of #includes. 107 MaxAllowedIncludeStackDepth = 200 108 }; 109 110 // State that is set before the preprocessor begins. 111 bool KeepComments : 1; 112 bool KeepMacroComments : 1; 113 bool SuppressIncludeNotFoundError : 1; 114 115 // State that changes while the preprocessor runs: 116 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 117 118 /// Whether the preprocessor owns the header search object. 119 bool OwnsHeaderSearch : 1; 120 121 /// DisableMacroExpansion - True if macro expansion is disabled. 122 bool DisableMacroExpansion : 1; 123 124 /// MacroExpansionInDirectivesOverride - Temporarily disables 125 /// DisableMacroExpansion (i.e. enables expansion) when parsing preprocessor 126 /// directives. 127 bool MacroExpansionInDirectivesOverride : 1; 128 129 class ResetMacroExpansionHelper; 130 131 /// \brief Whether we have already loaded macros from the external source. 132 mutable bool ReadMacrosFromExternalSource : 1; 133 134 /// \brief True if pragmas are enabled. 135 bool PragmasEnabled : 1; 136 137 /// \brief True if we are pre-expanding macro arguments. 138 bool InMacroArgPreExpansion; 139 140 /// Identifiers - This is mapping/lookup information for all identifiers in 141 /// the program, including program keywords. 142 mutable IdentifierTable Identifiers; 143 144 /// Selectors - This table contains all the selectors in the program. Unlike 145 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 146 /// It is declared/expanded here because it's role/lifetime is 147 /// conceptually similar the IdentifierTable. In addition, the current control 148 /// flow (in clang::ParseAST()), make it convenient to put here. 149 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 150 /// the lifetime of the preprocessor. 151 SelectorTable Selectors; 152 153 /// BuiltinInfo - Information about builtins. 154 Builtin::Context BuiltinInfo; 155 156 /// PragmaHandlers - This tracks all of the pragmas that the client registered 157 /// with this preprocessor. 158 PragmaNamespace *PragmaHandlers; 159 160 /// \brief Tracks all of the comment handlers that the client registered 161 /// with this preprocessor. 162 std::vector<CommentHandler *> CommentHandlers; 163 164 /// \brief True if we want to ignore EOF token and continue later on (thus 165 /// avoid tearing the Lexer and etc. down). 166 bool IncrementalProcessing; 167 168 /// \brief The code-completion handler. 169 CodeCompletionHandler *CodeComplete; 170 171 /// \brief The file that we're performing code-completion for, if any. 172 const FileEntry *CodeCompletionFile; 173 174 /// \brief The offset in file for the code-completion point. 175 unsigned CodeCompletionOffset; 176 177 /// \brief The location for the code-completion point. This gets instantiated 178 /// when the CodeCompletionFile gets #include'ed for preprocessing. 179 SourceLocation CodeCompletionLoc; 180 181 /// \brief The start location for the file of the code-completion point. 182 /// This gets instantiated when the CodeCompletionFile gets #include'ed 183 /// for preprocessing. 184 SourceLocation CodeCompletionFileLoc; 185 186 /// \brief The source location of the 'import' contextual keyword we just 187 /// lexed, if any. 188 SourceLocation ModuleImportLoc; 189 190 /// \brief The module import path that we're currently processing. 191 llvm::SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> 192 ModuleImportPath; 193 194 /// \brief Whether the module import expectes an identifier next. Otherwise, 195 /// it expects a '.' or ';'. 196 bool ModuleImportExpectsIdentifier; 197 198 /// \brief The source location of the currently-active 199 /// #pragma clang arc_cf_code_audited begin. 200 SourceLocation PragmaARCCFCodeAuditedLoc; 201 202 /// \brief True if we hit the code-completion point. 203 bool CodeCompletionReached; 204 205 /// \brief The number of bytes that we will initially skip when entering the 206 /// main file, which is used when loading a precompiled preamble, along 207 /// with a flag that indicates whether skipping this number of bytes will 208 /// place the lexer at the start of a line. 209 std::pair<unsigned, bool> SkipMainFilePreamble; 210 211 /// CurLexer - This is the current top of the stack that we're lexing from if 212 /// not expanding a macro and we are lexing directly from source code. 213 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 214 OwningPtr<Lexer> CurLexer; 215 216 /// CurPTHLexer - This is the current top of stack that we're lexing from if 217 /// not expanding from a macro and we are lexing from a PTH cache. 218 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 219 OwningPtr<PTHLexer> CurPTHLexer; 220 221 /// CurPPLexer - This is the current top of the stack what we're lexing from 222 /// if not expanding a macro. This is an alias for either CurLexer or 223 /// CurPTHLexer. 224 PreprocessorLexer *CurPPLexer; 225 226 /// CurLookup - The DirectoryLookup structure used to find the current 227 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 228 /// implement #include_next and find directory-specific properties. 229 const DirectoryLookup *CurDirLookup; 230 231 /// CurTokenLexer - This is the current macro we are expanding, if we are 232 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 233 OwningPtr<TokenLexer> CurTokenLexer; 234 235 /// \brief The kind of lexer we're currently working with. 236 enum CurLexerKind { 237 CLK_Lexer, 238 CLK_PTHLexer, 239 CLK_TokenLexer, 240 CLK_CachingLexer, 241 CLK_LexAfterModuleImport 242 } CurLexerKind; 243 244 /// IncludeMacroStack - This keeps track of the stack of files currently 245 /// #included, and macros currently being expanded from, not counting 246 /// CurLexer/CurTokenLexer. 247 struct IncludeStackInfo { 248 enum CurLexerKind CurLexerKind; 249 Lexer *TheLexer; 250 PTHLexer *ThePTHLexer; 251 PreprocessorLexer *ThePPLexer; 252 TokenLexer *TheTokenLexer; 253 const DirectoryLookup *TheDirLookup; 254 255 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 256 PreprocessorLexer* PPL, 257 TokenLexer* TL, const DirectoryLookup *D) 258 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 259 TheTokenLexer(TL), TheDirLookup(D) {} 260 }; 261 std::vector<IncludeStackInfo> IncludeMacroStack; 262 263 /// Callbacks - These are actions invoked when some preprocessor activity is 264 /// encountered (e.g. a file is #included, etc). 