Preprocessor.h revision 4d7e0ced7f16a04aabe2d8d91cbbb52fb1162810
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 IdentifierInfo *Ident__building_module; // __building_module 102 IdentifierInfo *Ident__MODULE__; // __MODULE__ 103 104 SourceLocation DATELoc, TIMELoc; 105 unsigned CounterValue; // Next __COUNTER__ value. 106 107 enum { 108 /// MaxIncludeStackDepth - Maximum depth of \#includes. 109 MaxAllowedIncludeStackDepth = 200 110 }; 111 112 // State that is set before the preprocessor begins. 113 bool KeepComments : 1; 114 bool KeepMacroComments : 1; 115 bool SuppressIncludeNotFoundError : 1; 116 117 // State that changes while the preprocessor runs: 118 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 119 120 /// Whether the preprocessor owns the header search object. 121 bool OwnsHeaderSearch : 1; 122 123 /// DisableMacroExpansion - True if macro expansion is disabled. 124 bool DisableMacroExpansion : 1; 125 126 /// MacroExpansionInDirectivesOverride - Temporarily disables 127 /// DisableMacroExpansion (i.e. enables expansion) when parsing preprocessor 128 /// directives. 129 bool MacroExpansionInDirectivesOverride : 1; 130 131 class ResetMacroExpansionHelper; 132 133 /// \brief Whether we have already loaded macros from the external source. 134 mutable bool ReadMacrosFromExternalSource : 1; 135 136 /// \brief True if pragmas are enabled. 137 bool PragmasEnabled : 1; 138 139 /// \brief True if we are pre-expanding macro arguments. 140 bool InMacroArgPreExpansion; 141 142 /// Identifiers - This is mapping/lookup information for all identifiers in 143 /// the program, including program keywords. 144 mutable IdentifierTable Identifiers; 145 146 /// Selectors - This table contains all the selectors in the program. Unlike 147 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 148 /// It is declared/expanded here because it's role/lifetime is 149 /// conceptually similar the IdentifierTable. In addition, the current control 150 /// flow (in clang::ParseAST()), make it convenient to put here. 151 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 152 /// the lifetime of the preprocessor. 153 SelectorTable Selectors; 154 155 /// BuiltinInfo - Information about builtins. 156 Builtin::Context BuiltinInfo; 157 158 /// PragmaHandlers - This tracks all of the pragmas that the client registered 159 /// with this preprocessor. 160 PragmaNamespace *PragmaHandlers; 161 162 /// \brief Tracks all of the comment handlers that the client registered 163 /// with this preprocessor. 164 std::vector<CommentHandler *> CommentHandlers; 165 166 /// \brief True if we want to ignore EOF token and continue later on (thus 167 /// avoid tearing the Lexer and etc. down). 168 bool IncrementalProcessing; 169 170 /// \brief The code-completion handler. 171 CodeCompletionHandler *CodeComplete; 172 173 /// \brief The file that we're performing code-completion for, if any. 174 const FileEntry *CodeCompletionFile; 175 176 /// \brief The offset in file for the code-completion point. 177 unsigned CodeCompletionOffset; 178 179 /// \brief The location for the code-completion point. This gets instantiated 180 /// when the CodeCompletionFile gets \#include'ed for preprocessing. 181 SourceLocation CodeCompletionLoc; 182 183 /// \brief The start location for the file of the code-completion point. 184 /// 185 /// This gets instantiated when the CodeCompletionFile gets \#include'ed 186 /// for preprocessing. 187 SourceLocation CodeCompletionFileLoc; 188 189 /// \brief The source location of the 'import' contextual keyword we just 190 /// lexed, if any. 191 SourceLocation ModuleImportLoc; 192 193 /// \brief The module import path that we're currently processing. 194 llvm::SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> 195 ModuleImportPath; 196 197 /// \brief Whether the module import expectes an identifier next. Otherwise, 198 /// it expects a '.' or ';'. 199 bool ModuleImportExpectsIdentifier; 200 201 /// \brief The source location of the currently-active 202 /// #pragma clang arc_cf_code_audited begin. 203 SourceLocation PragmaARCCFCodeAuditedLoc; 204 205 /// \brief True if we hit the code-completion point. 206 bool CodeCompletionReached; 207 208 /// \brief The number of bytes that we will initially skip when entering the 209 /// main file, which is used when loading a precompiled preamble, along 210 /// with a flag that indicates whether skipping this number of bytes will 211 /// place the lexer at the start of a line. 212 std::pair<unsigned, bool> SkipMainFilePreamble; 213 214 /// CurLexer - This is the current top of the stack that we're lexing from if 215 /// not expanding a macro and we are lexing directly from source code. 216 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 217 OwningPtr<Lexer> CurLexer; 218 219 /// CurPTHLexer - This is the current top of stack that we're lexing from if 220 /// not expanding from a macro and we are lexing from a PTH cache. 221 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 222 OwningPtr<PTHLexer> CurPTHLexer; 223 224 /// CurPPLexer - This is the current top of the stack what we're lexing from 225 /// if not expanding a macro. This is an alias for either CurLexer or 226 /// CurPTHLexer. 227 PreprocessorLexer *CurPPLexer; 228 229 /// CurLookup - The DirectoryLookup structure used to find the current 230 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 231 /// implement \#include_next and find directory-specific properties. 232 const DirectoryLookup *CurDirLookup; 233 234 /// CurTokenLexer - This is the current macro we are expanding, if we are 235 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 236 OwningPtr<TokenLexer> CurTokenLexer; 237 238 /// \brief The kind of lexer we're currently working with. 239 enum CurLexerKind { 240 CLK_Lexer, 241 CLK_PTHLexer, 242 CLK_TokenLexer, 243 CLK_CachingLexer, 244 CLK_LexAfterModuleImport 245 } CurLexerKind; 246 247 /// IncludeMacroStack - This keeps track of the stack of files currently 248 /// \#included, and macros currently being expanded from, not counting 249 /// CurLexer/CurTokenLexer. 250 struct IncludeStackInfo { 251 enum CurLexerKind CurLexerKind; 252 Lexer *TheLexer; 253 PTHLexer *ThePTHLexer; 254 PreprocessorLexer *ThePPLexer; 255 TokenLexer *TheTokenLexer; 256 const DirectoryLookup *TheDirLookup; 257 258 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 259 PreprocessorLexer* PPL, 260 TokenLexer* TL, const DirectoryLookup *D) 261 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 262 TheTokenLexer(TL), TheDirLookup(D) {} 263 }; 264 std::vector<IncludeStackInfo> IncludeMacroStack; 265 266 /// Callbacks - These are actions invoked when some preprocessor activity is 267 /// encountered (e.g. a file is \#included, etc). 