265 PPCallbacks *Callbacks; 266 267 struct MacroExpandsInfo { 268 Token Tok; 269 MacroInfo *MI; 270 SourceRange Range; 271 MacroExpandsInfo(Token Tok, MacroInfo *MI, SourceRange Range) 272 : Tok(Tok), MI(MI), Range(Range) { } 273 }; 274 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 275 276 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 277 /// to the actual definition of the macro. 278 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 279 280 /// \brief Macros that we want to warn because they are not used at the end 281 /// of the translation unit; we store just their SourceLocations instead 282 /// something like MacroInfo*. The benefit of this is that when we are 283 /// deserializing from PCH, we don't need to deserialize identifier & macros 284 /// just so that we can report that they are unused, we just warn using 285 /// the SourceLocations of this set (that will be filled by the ASTReader). 286 /// We are using SmallPtrSet instead of a vector for faster removal. 287 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 288 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 289 290 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 291 /// reused for quick allocation. 292 MacroArgs *MacroArgCache; 293 friend class MacroArgs; 294 295 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 296 /// push_macro directive, we keep a MacroInfo stack used to restore 297 /// previous macro value. 298 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 299 300 // Various statistics we track for performance analysis. 301 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 302 unsigned NumIf, NumElse, NumEndif; 303 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 304 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 305 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 306 unsigned NumSkipped; 307 308 /// Predefines - This string is the predefined macros that preprocessor 309 /// should use from the command line etc. 310 std::string Predefines; 311 312 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 313 enum { TokenLexerCacheSize = 8 }; 314 unsigned NumCachedTokenLexers; 315 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 316 317 /// \brief Keeps macro expanded tokens for TokenLexers. 318 // 319 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 320 /// going to lex in the cache and when it finishes the tokens are removed 321 /// from the end of the cache. 322 SmallVector<Token, 16> MacroExpandedTokens; 323 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 324 325 /// \brief A record of the macro definitions and expansions that 326 /// occurred during preprocessing. 327 /// 328 /// This is an optional side structure that can be enabled with 329 /// \c createPreprocessingRecord() prior to preprocessing. 330 PreprocessingRecord *Record; 331 332private: // Cached tokens state. 333 typedef SmallVector<Token, 1> CachedTokensTy; 334 335 /// CachedTokens - Cached tokens are stored here when we do backtracking or 336 /// lookahead. They are "lexed" by the CachingLex() method. 337 CachedTokensTy CachedTokens; 338 339 /// CachedLexPos - The position of the cached token that CachingLex() should 340 /// "lex" next. If it points beyond the CachedTokens vector, it means that 341 /// a normal Lex() should be invoked. 342 CachedTokensTy::size_type CachedLexPos; 343 344 /// BacktrackPositions - Stack of backtrack positions, allowing nested 345 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 346 /// indicate where CachedLexPos should be set when the BackTrack() method is 347 /// invoked (at which point the last position is popped). 348 std::vector<CachedTokensTy::size_type> BacktrackPositions; 349 350 struct MacroInfoChain { 351 MacroInfo MI; 352 MacroInfoChain *Next; 353 MacroInfoChain *Prev; 354 }; 355 356 /// MacroInfos are managed as a chain for easy disposal. This is the head 357 /// of that list. 358 MacroInfoChain *MIChainHead; 359 360 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 361 /// allocation. 362 MacroInfoChain *MICache; 363 364 MacroInfo *getInfoForMacro(IdentifierInfo *II) const; 365 366public: 367 Preprocessor(DiagnosticsEngine &diags, LangOptions &opts, 368 const TargetInfo *target, 369 SourceManager &SM, HeaderSearch &Headers, 370 ModuleLoader &TheModuleLoader, 371 IdentifierInfoLookup *IILookup = 0, 372 bool OwnsHeaderSearch = false, 373 bool DelayInitialization = false, 374 bool IncrProcessing = false); 375 376 ~Preprocessor(); 377 378 /// \brief Initialize the preprocessor, if the constructor did not already 379 /// perform the initialization. 380 /// 381 /// \param Target Information about the target. 382 void Initialize(const TargetInfo &Target); 383 384 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 385 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 386 387 const LangOptions &getLangOpts() const { return LangOpts; } 388 const TargetInfo &getTargetInfo() const { return *Target; } 389 FileManager &getFileManager() const { return FileMgr; } 390 SourceManager &getSourceManager() const { return SourceMgr; } 391 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 392 393 IdentifierTable &getIdentifierTable() { return Identifiers; } 394 SelectorTable &getSelectorTable() { return Selectors; } 395 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 396 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 397 398 void setPTHManager(PTHManager* pm); 399 400 PTHManager *getPTHManager() { return PTH.get(); } 401 402 void setExternalSource(ExternalPreprocessorSource *Source) { 403 ExternalSource = Source; 404 } 405 406 ExternalPreprocessorSource *getExternalSource() const { 407 return ExternalSource; 408 } 409 410 /// \brief Retrieve the module loader associated with this preprocessor. 411 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 412 413 /// SetCommentRetentionState - Control whether or not the preprocessor retains 414 /// comments in output. 415 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 416 this->KeepComments = KeepComments | KeepMacroComments; 417 this->KeepMacroComments = KeepMacroComments; 418 } 419 420 bool getCommentRetentionState() const { return KeepComments; } 421 422 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 423 bool getPragmasEnabled() const { return PragmasEnabled; } 424 425 void SetSuppressIncludeNotFoundError(bool Suppress) { 426 SuppressIncludeNotFoundError = Suppress; 427 } 428 429 bool GetSuppressIncludeNotFoundError() { 430 return SuppressIncludeNotFoundError; 431 } 432 433 /// isCurrentLexer - Return true if we are lexing directly from the specified 434 /// lexer. 435 bool isCurrentLexer(const PreprocessorLexer *L) const { 436 return CurPPLexer == L; 437 } 438 439 /// getCurrentLexer - Return the current lexer being lexed from. Note 440 /// that this ignores any potentially active macro expansions and _Pragma 441 /// expansions going on at the time. 442 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 443 444 /// getCurrentFileLexer - Return the current file lexer being lexed from. 445 /// Note that this ignores any potentially active macro expansions and _Pragma 446 /// expansions going on at the time. 447 PreprocessorLexer *getCurrentFileLexer() const; 448 449 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 450 /// Note that this class takes ownership of any PPCallbacks object given to 451 /// it. 452 PPCallbacks *getPPCallbacks() const { return Callbacks; } 453 void addPPCallbacks(PPCallbacks *C) { 454 if (Callbacks) 455 C = new PPChainedCallbacks(C, Callbacks); 456 Callbacks = C; 457 } 458 459 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 460 /// or null if it isn't #define'd. 461 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 462 if (!II->hasMacroDefinition()) 463 return 0; 464 465 return getInfoForMacro(II); 466 } 467 468 /// setMacroInfo - Specify a macro for this identifier. 469 /// 470 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI, 471 bool LoadedFromAST = false); 472 473 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 474 /// state of the macro table. This visits every currently-defined macro. 475 typedef llvm::DenseMap<IdentifierInfo*, 476 MacroInfo*>::const_iterator macro_iterator; 477 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 478 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 479 480 const std::string &getPredefines() const { return Predefines; } 481 /// setPredefines - Set the predefines for this Preprocessor. These 482 /// predefines are automatically injected when parsing the main file. 483 void setPredefines(const char *P) { Predefines = P; } 484 void setPredefines(const std::string &P) { Predefines = P; } 485 486 /// getIdentifierInfo - Return information about the specified preprocessor 487 /// identifier token. The version of this method that takes two character 488 /// pointers is preferred unless the identifier is already available as a 489 /// string (this avoids allocation and copying of memory to construct an 490 /// std::string). 491 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 492 return &Identifiers.get(Name); 493 } 494 495 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 496 /// If 'Namespace' is non-null, then it is a token required to exist on the 497 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 498 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 499 void AddPragmaHandler(PragmaHandler *Handler) { 500 AddPragmaHandler(StringRef(), Handler); 501 } 502 503 /// RemovePragmaHandler - Remove the specific pragma handler from 504 /// the preprocessor. If \arg Namespace is non-null, then it should 505 /// be the namespace that \arg Handler was added to. It is an error 506 /// to remove a handler that has not been registered. 507 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 508 void RemovePragmaHandler(PragmaHandler *Handler) { 509 RemovePragmaHandler(StringRef(), Handler); 510 } 511 512 /// \brief Add the specified comment handler to the preprocessor. 513 void AddCommentHandler(CommentHandler *Handler); 514 515 /// \brief Remove the specified comment handler. 516 /// 517 /// It is an error to remove a handler that has not been registered. 518 void RemoveCommentHandler(CommentHandler *Handler); 519 520 /// \brief Set the code completion handler to the given object. 521 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 522 CodeComplete = &Handler; 523 } 524 525 /// \brief Retrieve the current code-completion handler. 526 CodeCompletionHandler *getCodeCompletionHandler() const { 527 return CodeComplete; 528 } 529 530 /// \brief Clear out the code completion handler. 531 void clearCodeCompletionHandler() { 532 CodeComplete = 0; 533 } 534 535 /// \brief Hook used by the lexer to invoke the "natural language" code 536 /// completion point. 537 void CodeCompleteNaturalLanguage(); 538 539 /// \brief Retrieve the preprocessing record, or NULL if there is no 540 /// preprocessing record. 541 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 542 543 /// \brief Create a new preprocessing record, which will keep track of 544 /// all macro expansions, macro definitions, etc. 545 void createPreprocessingRecord(bool RecordConditionalDirectives); 546 547 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 548 /// which implicitly adds the builtin defines etc. 549 void EnterMainSourceFile(); 550 551 /// EndSourceFile - Inform the preprocessor callbacks that processing is 552 /// complete. 553 void EndSourceFile(); 554 555 /// EnterSourceFile - Add a source file to the top of the include stack and 556 /// start lexing tokens from it instead of the current buffer. Emit an error 557 /// and don't enter the file on error. 558 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 559 SourceLocation Loc); 560 561 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 562 /// tokens from it instead of the current buffer. Args specifies the 563 /// tokens input to a function-like macro. 564 /// 565 /// ILEnd specifies the location of the ')' for a function-like macro or the 566 /// identifier for an object-like macro. 567 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 568 569 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 570 /// which will cause the lexer to start returning the specified tokens. 571 /// 572 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 573 /// not be subject to further macro expansion. Otherwise, these tokens will 574 /// be re-macro-expanded when/if expansion is enabled. 575 /// 576 /// If OwnsTokens is false, this method assumes that the specified stream of 577 /// tokens has a permanent owner somewhere, so they do not need to be copied. 578 /// If it is true, it assumes the array of tokens is allocated with new[] and 579 /// must be freed. 