268 PPCallbacks *Callbacks; 269 270 struct MacroExpandsInfo { 271 Token Tok; 272 MacroInfo *MI; 273 SourceRange Range; 274 MacroExpandsInfo(Token Tok, MacroInfo *MI, SourceRange Range) 275 : Tok(Tok), MI(MI), Range(Range) { } 276 }; 277 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 278 279 /// Macros - For each IdentifierInfo that was associated with a macro, we 280 /// keep a mapping to the history of all macro definitions and #undefs in 281 /// the reverse order (the latest one is in the head of the list). 282 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 283 friend class ASTReader; 284 285 /// \brief Macros that we want to warn because they are not used at the end 286 /// of the translation unit; we store just their SourceLocations instead 287 /// something like MacroInfo*. The benefit of this is that when we are 288 /// deserializing from PCH, we don't need to deserialize identifier & macros 289 /// just so that we can report that they are unused, we just warn using 290 /// the SourceLocations of this set (that will be filled by the ASTReader). 291 /// We are using SmallPtrSet instead of a vector for faster removal. 292 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 293 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 294 295 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 296 /// reused for quick allocation. 297 MacroArgs *MacroArgCache; 298 friend class MacroArgs; 299 300 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 301 /// push_macro directive, we keep a MacroInfo stack used to restore 302 /// previous macro value. 303 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 304 305 // Various statistics we track for performance analysis. 306 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 307 unsigned NumIf, NumElse, NumEndif; 308 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 309 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 310 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 311 unsigned NumSkipped; 312 313 /// Predefines - This string is the predefined macros that preprocessor 314 /// should use from the command line etc. 315 std::string Predefines; 316 317 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 318 enum { TokenLexerCacheSize = 8 }; 319 unsigned NumCachedTokenLexers; 320 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 321 322 /// \brief Keeps macro expanded tokens for TokenLexers. 323 // 324 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 325 /// going to lex in the cache and when it finishes the tokens are removed 326 /// from the end of the cache. 327 SmallVector<Token, 16> MacroExpandedTokens; 328 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 329 330 /// \brief A record of the macro definitions and expansions that 331 /// occurred during preprocessing. 332 /// 333 /// This is an optional side structure that can be enabled with 334 /// \c createPreprocessingRecord() prior to preprocessing. 335 PreprocessingRecord *Record; 336 337private: // Cached tokens state. 338 typedef SmallVector<Token, 1> CachedTokensTy; 339 340 /// CachedTokens - Cached tokens are stored here when we do backtracking or 341 /// lookahead. They are "lexed" by the CachingLex() method. 342 CachedTokensTy CachedTokens; 343 344 /// CachedLexPos - The position of the cached token that CachingLex() should 345 /// "lex" next. If it points beyond the CachedTokens vector, it means that 346 /// a normal Lex() should be invoked. 347 CachedTokensTy::size_type CachedLexPos; 348 349 /// BacktrackPositions - Stack of backtrack positions, allowing nested 350 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 351 /// indicate where CachedLexPos should be set when the BackTrack() method is 352 /// invoked (at which point the last position is popped). 353 std::vector<CachedTokensTy::size_type> BacktrackPositions; 354 355 struct MacroInfoChain { 356 MacroInfo MI; 357 MacroInfoChain *Next; 358 MacroInfoChain *Prev; 359 }; 360 361 /// MacroInfos are managed as a chain for easy disposal. This is the head 362 /// of that list. 363 MacroInfoChain *MIChainHead; 364 365 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 366 /// allocation. 367 MacroInfoChain *MICache; 368 369public: 370 Preprocessor(DiagnosticsEngine &diags, LangOptions &opts, 371 const TargetInfo *target, 372 SourceManager &SM, HeaderSearch &Headers, 373 ModuleLoader &TheModuleLoader, 374 IdentifierInfoLookup *IILookup = 0, 375 bool OwnsHeaderSearch = false, 376 bool DelayInitialization = false, 377 bool IncrProcessing = false); 378 379 ~Preprocessor(); 380 381 /// \brief Initialize the preprocessor, if the constructor did not already 382 /// perform the initialization. 383 /// 384 /// \param Target Information about the target. 385 void Initialize(const TargetInfo &Target); 386 387 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 388 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 389 390 const LangOptions &getLangOpts() const { return LangOpts; } 391 const TargetInfo &getTargetInfo() const { return *Target; } 392 FileManager &getFileManager() const { return FileMgr; } 393 SourceManager &getSourceManager() const { return SourceMgr; } 394 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 395 396 IdentifierTable &getIdentifierTable() { return Identifiers; } 397 SelectorTable &getSelectorTable() { return Selectors; } 398 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 399 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 400 401 void setPTHManager(PTHManager* pm); 402 403 PTHManager *getPTHManager() { return PTH.get(); } 404 405 void setExternalSource(ExternalPreprocessorSource *Source) { 406 ExternalSource = Source; 407 } 408 409 ExternalPreprocessorSource *getExternalSource() const { 410 return ExternalSource; 411 } 412 413 /// \brief Retrieve the module loader associated with this preprocessor. 414 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 415 416 /// SetCommentRetentionState - Control whether or not the preprocessor retains 417 /// comments in output. 