580 /// 581 void EnterTokenStream(const Token *Toks, unsigned NumToks, 582 bool DisableMacroExpansion, bool OwnsTokens); 583 584 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 585 /// lexer stack. This should only be used in situations where the current 586 /// state of the top-of-stack lexer is known. 587 void RemoveTopOfLexerStack(); 588 589 /// EnableBacktrackAtThisPos - From the point that this method is called, and 590 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 591 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 592 /// make the Preprocessor re-lex the same tokens. 593 /// 594 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 595 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 596 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 597 /// 598 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 599 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 600 /// tokens will continue indefinitely. 601 /// 602 void EnableBacktrackAtThisPos(); 603 604 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 605 void CommitBacktrackedTokens(); 606 607 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 608 /// EnableBacktrackAtThisPos() was previously called. 609 void Backtrack(); 610 611 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 612 /// caching of tokens is on. 613 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 614 615 /// Lex - To lex a token from the preprocessor, just pull a token from the 616 /// current lexer or macro object. 617 void Lex(Token &Result) { 618 switch (CurLexerKind) { 619 case CLK_Lexer: CurLexer->Lex(Result); break; 620 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 621 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 622 case CLK_CachingLexer: CachingLex(Result); break; 623 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 624 } 625 } 626 627 void LexAfterModuleImport(Token &Result); 628 629 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 630 /// something not a comment. This is useful in -E -C mode where comments 631 /// would foul up preprocessor directive handling. 632 void LexNonComment(Token &Result) { 633 do 634 Lex(Result); 635 while (Result.getKind() == tok::comment); 636 } 637 638 /// LexUnexpandedToken - This is just like Lex, but this disables macro 639 /// expansion of identifier tokens. 640 void LexUnexpandedToken(Token &Result) { 641 // Disable macro expansion. 642 bool OldVal = DisableMacroExpansion; 643 DisableMacroExpansion = true; 644 // Lex the token. 645 Lex(Result); 646 647 // Reenable it. 648 DisableMacroExpansion = OldVal; 649 } 650 651 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 652 /// expansion of identifier tokens. 653 void LexUnexpandedNonComment(Token &Result) { 654 do 655 LexUnexpandedToken(Result); 656 while (Result.getKind() == tok::comment); 657 } 658 659 /// Disables macro expansion everywhere except for preprocessor directives. 660 void SetMacroExpansionOnlyInDirectives() { 661 DisableMacroExpansion = true; 662 MacroExpansionInDirectivesOverride = true; 663 } 664 665 /// LookAhead - This peeks ahead N tokens and returns that token without 666 /// consuming any tokens. LookAhead(0) returns the next token that would be 667 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 668 /// returns normal tokens after phase 5. As such, it is equivalent to using 669 /// 'Lex', not 'LexUnexpandedToken'. 670 const Token &LookAhead(unsigned N) { 671 if (CachedLexPos + N < CachedTokens.size()) 672 return CachedTokens[CachedLexPos+N]; 673 else 674 return PeekAhead(N+1); 675 } 676 677 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 678 /// this allows to revert a specific number of tokens. 679 /// Note that the number of tokens being reverted should be up to the last 680 /// backtrack position, not more. 681 void RevertCachedTokens(unsigned N) { 682 assert(isBacktrackEnabled() && 683 "Should only be called when tokens are cached for backtracking"); 684 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 685 && "Should revert tokens up to the last backtrack position, not more"); 686 assert(signed(CachedLexPos) - signed(N) >= 0 && 687 "Corrupted backtrack positions ?"); 688 CachedLexPos -= N; 689 } 690 691 /// EnterToken - Enters a token in the token stream to be lexed next. If 692 /// BackTrack() is called afterwards, the token will remain at the insertion 693 /// point. 694 void EnterToken(const Token &Tok) { 695 EnterCachingLexMode(); 696 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 697 } 698 699 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 700 /// tokens (because backtrack is enabled) it should replace the most recent 701 /// cached tokens with the given annotation token. This function has no effect 702 /// if backtracking is not enabled. 703 /// 704 /// Note that the use of this function is just for optimization; so that the 705 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 706 /// invoked. 707 void AnnotateCachedTokens(const Token &Tok) { 708 assert(Tok.isAnnotation() && "Expected annotation token"); 709 if (CachedLexPos != 0 && isBacktrackEnabled()) 710 AnnotatePreviousCachedTokens(Tok); 711 } 712 713 /// \brief Replace the last token with an annotation token. 714 /// 715 /// Like AnnotateCachedTokens(), this routine replaces an 716 /// already-parsed (and resolved) token with an annotation 717 /// token. However, this routine only replaces the last token with 718 /// the annotation token; it does not affect any other cached 719 /// tokens. This function has no effect if backtracking is not 720 /// enabled. 721 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 722 assert(Tok.isAnnotation() && "Expected annotation token"); 723 if (CachedLexPos != 0 && isBacktrackEnabled()) 724 CachedTokens[CachedLexPos-1] = Tok; 725 } 726 727 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 728 /// CurTokenLexer pointers. 729 void recomputeCurLexerKind(); 730 731 /// \brief Returns true if incremental processing is enabled 732 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 733 734 /// \brief Enables the incremental processing 735 void enableIncrementalProcessing(bool value = true) { 736 IncrementalProcessing = value; 737 } 738 739 /// \brief Specify the point at which code-completion will be performed. 740 /// 741 /// \param File the file in which code completion should occur. If 742 /// this file is included multiple times, code-completion will 743 /// perform completion the first time it is included. If NULL, this 744 /// function clears out the code-completion point. 745 /// 746 /// \param Line the line at which code completion should occur 747 /// (1-based). 748 /// 749 /// \param Column the column at which code completion should occur 750 /// (1-based). 751 /// 752 /// \returns true if an error occurred, false otherwise. 753 bool SetCodeCompletionPoint(const FileEntry *File, 754 unsigned Line, unsigned Column); 755 756 /// \brief Determine if we are performing code completion. 757 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 758 759 /// \brief Returns the location of the code-completion point. 760 /// Returns an invalid location if code-completion is not enabled or the file 761 /// containing the code-completion point has not been lexed yet. 762 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 763 764 /// \brief Returns the start location of the file of code-completion point. 765 /// Returns an invalid location if code-completion is not enabled or the file 766 /// containing the code-completion point has not been lexed yet. 767 SourceLocation getCodeCompletionFileLoc() const { 768 return CodeCompletionFileLoc; 769 } 770 771 /// \brief Returns true if code-completion is enabled and we have hit the 772 /// code-completion point. 773 bool isCodeCompletionReached() const { return CodeCompletionReached; } 774 775 /// \brief Note that we hit the code-completion point. 776 void setCodeCompletionReached() { 777 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 778 CodeCompletionReached = true; 779 // Silence any diagnostics that occur after we hit the code-completion. 780 getDiagnostics().setSuppressAllDiagnostics(true); 781 } 782 783 /// \brief The location of the currently-active #pragma clang 784 /// arc_cf_code_audited begin. Returns an invalid location if there 785 /// is no such pragma active. 786 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 787 return PragmaARCCFCodeAuditedLoc; 788 } 789 790 /// \brief Set the location of the currently-active #pragma clang 791 /// arc_cf_code_audited begin. An invalid location ends the pragma. 792 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 793 PragmaARCCFCodeAuditedLoc = Loc; 794 } 795 796 /// \brief Instruct the preprocessor to skip part of the main 797 /// the main source file. 798 /// 799 /// \brief Bytes The number of bytes in the preamble to skip. 800 /// 801 /// \brief StartOfLine Whether skipping these bytes puts the lexer at the 802 /// start of a line. 803 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 804 SkipMainFilePreamble.first = Bytes; 805 SkipMainFilePreamble.second = StartOfLine; 806 } 807 808 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 809 /// the specified Token's location, translating the token's start 810 /// position in the current buffer into a SourcePosition object for rendering. 811 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 812 return Diags->Report(Loc, DiagID); 813 } 814 815 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 816 return Diags->Report(Tok.getLocation(), DiagID); 817 } 818 819 /// getSpelling() - Return the 'spelling' of the token at the given 820 /// location; does not go up to the spelling location or down to the 821 /// expansion location. 822 /// 823 /// \param buffer A buffer which will be used only if the token requires 824 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 825 /// \param invalid If non-null, will be set \c true if an error occurs. 826 StringRef getSpelling(SourceLocation loc, 827 SmallVectorImpl<char> &buffer, 828 bool *invalid = 0) const { 829 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 830 } 831 832 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 833 /// token is the characters used to represent the token in the source file 834 /// after trigraph expansion and escaped-newline folding. In particular, this 835 /// wants to get the true, uncanonicalized, spelling of things like digraphs 836 /// UCNs, etc. 837 /// 838 /// \param Invalid If non-null, will be set \c true if an error occurs. 839 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 840 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 841 } 842 843 /// getSpelling - This method is used to get the spelling of a token into a 844 /// preallocated buffer, instead of as an std::string. The caller is required 845 /// to allocate enough space for the token, which is guaranteed to be at least 846 /// Tok.getLength() bytes long. The length of the actual result is returned. 847 /// 848 /// Note that this method may do two possible things: it may either fill in 849 /// the buffer specified with characters, or it may *change the input pointer* 850 /// to point to a constant buffer with the data already in it (avoiding a 851 /// copy). The caller is not allowed to modify the returned buffer pointer 852 /// if an internal buffer is returned. 853 unsigned getSpelling(const Token &Tok, const char *&Buffer, 854 bool *Invalid = 0) const { 855 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 856 } 857 858 /// getSpelling - This method is used to get the spelling of a token into a 859 /// SmallVector. Note that the returned StringRef may not point to the 860 /// supplied buffer if a copy can be avoided. 861 StringRef getSpelling(const Token &Tok, 862 SmallVectorImpl<char> &Buffer, 863 bool *Invalid = 0) const; 864 865 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 866 /// with length 1, return the character. 867 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 868 bool *Invalid = 0) const { 869 assert(Tok.is(tok::numeric_constant) && 870 Tok.getLength() == 1 && "Called on unsupported token"); 871 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 872 873 // If the token is carrying a literal data pointer, just use it. 874 if (const char *D = Tok.getLiteralData()) 875 return *D; 876 877 // Otherwise, fall back on getCharacterData, which is slower, but always 878 // works. 879 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 880 } 881 882 /// \brief Retrieve the name of the immediate macro expansion. 