418 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 419 this->KeepComments = KeepComments | KeepMacroComments; 420 this->KeepMacroComments = KeepMacroComments; 421 } 422 423 bool getCommentRetentionState() const { return KeepComments; } 424 425 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 426 bool getPragmasEnabled() const { return PragmasEnabled; } 427 428 void SetSuppressIncludeNotFoundError(bool Suppress) { 429 SuppressIncludeNotFoundError = Suppress; 430 } 431 432 bool GetSuppressIncludeNotFoundError() { 433 return SuppressIncludeNotFoundError; 434 } 435 436 /// isCurrentLexer - Return true if we are lexing directly from the specified 437 /// lexer. 438 bool isCurrentLexer(const PreprocessorLexer *L) const { 439 return CurPPLexer == L; 440 } 441 442 /// getCurrentLexer - Return the current lexer being lexed from. Note 443 /// that this ignores any potentially active macro expansions and _Pragma 444 /// expansions going on at the time. 445 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 446 447 /// getCurrentFileLexer - Return the current file lexer being lexed from. 448 /// Note that this ignores any potentially active macro expansions and _Pragma 449 /// expansions going on at the time. 450 PreprocessorLexer *getCurrentFileLexer() const; 451 452 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 453 /// Note that this class takes ownership of any PPCallbacks object given to 454 /// it. 455 PPCallbacks *getPPCallbacks() const { return Callbacks; } 456 void addPPCallbacks(PPCallbacks *C) { 457 if (Callbacks) 458 C = new PPChainedCallbacks(C, Callbacks); 459 Callbacks = C; 460 } 461 462 /// \brief Given an identifier, return the MacroInfo it is \#defined to 463 /// or null if it isn't \#define'd. 464 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 465 if (!II->hasMacroDefinition()) 466 return 0; 467 468 MacroInfo *MI = getMacroInfoHistory(II); 469 assert(MI->getUndefLoc().isInvalid() && "Macro is undefined!"); 470 return MI; 471 } 472 473 /// \brief Given an identifier, return the (probably #undef'd) MacroInfo 474 /// representing the most recent macro definition. One can iterate over all 475 /// previous macro definitions from it. This method should only be called for 476 /// identifiers that hadMacroDefinition(). 477 MacroInfo *getMacroInfoHistory(IdentifierInfo *II) const; 478 479 /// \brief Specify a macro for this identifier. 480 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI, 481 bool LoadedFromAST = false); 482 /// \brief Undefine a macro for this identifier. 483 void clearMacroInfo(IdentifierInfo *II); 484 485 /// macro_iterator/macro_begin/macro_end - This allows you to walk the macro 486 /// history table. Currently defined macros have 487 /// IdentifierInfo::hasMacroDefinition() set and an empty 488 /// MacroInfo::getUndefLoc() at the head of the list. 489 typedef llvm::DenseMap<IdentifierInfo*, 490 MacroInfo*>::const_iterator macro_iterator; 491 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 492 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 493 494 const std::string &getPredefines() const { return Predefines; } 495 /// setPredefines - Set the predefines for this Preprocessor. These 496 /// predefines are automatically injected when parsing the main file. 497 void setPredefines(const char *P) { Predefines = P; } 498 void setPredefines(const std::string &P) { Predefines = P; } 499 500 /// getIdentifierInfo - Return information about the specified preprocessor 501 /// identifier token. The version of this method that takes two character 502 /// pointers is preferred unless the identifier is already available as a 503 /// string (this avoids allocation and copying of memory to construct an 504 /// std::string). 505 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 506 return &Identifiers.get(Name); 507 } 508 509 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 510 /// If 'Namespace' is non-null, then it is a token required to exist on the 511 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 512 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 513 void AddPragmaHandler(PragmaHandler *Handler) { 514 AddPragmaHandler(StringRef(), Handler); 515 } 516 517 /// RemovePragmaHandler - Remove the specific pragma handler from 518 /// the preprocessor. If \p Namespace is non-null, then it should 519 /// be the namespace that \p Handler was added to. It is an error 520 /// to remove a handler that has not been registered. 521 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 522 void RemovePragmaHandler(PragmaHandler *Handler) { 523 RemovePragmaHandler(StringRef(), Handler); 524 } 525 526 /// \brief Add the specified comment handler to the preprocessor. 527 void addCommentHandler(CommentHandler *Handler); 528 529 /// \brief Remove the specified comment handler. 530 /// 531 /// It is an error to remove a handler that has not been registered. 532 void removeCommentHandler(CommentHandler *Handler); 533 534 /// \brief Set the code completion handler to the given object. 535 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 536 CodeComplete = &Handler; 537 } 538 539 /// \brief Retrieve the current code-completion handler. 540 CodeCompletionHandler *getCodeCompletionHandler() const { 541 return CodeComplete; 542 } 543 544 /// \brief Clear out the code completion handler. 545 void clearCodeCompletionHandler() { 546 CodeComplete = 0; 547 } 548 549 /// \brief Hook used by the lexer to invoke the "natural language" code 550 /// completion point. 551 void CodeCompleteNaturalLanguage(); 552 553 /// \brief Retrieve the preprocessing record, or NULL if there is no 554 /// preprocessing record. 555 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 556 557 /// \brief Create a new preprocessing record, which will keep track of 558 /// all macro expansions, macro definitions, etc. 559 void createPreprocessingRecord(bool RecordConditionalDirectives); 560 561 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 562 /// which implicitly adds the builtin defines etc. 563 void EnterMainSourceFile(); 564 565 /// EndSourceFile - Inform the preprocessor callbacks that processing is 566 /// complete. 567 void EndSourceFile(); 568 569 /// EnterSourceFile - Add a source file to the top of the include stack and 570 /// start lexing tokens from it instead of the current buffer. Emit an error 571 /// and don't enter the file on error. 572 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 573 SourceLocation Loc); 574 575 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 576 /// tokens from it instead of the current buffer. Args specifies the 577 /// tokens input to a function-like macro. 578 /// 579 /// ILEnd specifies the location of the ')' for a function-like macro or the 580 /// identifier for an object-like macro. 