883 /// 884 /// This routine starts from a source location, and finds the name of the macro 885 /// responsible for its immediate expansion. It looks through any intervening 886 /// macro argument expansions to compute this. It returns a StringRef which 887 /// refers to the SourceManager-owned buffer of the source where that macro 888 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 889 StringRef getImmediateMacroName(SourceLocation Loc) { 890 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 891 } 892 893 /// CreateString - Plop the specified string into a scratch buffer and set the 894 /// specified token's location and length to it. If specified, the source 895 /// location provides a location of the expansion point of the token. 896 void CreateString(const char *Buf, unsigned Len, Token &Tok, 897 SourceLocation ExpansionLocStart = SourceLocation(), 898 SourceLocation ExpansionLocEnd = SourceLocation()); 899 900 /// \brief Computes the source location just past the end of the 901 /// token at this source location. 902 /// 903 /// This routine can be used to produce a source location that 904 /// points just past the end of the token referenced by \p Loc, and 905 /// is generally used when a diagnostic needs to point just after a 906 /// token where it expected something different that it received. If 907 /// the returned source location would not be meaningful (e.g., if 908 /// it points into a macro), this routine returns an invalid 909 /// source location. 910 /// 911 /// \param Offset an offset from the end of the token, where the source 912 /// location should refer to. The default offset (0) produces a source 913 /// location pointing just past the end of the token; an offset of 1 produces 914 /// a source location pointing to the last character in the token, etc. 915 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 916 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 917 } 918 919 /// \brief Returns true if the given MacroID location points at the first 920 /// token of the macro expansion. 921 /// 922 /// \param MacroBegin If non-null and function returns true, it is set to 923 /// begin location of the macro. 924 bool isAtStartOfMacroExpansion(SourceLocation loc, 925 SourceLocation *MacroBegin = 0) const { 926 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 927 MacroBegin); 928 } 929 930 /// \brief Returns true if the given MacroID location points at the last 931 /// token of the macro expansion. 932 /// 933 /// \param MacroBegin If non-null and function returns true, it is set to 934 /// end location of the macro. 935 bool isAtEndOfMacroExpansion(SourceLocation loc, 936 SourceLocation *MacroEnd = 0) const { 937 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 938 } 939 940 /// DumpToken - Print the token to stderr, used for debugging. 941 /// 942 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 943 void DumpLocation(SourceLocation Loc) const; 944 void DumpMacro(const MacroInfo &MI) const; 945 946 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 947 /// token, return a new location that specifies a character within the token. 948 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 949 unsigned Char) const { 950 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 951 } 952 953 /// IncrementPasteCounter - Increment the counters for the number of token 954 /// paste operations performed. If fast was specified, this is a 'fast paste' 955 /// case we handled. 956 /// 957 void IncrementPasteCounter(bool isFast) { 958 if (isFast) 959 ++NumFastTokenPaste; 960 else 961 ++NumTokenPaste; 962 } 963 964 void PrintStats(); 965 966 size_t getTotalMemory() const; 967 968 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 969 /// comment (/##/) in microsoft mode, this method handles updating the current 970 /// state, returning the token on the next source line. 971 void HandleMicrosoftCommentPaste(Token &Tok); 972 973 //===--------------------------------------------------------------------===// 974 // Preprocessor callback methods. These are invoked by a lexer as various 975 // directives and events are found. 976 977 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 978 /// identifier information for the token and install it into the token, 979 /// updating the token kind accordingly. 980 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 981 982private: 983 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 984 985public: 986 987 // SetPoisonReason - Call this function to indicate the reason for 988 // poisoning an identifier. If that identifier is accessed while 989 // poisoned, then this reason will be used instead of the default 990 // "poisoned" diagnostic. 991 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 992 993 // HandlePoisonedIdentifier - Display reason for poisoned 994 // identifier. 995 void HandlePoisonedIdentifier(Token & Tok); 996 997 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 998 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 999 if(II->isPoisoned()) { 1000 HandlePoisonedIdentifier(Identifier); 1001 } 1002 } 1003 } 1004 1005private: 1006 /// Identifiers used for SEH handling in Borland. These are only 1007 /// allowed in particular circumstances 1008 // __except block 1009 IdentifierInfo *Ident__exception_code, 1010 *Ident___exception_code, 1011 *Ident_GetExceptionCode; 1012 // __except filter expression 1013 IdentifierInfo *Ident__exception_info, 1014 *Ident___exception_info, 1015 *Ident_GetExceptionInfo; 1016 // __finally 1017 IdentifierInfo *Ident__abnormal_termination, 1018 *Ident___abnormal_termination, 1019 *Ident_AbnormalTermination; 1020public: 1021 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1022 1023 /// HandleIdentifier - This callback is invoked when the lexer reads an 1024 /// identifier and has filled in the tokens IdentifierInfo member. This 1025 /// callback potentially macro expands it or turns it into a named token (like 1026 /// 'for'). 1027 void HandleIdentifier(Token &Identifier); 1028 1029 1030 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1031 /// the current file. This either returns the EOF token and returns true, or 1032 /// pops a level off the include stack and returns false, at which point the 1033 /// client should call lex again. 1034 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1035 1036 /// HandleEndOfTokenLexer - This callback is invoked when the current 1037 /// TokenLexer hits the end of its token stream. 