581 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 582 MacroArgs *Args); 583 584 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 585 /// which will cause the lexer to start returning the specified tokens. 586 /// 587 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 588 /// not be subject to further macro expansion. Otherwise, these tokens will 589 /// be re-macro-expanded when/if expansion is enabled. 590 /// 591 /// If OwnsTokens is false, this method assumes that the specified stream of 592 /// tokens has a permanent owner somewhere, so they do not need to be copied. 593 /// If it is true, it assumes the array of tokens is allocated with new[] and 594 /// must be freed. 595 /// 596 void EnterTokenStream(const Token *Toks, unsigned NumToks, 597 bool DisableMacroExpansion, bool OwnsTokens); 598 599 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 600 /// lexer stack. This should only be used in situations where the current 601 /// state of the top-of-stack lexer is known. 602 void RemoveTopOfLexerStack(); 603 604 /// EnableBacktrackAtThisPos - From the point that this method is called, and 605 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 606 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 607 /// make the Preprocessor re-lex the same tokens. 608 /// 609 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 610 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 611 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 612 /// 613 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 614 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 615 /// tokens will continue indefinitely. 616 /// 617 void EnableBacktrackAtThisPos(); 618 619 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 620 void CommitBacktrackedTokens(); 621 622 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 623 /// EnableBacktrackAtThisPos() was previously called. 624 void Backtrack(); 625 626 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 627 /// caching of tokens is on. 628 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 629 630 /// Lex - To lex a token from the preprocessor, just pull a token from the 631 /// current lexer or macro object. 632 void Lex(Token &Result) { 633 switch (CurLexerKind) { 634 case CLK_Lexer: CurLexer->Lex(Result); break; 635 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 636 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 637 case CLK_CachingLexer: CachingLex(Result); break; 638 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 639 } 640 } 641 642 void LexAfterModuleImport(Token &Result); 643 644 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 645 /// something not a comment. This is useful in -E -C mode where comments 646 /// would foul up preprocessor directive handling. 647 void LexNonComment(Token &Result) { 648 do 649 Lex(Result); 650 while (Result.getKind() == tok::comment); 651 } 652 653 /// LexUnexpandedToken - This is just like Lex, but this disables macro 654 /// expansion of identifier tokens. 655 void LexUnexpandedToken(Token &Result) { 656 // Disable macro expansion. 657 bool OldVal = DisableMacroExpansion; 658 DisableMacroExpansion = true; 659 // Lex the token. 660 Lex(Result); 661 662 // Reenable it. 663 DisableMacroExpansion = OldVal; 664 } 665 666 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 667 /// expansion of identifier tokens. 668 void LexUnexpandedNonComment(Token &Result) { 669 do 670 LexUnexpandedToken(Result); 671 while (Result.getKind() == tok::comment); 672 } 673 674 /// Disables macro expansion everywhere except for preprocessor directives. 675 void SetMacroExpansionOnlyInDirectives() { 676 DisableMacroExpansion = true; 677 MacroExpansionInDirectivesOverride = true; 678 } 679 680 /// LookAhead - This peeks ahead N tokens and returns that token without 681 /// consuming any tokens. LookAhead(0) returns the next token that would be 682 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 683 /// returns normal tokens after phase 5. As such, it is equivalent to using 684 /// 'Lex', not 'LexUnexpandedToken'. 685 const Token &LookAhead(unsigned N) { 686 if (CachedLexPos + N < CachedTokens.size()) 687 return CachedTokens[CachedLexPos+N]; 688 else 689 return PeekAhead(N+1); 690 } 691 692 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 693 /// this allows to revert a specific number of tokens. 694 /// Note that the number of tokens being reverted should be up to the last 695 /// backtrack position, not more. 696 void RevertCachedTokens(unsigned N) { 697 assert(isBacktrackEnabled() && 698 "Should only be called when tokens are cached for backtracking"); 699 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 700 && "Should revert tokens up to the last backtrack position, not more"); 701 assert(signed(CachedLexPos) - signed(N) >= 0 && 702 "Corrupted backtrack positions ?"); 703 CachedLexPos -= N; 704 } 705 706 /// EnterToken - Enters a token in the token stream to be lexed next. If 707 /// BackTrack() is called afterwards, the token will remain at the insertion 708 /// point. 709 void EnterToken(const Token &Tok) { 710 EnterCachingLexMode(); 711 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 712 } 713 714 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 715 /// tokens (because backtrack is enabled) it should replace the most recent 716 /// cached tokens with the given annotation token. This function has no effect 717 /// if backtracking is not enabled. 718 /// 719 /// Note that the use of this function is just for optimization; so that the 720 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 721 /// invoked. 722 void AnnotateCachedTokens(const Token &Tok) { 723 assert(Tok.isAnnotation() && "Expected annotation token"); 724 if (CachedLexPos != 0 && isBacktrackEnabled()) 725 AnnotatePreviousCachedTokens(Tok); 726 } 727 728 /// \brief Replace the last token with an annotation token. 729 /// 730 /// Like AnnotateCachedTokens(), this routine replaces an 731 /// already-parsed (and resolved) token with an annotation 732 /// token. However, this routine only replaces the last token with 733 /// the annotation token; it does not affect any other cached 734 /// tokens. This function has no effect if backtracking is not 735 /// enabled. 736 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 737 assert(Tok.isAnnotation() && "Expected annotation token"); 738 if (CachedLexPos != 0 && isBacktrackEnabled()) 739 CachedTokens[CachedLexPos-1] = Tok; 740 } 741 742 /// TypoCorrectToken - Update the current token to represent the provided 743 /// identifier, in order to cache an action performed by typo correction. 744 void TypoCorrectToken(const Token &Tok) { 745 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 746 if (CachedLexPos != 0 && isBacktrackEnabled()) 747 CachedTokens[CachedLexPos-1] = Tok; 748 } 749 750 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 751 /// CurTokenLexer pointers. 752 void recomputeCurLexerKind(); 753 754 /// \brief Returns true if incremental processing is enabled 755 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 756 757 /// \brief Enables the incremental processing 758 void enableIncrementalProcessing(bool value = true) { 759 IncrementalProcessing = value; 760 } 761 762 /// \brief Specify the point at which code-completion will be performed. 763 /// 764 /// \param File the file in which code completion should occur. If 765 /// this file is included multiple times, code-completion will 766 /// perform completion the first time it is included. If NULL, this 767 /// function clears out the code-completion point. 768 /// 769 /// \param Line the line at which code completion should occur 770 /// (1-based). 771 /// 772 /// \param Column the column at which code completion should occur 773 /// (1-based). 774 /// 775 /// \returns true if an error occurred, false otherwise. 776 bool SetCodeCompletionPoint(const FileEntry *File, 777 unsigned Line, unsigned Column); 778 779 /// \brief Determine if we are performing code completion. 780 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 781 782 /// \brief Returns the location of the code-completion point. 783 /// Returns an invalid location if code-completion is not enabled or the file 784 /// containing the code-completion point has not been lexed yet. 785 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 786 787 /// \brief Returns the start location of the file of code-completion point. 788 /// Returns an invalid location if code-completion is not enabled or the file 789 /// containing the code-completion point has not been lexed yet. 790 SourceLocation getCodeCompletionFileLoc() const { 791 return CodeCompletionFileLoc; 792 } 793 794 /// \brief Returns true if code-completion is enabled and we have hit the 795 /// code-completion point. 796 bool isCodeCompletionReached() const { return CodeCompletionReached; } 797 798 /// \brief Note that we hit the code-completion point. 799 void setCodeCompletionReached() { 800 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 801 CodeCompletionReached = true; 802 // Silence any diagnostics that occur after we hit the code-completion. 803 getDiagnostics().setSuppressAllDiagnostics(true); 804 } 805 806 /// \brief The location of the currently-active \#pragma clang 807 /// arc_cf_code_audited begin. Returns an invalid location if there 808 /// is no such pragma active. 809 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 810 return PragmaARCCFCodeAuditedLoc; 811 } 812 813 /// \brief Set the location of the currently-active \#pragma clang 814 /// arc_cf_code_audited begin. An invalid location ends the pragma. 815 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 816 PragmaARCCFCodeAuditedLoc = Loc; 817 } 818 819 /// \brief Instruct the preprocessor to skip part of the main source file. 820 /// 821 /// \param Bytes The number of bytes in the preamble to skip. 822 /// 823 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 824 /// start of a line. 825 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 826 SkipMainFilePreamble.first = Bytes; 827 SkipMainFilePreamble.second = StartOfLine; 828 } 829 830 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 831 /// the specified Token's location, translating the token's start 832 /// position in the current buffer into a SourcePosition object for rendering. 833 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 834 return Diags->Report(Loc, DiagID); 835 } 836 837 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 838 return Diags->Report(Tok.getLocation(), DiagID); 839 } 840 841 /// getSpelling() - Return the 'spelling' of the token at the given 842 /// location; does not go up to the spelling location or down to the 843 /// expansion location. 844 /// 845 /// \param buffer A buffer which will be used only if the token requires 846 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 847 /// \param invalid If non-null, will be set \c true if an error occurs. 848 StringRef getSpelling(SourceLocation loc, 849 SmallVectorImpl<char> &buffer, 850 bool *invalid = 0) const { 851 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 852 } 853 854 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 855 /// token is the characters used to represent the token in the source file 856 /// after trigraph expansion and escaped-newline folding. In particular, this 857 /// wants to get the true, uncanonicalized, spelling of things like digraphs 858 /// UCNs, etc. 859 /// 860 /// \param Invalid If non-null, will be set \c true if an error occurs. 861 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 862 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 863 } 864 865 /// getSpelling - This method is used to get the spelling of a token into a 866 /// preallocated buffer, instead of as an std::string. The caller is required 867 /// to allocate enough space for the token, which is guaranteed to be at least 868 /// Tok.getLength() bytes long. The length of the actual result is returned. 869 /// 870 /// Note that this method may do two possible things: it may either fill in 871 /// the buffer specified with characters, or it may *change the input pointer* 872 /// to point to a constant buffer with the data already in it (avoiding a 873 /// copy). The caller is not allowed to modify the returned buffer pointer 874 /// if an internal buffer is returned. 875 unsigned getSpelling(const Token &Tok, const char *&Buffer, 876 bool *Invalid = 0) const { 877 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 878 } 879 880 /// getSpelling - This method is used to get the spelling of a token into a 881 /// SmallVector. Note that the returned StringRef may not point to the 882 /// supplied buffer if a copy can be avoided. 883 StringRef getSpelling(const Token &Tok, 884 SmallVectorImpl<char> &Buffer, 885 bool *Invalid = 0) const; 886 887 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 888 /// with length 1, return the character. 