1038 bool HandleEndOfTokenLexer(Token &Result); 1039 1040 /// HandleDirective - This callback is invoked when the lexer sees a # token 1041 /// at the start of a line. This consumes the directive, modifies the 1042 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1043 /// read is the correct one. 1044 void HandleDirective(Token &Result); 1045 1046 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1047 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1048 /// true, then we consider macros that expand to zero tokens as being ok. 1049 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1050 1051 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1052 /// current line until the tok::eod token is found. 1053 void DiscardUntilEndOfDirective(); 1054 1055 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1056 /// __DATE__ or __TIME__ in the file so far. 1057 bool SawDateOrTime() const { 1058 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1059 } 1060 unsigned getCounterValue() const { return CounterValue; } 1061 void setCounterValue(unsigned V) { CounterValue = V; } 1062 1063 /// \brief Retrieves the module that we're currently building, if any. 1064 Module *getCurrentModule(); 1065 1066 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 1067 /// SourceLocation. 1068 MacroInfo *AllocateMacroInfo(SourceLocation L); 1069 1070 /// CloneMacroInfo - Allocate a new MacroInfo object which is clone of MI. 1071 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1072 1073 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 1074 /// checked and spelled filename, e.g. as an operand of #include. This returns 1075 /// true if the input filename was in <>'s or false if it were in ""'s. The 1076 /// caller is expected to provide a buffer that is large enough to hold the 1077 /// spelling of the filename, but is also expected to handle the case when 1078 /// this method decides to use a different buffer. 1079 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1080 1081 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 1082 /// return null on failure. isAngled indicates whether the file reference is 1083 /// for system #include's or not (i.e. using <> instead of ""). 1084 const FileEntry *LookupFile(StringRef Filename, 1085 bool isAngled, const DirectoryLookup *FromDir, 1086 const DirectoryLookup *&CurDir, 1087 SmallVectorImpl<char> *SearchPath, 1088 SmallVectorImpl<char> *RelativePath, 1089 Module **SuggestedModule, 1090 bool SkipCache = false); 1091 1092 /// GetCurLookup - The DirectoryLookup structure used to find the current 1093 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1094 /// implement #include_next and find directory-specific properties. 1095 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1096 1097 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 1098 /// #include. 1099 bool isInPrimaryFile() const; 1100 1101 /// ConcatenateIncludeName - Handle cases where the #include name is expanded 1102 /// from a macro as multiple tokens, which need to be glued together. This 1103 /// occurs for code like: 1104 /// #define FOO <a/b.h> 1105 /// #include FOO 1106 /// because in this case, "<a/b.h>" is returned as 7 tokens, not one. 1107 /// 1108 /// This code concatenates and consumes tokens up to the '>' token. It 1109 /// returns false if the > was found, otherwise it returns true if it finds 1110 /// and consumes the EOD marker. 1111 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1112 SourceLocation &End); 1113 1114 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1115 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1116 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1117 1118private: 1119 1120 void PushIncludeMacroStack() { 1121 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1122 CurLexer.take(), 1123 CurPTHLexer.take(), 1124 CurPPLexer, 1125 CurTokenLexer.take(), 1126 CurDirLookup)); 1127 CurPPLexer = 0; 1128 } 1129 1130 void PopIncludeMacroStack() { 1131 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1132 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1133 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1134 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1135 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1136 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1137 IncludeMacroStack.pop_back(); 1138 } 1139 1140 /// AllocateMacroInfo - Allocate a new MacroInfo object. 1141 MacroInfo *AllocateMacroInfo(); 1142 1143 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 1144 /// be reused for allocating new MacroInfo objects. 1145 void ReleaseMacroInfo(MacroInfo* MI); 1146 1147 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1148 /// #define or #undef. This emits a diagnostic, sets the token kind to eod, 1149 /// and discards the rest of the macro line if the macro name is invalid. 1150 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1151 1152 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1153 /// definition has just been read. Lex the rest of the arguments and the 1154 /// closing ), updating MI with what we learn and saving in LastTok the 1155 /// last token read. 1156 /// Return true if an error occurs parsing the arg list. 1157 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1158 1159 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 1160 /// decided that the subsequent tokens are in the #if'd out portion of the 1161 /// file. Lex the rest of the file, until we see an #endif. If 1162 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1163 /// this #if directive, so #else/#elif blocks should never be entered. If 1164 /// FoundElse is false, then #else directives are ok, if not, then we have 1165 /// already seen one so a #else directive is a duplicate. When this returns, 1166 /// the caller can lex the first valid token. 1167 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1168 bool FoundNonSkipPortion, bool FoundElse, 1169 SourceLocation ElseLoc = SourceLocation()); 1170 1171 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 1172 /// SkipExcludedConditionalBlock. 1173 void PTHSkipExcludedConditionalBlock(); 1174 1175 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1176 /// may occur after a #if or #elif directive and return it as a bool. If the 1177 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1178 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1179 1180 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1181 /// #pragma GCC poison/system_header/dependency and #pragma once. 1182 void RegisterBuiltinPragmas(); 1183 1184 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 1185 /// identifier table. 