889 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 890 bool *Invalid = 0) const { 891 assert(Tok.is(tok::numeric_constant) && 892 Tok.getLength() == 1 && "Called on unsupported token"); 893 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 894 895 // If the token is carrying a literal data pointer, just use it. 896 if (const char *D = Tok.getLiteralData()) 897 return *D; 898 899 // Otherwise, fall back on getCharacterData, which is slower, but always 900 // works. 901 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 902 } 903 904 /// \brief Retrieve the name of the immediate macro expansion. 905 /// 906 /// This routine starts from a source location, and finds the name of the macro 907 /// responsible for its immediate expansion. It looks through any intervening 908 /// macro argument expansions to compute this. It returns a StringRef which 909 /// refers to the SourceManager-owned buffer of the source where that macro 910 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 911 StringRef getImmediateMacroName(SourceLocation Loc) { 912 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 913 } 914 915 /// CreateString - Plop the specified string into a scratch buffer and set the 916 /// specified token's location and length to it. If specified, the source 917 /// location provides a location of the expansion point of the token. 918 void CreateString(StringRef Str, Token &Tok, 919 SourceLocation ExpansionLocStart = SourceLocation(), 920 SourceLocation ExpansionLocEnd = SourceLocation()); 921 922 /// \brief Computes the source location just past the end of the 923 /// token at this source location. 924 /// 925 /// This routine can be used to produce a source location that 926 /// points just past the end of the token referenced by \p Loc, and 927 /// is generally used when a diagnostic needs to point just after a 928 /// token where it expected something different that it received. If 929 /// the returned source location would not be meaningful (e.g., if 930 /// it points into a macro), this routine returns an invalid 931 /// source location. 932 /// 933 /// \param Offset an offset from the end of the token, where the source 934 /// location should refer to. The default offset (0) produces a source 935 /// location pointing just past the end of the token; an offset of 1 produces 936 /// a source location pointing to the last character in the token, etc. 937 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 938 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 939 } 940 941 /// \brief Returns true if the given MacroID location points at the first 942 /// token of the macro expansion. 943 /// 944 /// \param MacroBegin If non-null and function returns true, it is set to 945 /// begin location of the macro. 946 bool isAtStartOfMacroExpansion(SourceLocation loc, 947 SourceLocation *MacroBegin = 0) const { 948 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 949 MacroBegin); 950 } 951 952 /// \brief Returns true if the given MacroID location points at the last 953 /// token of the macro expansion. 954 /// 955 /// \param MacroEnd If non-null and function returns true, it is set to 956 /// end location of the macro. 957 bool isAtEndOfMacroExpansion(SourceLocation loc, 958 SourceLocation *MacroEnd = 0) const { 959 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 960 } 961 962 /// DumpToken - Print the token to stderr, used for debugging. 963 /// 964 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 965 void DumpLocation(SourceLocation Loc) const; 966 void DumpMacro(const MacroInfo &MI) const; 967 968 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 969 /// token, return a new location that specifies a character within the token. 970 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 971 unsigned Char) const { 972 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 973 } 974 975 /// IncrementPasteCounter - Increment the counters for the number of token 976 /// paste operations performed. If fast was specified, this is a 'fast paste' 977 /// case we handled. 978 /// 979 void IncrementPasteCounter(bool isFast) { 980 if (isFast) 981 ++NumFastTokenPaste; 982 else 983 ++NumTokenPaste; 984 } 985 986 void PrintStats(); 987 988 size_t getTotalMemory() const; 989 990 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 991 /// comment (/##/) in microsoft mode, this method handles updating the current 992 /// state, returning the token on the next source line. 993 void HandleMicrosoftCommentPaste(Token &Tok); 994 995 //===--------------------------------------------------------------------===// 996 // Preprocessor callback methods. These are invoked by a lexer as various 997 // directives and events are found. 998 999 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1000 /// identifier information for the token and install it into the token, 1001 /// updating the token kind accordingly. 1002 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1003 1004private: 1005 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1006 1007public: 1008 1009 // SetPoisonReason - Call this function to indicate the reason for 1010 // poisoning an identifier. If that identifier is accessed while 1011 // poisoned, then this reason will be used instead of the default 1012 // "poisoned" diagnostic. 1013 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1014 1015 // HandlePoisonedIdentifier - Display reason for poisoned 1016 // identifier. 1017 void HandlePoisonedIdentifier(Token & Tok); 1018 1019 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1020 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1021 if(II->isPoisoned()) { 1022 HandlePoisonedIdentifier(Identifier); 1023 } 1024 } 1025 } 1026 1027private: 1028 /// Identifiers used for SEH handling in Borland. These are only 1029 /// allowed in particular circumstances 1030 // __except block 1031 IdentifierInfo *Ident__exception_code, 1032 *Ident___exception_code, 1033 *Ident_GetExceptionCode; 1034 // __except filter expression 1035 IdentifierInfo *Ident__exception_info, 1036 *Ident___exception_info, 1037 *Ident_GetExceptionInfo; 1038 // __finally 1039 IdentifierInfo *Ident__abnormal_termination, 1040 *Ident___abnormal_termination, 1041 *Ident_AbnormalTermination; 1042public: 1043 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1044 1045 /// HandleIdentifier - This callback is invoked when the lexer reads an 1046 /// identifier and has filled in the tokens IdentifierInfo member. This 1047 /// callback potentially macro expands it or turns it into a named token (like 1048 /// 'for'). 1049 void HandleIdentifier(Token &Identifier); 1050 1051 1052 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1053 /// the current file. This either returns the EOF token and returns true, or 1054 /// pops a level off the include stack and returns false, at which point the 1055 /// client should call lex again. 1056 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1057 1058 /// HandleEndOfTokenLexer - This callback is invoked when the current 1059 /// TokenLexer hits the end of its token stream. 1060 bool HandleEndOfTokenLexer(Token &Result); 1061 1062 /// HandleDirective - This callback is invoked when the lexer sees a # token 1063 /// at the start of a line. This consumes the directive, modifies the 1064 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1065 /// read is the correct one. 1066 void HandleDirective(Token &Result); 1067 1068 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1069 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1070 /// true, then we consider macros that expand to zero tokens as being ok. 1071 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1072 1073 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1074 /// current line until the tok::eod token is found. 1075 void DiscardUntilEndOfDirective(); 1076 1077 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1078 /// __DATE__ or __TIME__ in the file so far. 1079 bool SawDateOrTime() const { 1080 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1081 } 1082 unsigned getCounterValue() const { return CounterValue; } 1083 void setCounterValue(unsigned V) { CounterValue = V; } 1084 1085 /// \brief Retrieves the module that we're currently building, if any. 1086 Module *getCurrentModule(); 1087 1088 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1089 MacroInfo *AllocateMacroInfo(SourceLocation L); 1090 1091 /// \brief Allocate a new MacroInfo object which is clone of \p MI. 1092 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1093 1094 /// \brief Turn the specified lexer token into a fully checked and spelled 1095 /// filename, e.g. as an operand of \#include. 1096 /// 1097 /// The caller is expected to provide a buffer that is large enough to hold 1098 /// the spelling of the filename, but is also expected to handle the case 1099 /// when this method decides to use a different buffer. 1100 /// 1101 /// \returns true if the input filename was in <>'s or false if it was 1102 /// in ""'s. 1103 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1104 1105 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1106 /// 1107 /// Returns null on failure. \p isAngled indicates whether the file 1108 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1109 const FileEntry *LookupFile(StringRef Filename, 1110 bool isAngled, const DirectoryLookup *FromDir, 1111 const DirectoryLookup *&CurDir, 1112 SmallVectorImpl<char> *SearchPath, 1113 SmallVectorImpl<char> *RelativePath, 1114 Module **SuggestedModule, 1115 bool SkipCache = false); 1116 1117 /// GetCurLookup - The DirectoryLookup structure used to find the current 1118 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1119 /// implement \#include_next and find directory-specific properties. 1120 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1121 1122 /// \brief Return true if we're in the top-level file, not in a \#include. 1123 bool isInPrimaryFile() const; 1124 1125 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1126 /// from a macro as multiple tokens, which need to be glued together. This 1127 /// occurs for code like: 1128 /// \code 1129 /// \#define FOO <x/y.h> 1130 /// \#include FOO 1131 /// \endcode 1132 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1133 /// 1134 /// This code concatenates and consumes tokens up to the '>' token. It 1135 /// returns false if the > was found, otherwise it returns true if it finds 1136 /// and consumes the EOD marker. 1137 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1138 SourceLocation &End); 1139 1140 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1141 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1142 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1143 1144private: 1145 1146 void PushIncludeMacroStack() { 1147 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1148 CurLexer.take(), 1149 CurPTHLexer.take(), 1150 CurPPLexer, 1151 CurTokenLexer.take(), 1152 CurDirLookup)); 1153 CurPPLexer = 0; 1154 } 1155 1156 void PopIncludeMacroStack() { 1157 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1158 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1159 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1160 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1161 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1162 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1163 IncludeMacroStack.pop_back(); 1164 } 1165 1166 /// \brief Allocate a new MacroInfo object. 1167 MacroInfo *AllocateMacroInfo(); 1168 1169 /// \brief Release the specified MacroInfo for re-use. 1170 /// 1171 /// This memory will be reused for allocating new MacroInfo objects. 1172 void ReleaseMacroInfo(MacroInfo* MI); 1173 1174 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1175 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1176 /// and discards the rest of the macro line if the macro name is invalid. 1177 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1178 1179 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1180 /// definition has just been read. Lex the rest of the arguments and the 1181 /// closing ), updating MI with what we learn and saving in LastTok the 1182 /// last token read. 1183 /// Return true if an error occurs parsing the arg list. 1184 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1185 1186 /// We just read a \#if or related directive and decided that the 1187 /// subsequent tokens are in the \#if'd out portion of the 1188 /// file. Lex the rest of the file, until we see an \#endif. If \p 1189 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1190 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1191 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1192 /// already seen one so a \#else directive is a duplicate. When this returns, 1193 /// the caller can lex the first valid token. 