1186 void RegisterBuiltinMacros(); 1187 1188 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1189 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1190 /// the macro should not be expanded return true, otherwise return false. 1191 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1192 1193 /// \brief Cache macro expanded tokens for TokenLexers. 1194 // 1195 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1196 /// going to lex in the cache and when it finishes the tokens are removed 1197 /// from the end of the cache. 1198 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1199 ArrayRef<Token> tokens); 1200 void removeCachedMacroExpandedTokensOfLastLexer(); 1201 friend void TokenLexer::ExpandFunctionArguments(); 1202 1203 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1204 /// lexed is a '('. If so, consume the token and return true, if not, this 1205 /// method should have no observable side-effect on the lexed tokens. 1206 bool isNextPPTokenLParen(); 1207 1208 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1209 /// invoked to read all of the formal arguments specified for the macro 1210 /// invocation. This returns null on error. 1211 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1212 SourceLocation &ExpansionEnd); 1213 1214 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1215 /// as a builtin macro, handle it and return the next token as 'Tok'. 1216 void ExpandBuiltinMacro(Token &Tok); 1217 1218 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1219 /// return the first token after the directive. The _Pragma token has just 1220 /// been read into 'Tok'. 1221 void Handle_Pragma(Token &Tok); 1222 1223 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1224 /// is not enclosed within a string literal. 1225 void HandleMicrosoft__pragma(Token &Tok); 1226 1227 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1228 /// start lexing tokens from it instead of the current buffer. 1229 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1230 1231 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1232 /// start getting tokens from it using the PTH cache. 1233 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1234 1235 /// IsFileLexer - Returns true if we are lexing from a file and not a 1236 /// pragma or a macro. 1237 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1238 return L ? !L->isPragmaLexer() : P != 0; 1239 } 1240 1241 static bool IsFileLexer(const IncludeStackInfo& I) { 1242 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1243 } 1244 1245 bool IsFileLexer() const { 1246 return IsFileLexer(CurLexer.get(), CurPPLexer); 1247 } 1248 1249 //===--------------------------------------------------------------------===// 1250 // Caching stuff. 1251 void CachingLex(Token &Result); 1252 bool InCachingLexMode() const { 1253 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1254 // that we are past EOF, not that we are in CachingLex mode. 1255 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1256 !IncludeMacroStack.empty(); 1257 } 1258 void EnterCachingLexMode(); 1259 void ExitCachingLexMode() { 1260 if (InCachingLexMode()) 1261 RemoveTopOfLexerStack(); 1262 } 1263 const Token &PeekAhead(unsigned N); 1264 void AnnotatePreviousCachedTokens(const Token &Tok); 1265 1266 //===--------------------------------------------------------------------===// 1267 /// Handle*Directive - implement the various preprocessor directives. These 1268 /// should side-effect the current preprocessor object so that the next call 1269 /// to Lex() will return the appropriate token next. 1270 void HandleLineDirective(Token &Tok); 1271 void HandleDigitDirective(Token &Tok); 1272 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1273 void HandleIdentSCCSDirective(Token &Tok); 1274 void HandleMacroPublicDirective(Token &Tok); 1275 void HandleMacroPrivateDirective(Token &Tok); 1276 1277 // File inclusion. 1278 void HandleIncludeDirective(SourceLocation HashLoc, 1279 Token &Tok, 1280 const DirectoryLookup *LookupFrom = 0, 1281 bool isImport = false); 1282 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1283 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1284 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1285 void HandleMicrosoftImportDirective(Token &Tok); 1286 1287 // Macro handling. 1288 void HandleDefineDirective(Token &Tok); 1289 void HandleUndefDirective(Token &Tok); 1290 1291 // Conditional Inclusion. 1292 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1293 bool ReadAnyTokensBeforeDirective); 1294 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1295 void HandleEndifDirective(Token &Tok); 1296 void HandleElseDirective(Token &Tok); 1297 void HandleElifDirective(Token &Tok); 1298 1299 // Pragmas. 1300 void HandlePragmaDirective(unsigned Introducer); 1301public: 1302 void HandlePragmaOnce(Token &OnceTok); 1303 void HandlePragmaMark(); 1304 void HandlePragmaPoison(Token &PoisonTok); 1305 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1306 void HandlePragmaDependency(Token &DependencyTok); 1307 void HandlePragmaComment(Token &CommentTok); 1308 void HandlePragmaMessage(Token &MessageTok); 1309 void HandlePragmaPushMacro(Token &Tok); 1310 void HandlePragmaPopMacro(Token &Tok); 1311 void HandlePragmaIncludeAlias(Token &Tok); 1312 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1313 1314 // Return true and store the first token only if any CommentHandler 1315 // has inserted some tokens and getCommentRetentionState() is false. 1316 bool HandleComment(Token &Token, SourceRange Comment); 1317 1318 /// \brief A macro is used, update information about macros that need unused 1319 /// warnings. 1320 void markMacroAsUsed(MacroInfo *MI); 1321}; 1322 1323/// \brief Abstract base class that describes a handler that will receive 1324/// source ranges for each of the comments encountered in the source file. 1325class CommentHandler { 1326public: 1327 virtual ~CommentHandler(); 1328 1329 // The handler shall return true if it has pushed any tokens 1330 // to be read using e.g. EnterToken or EnterTokenStream. 1331 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1332}; 1333 1334} // end namespace clang 1335 1336#endif 1337