1194 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1195 bool FoundNonSkipPortion, bool FoundElse, 1196 SourceLocation ElseLoc = SourceLocation()); 1197 1198 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1199 void PTHSkipExcludedConditionalBlock(); 1200 1201 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1202 /// may occur after a #if or #elif directive and return it as a bool. If the 1203 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1204 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1205 1206 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1207 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1208 void RegisterBuiltinPragmas(); 1209 1210 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1211 void RegisterBuiltinMacros(); 1212 1213 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1214 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1215 /// the macro should not be expanded return true, otherwise return false. 1216 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1217 1218 /// \brief Cache macro expanded tokens for TokenLexers. 1219 // 1220 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1221 /// going to lex in the cache and when it finishes the tokens are removed 1222 /// from the end of the cache. 1223 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1224 ArrayRef<Token> tokens); 1225 void removeCachedMacroExpandedTokensOfLastLexer(); 1226 friend void TokenLexer::ExpandFunctionArguments(); 1227 1228 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1229 /// lexed is a '('. If so, consume the token and return true, if not, this 1230 /// method should have no observable side-effect on the lexed tokens. 1231 bool isNextPPTokenLParen(); 1232 1233 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1234 /// invoked to read all of the formal arguments specified for the macro 1235 /// invocation. This returns null on error. 1236 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1237 SourceLocation &ExpansionEnd); 1238 1239 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1240 /// as a builtin macro, handle it and return the next token as 'Tok'. 1241 void ExpandBuiltinMacro(Token &Tok); 1242 1243 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1244 /// return the first token after the directive. The _Pragma token has just 1245 /// been read into 'Tok'. 1246 void Handle_Pragma(Token &Tok); 1247 1248 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1249 /// is not enclosed within a string literal. 1250 void HandleMicrosoft__pragma(Token &Tok); 1251 1252 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1253 /// start lexing tokens from it instead of the current buffer. 1254 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1255 1256 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1257 /// start getting tokens from it using the PTH cache. 1258 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1259 1260 /// IsFileLexer - Returns true if we are lexing from a file and not a 1261 /// pragma or a macro. 1262 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1263 return L ? !L->isPragmaLexer() : P != 0; 1264 } 1265 1266 static bool IsFileLexer(const IncludeStackInfo& I) { 1267 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1268 } 1269 1270 bool IsFileLexer() const { 1271 return IsFileLexer(CurLexer.get(), CurPPLexer); 1272 } 1273 1274 //===--------------------------------------------------------------------===// 1275 // Caching stuff. 1276 void CachingLex(Token &Result); 1277 bool InCachingLexMode() const { 1278 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1279 // that we are past EOF, not that we are in CachingLex mode. 1280 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1281 !IncludeMacroStack.empty(); 1282 } 1283 void EnterCachingLexMode(); 1284 void ExitCachingLexMode() { 1285 if (InCachingLexMode()) 1286 RemoveTopOfLexerStack(); 1287 } 1288 const Token &PeekAhead(unsigned N); 1289 void AnnotatePreviousCachedTokens(const Token &Tok); 1290 1291 //===--------------------------------------------------------------------===// 1292 /// Handle*Directive - implement the various preprocessor directives. These 1293 /// should side-effect the current preprocessor object so that the next call 1294 /// to Lex() will return the appropriate token next. 1295 void HandleLineDirective(Token &Tok); 1296 void HandleDigitDirective(Token &Tok); 1297 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1298 void HandleIdentSCCSDirective(Token &Tok); 1299 void HandleMacroPublicDirective(Token &Tok); 1300 void HandleMacroPrivateDirective(Token &Tok); 1301 1302 // File inclusion. 1303 void HandleIncludeDirective(SourceLocation HashLoc, 1304 Token &Tok, 1305 const DirectoryLookup *LookupFrom = 0, 1306 bool isImport = false); 1307 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1308 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1309 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1310 void HandleMicrosoftImportDirective(Token &Tok); 1311 1312 // Macro handling. 1313 void HandleDefineDirective(Token &Tok); 1314 void HandleUndefDirective(Token &Tok); 1315 1316 // Conditional Inclusion. 1317 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1318 bool ReadAnyTokensBeforeDirective); 1319 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1320 void HandleEndifDirective(Token &Tok); 1321 void HandleElseDirective(Token &Tok); 1322 void HandleElifDirective(Token &Tok); 1323 1324 // Pragmas. 1325 void HandlePragmaDirective(unsigned Introducer); 1326public: 1327 void HandlePragmaOnce(Token &OnceTok); 1328 void HandlePragmaMark(); 1329 void HandlePragmaPoison(Token &PoisonTok); 1330 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1331 void HandlePragmaDependency(Token &DependencyTok); 1332 void HandlePragmaComment(Token &CommentTok); 1333 void HandlePragmaMessage(Token &MessageTok); 1334 void HandlePragmaPushMacro(Token &Tok); 1335 void HandlePragmaPopMacro(Token &Tok); 1336 void HandlePragmaIncludeAlias(Token &Tok); 1337 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1338 1339 // Return true and store the first token only if any CommentHandler 1340 // has inserted some tokens and getCommentRetentionState() is false. 1341 bool HandleComment(Token &Token, SourceRange Comment); 1342 1343 /// \brief A macro is used, update information about macros that need unused 1344 /// warnings. 1345 void markMacroAsUsed(MacroInfo *MI); 1346}; 1347 1348/// \brief Abstract base class that describes a handler that will receive 1349/// source ranges for each of the comments encountered in the source file. 1350class CommentHandler { 1351public: 1352 virtual ~CommentHandler(); 1353 1354 // The handler shall return true if it has pushed any tokens 1355 // to be read using e.g. EnterToken or EnterTokenStream. 1356 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1357}; 1358 1359} // end namespace clang 1360 1361#endif 1362