Preprocessor.h revision a71a7d8a1ce4474e7bdb680658fb58b6caf391d3
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/PPMutationListener.h" 22#include "clang/Lex/TokenLexer.h" 23#include "clang/Lex/PTHManager.h" 24#include "clang/Basic/Builtins.h" 25#include "clang/Basic/Diagnostic.h" 26#include "clang/Basic/IdentifierTable.h" 27#include "clang/Basic/SourceLocation.h" 28#include "llvm/ADT/DenseMap.h" 29#include "llvm/ADT/IntrusiveRefCntPtr.h" 30#include "llvm/ADT/SmallPtrSet.h" 31#include "llvm/ADT/OwningPtr.h" 32#include "llvm/ADT/SmallVector.h" 33#include "llvm/ADT/ArrayRef.h" 34#include "llvm/Support/Allocator.h" 35#include <vector> 36 37namespace llvm { 38 template<unsigned InternalLen> class SmallString; 39} 40 41namespace clang { 42 43class SourceManager; 44class ExternalPreprocessorSource; 45class FileManager; 46class FileEntry; 47class HeaderSearch; 48class PragmaNamespace; 49class PragmaHandler; 50class CommentHandler; 51class ScratchBuffer; 52class TargetInfo; 53class PPCallbacks; 54class CodeCompletionHandler; 55class DirectoryLookup; 56class PreprocessingRecord; 57class ModuleLoader; 58class PreprocessorOptions; 59 60/// \brief Stores token information for comparing actual tokens with 61/// predefined values. Only handles simple tokens and identifiers. 62class TokenValue { 63 tok::TokenKind Kind; 64 IdentifierInfo *II; 65 66public: 67 TokenValue(tok::TokenKind Kind) : Kind(Kind), II(0) { 68 assert(Kind != tok::raw_identifier && "Raw identifiers are not supported."); 69 assert(Kind != tok::identifier && 70 "Identifiers should be created by TokenValue(IdentifierInfo *)"); 71 assert(!tok::isLiteral(Kind) && "Literals are not supported."); 72 assert(!tok::isAnnotation(Kind) && "Annotations are not supported."); 73 } 74 TokenValue(IdentifierInfo *II) : Kind(tok::identifier), II(II) {} 75 bool operator==(const Token &Tok) const { 76 return Tok.getKind() == Kind && 77 (!II || II == Tok.getIdentifierInfo()); 78 } 79}; 80 81/// Preprocessor - This object engages in a tight little dance with the lexer to 82/// efficiently preprocess tokens. Lexers know only about tokens within a 83/// single source file, and don't know anything about preprocessor-level issues 84/// like the \#include stack, token expansion, etc. 85/// 86class Preprocessor : public RefCountedBase<Preprocessor> { 87 llvm::IntrusiveRefCntPtr<PreprocessorOptions> PPOpts; 88 DiagnosticsEngine *Diags; 89 LangOptions &LangOpts; 90 const TargetInfo *Target; 91 FileManager &FileMgr; 92 SourceManager &SourceMgr; 93 ScratchBuffer *ScratchBuf; 94 HeaderSearch &HeaderInfo; 95 ModuleLoader &TheModuleLoader; 96 97 /// \brief External source of macros. 98 ExternalPreprocessorSource *ExternalSource; 99 100 101 /// PTH - An optional PTHManager object used for getting tokens from 102 /// a token cache rather than lexing the original source file. 103 OwningPtr<PTHManager> PTH; 104 105 /// BP - A BumpPtrAllocator object used to quickly allocate and release 106 /// objects internal to the Preprocessor. 107 llvm::BumpPtrAllocator BP; 108 109 /// Identifiers for builtin macros and other builtins. 110 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 111 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 112 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 113 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 114 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 115 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 116 IdentifierInfo *Ident_Pragma, *Ident__pragma; // _Pragma, __pragma 117 IdentifierInfo *Ident__VA_ARGS__; // __VA_ARGS__ 118 IdentifierInfo *Ident__has_feature; // __has_feature 119 IdentifierInfo *Ident__has_extension; // __has_extension 120 IdentifierInfo *Ident__has_builtin; // __has_builtin 121 IdentifierInfo *Ident__has_attribute; // __has_attribute 122 IdentifierInfo *Ident__has_include; // __has_include 123 IdentifierInfo *Ident__has_include_next; // __has_include_next 124 IdentifierInfo *Ident__has_warning; // __has_warning 125 IdentifierInfo *Ident__building_module; // __building_module 126 IdentifierInfo *Ident__MODULE__; // __MODULE__ 127 128 SourceLocation DATELoc, TIMELoc; 129 unsigned CounterValue; // Next __COUNTER__ value. 130 131 enum { 132 /// MaxIncludeStackDepth - Maximum depth of \#includes. 133 MaxAllowedIncludeStackDepth = 200 134 }; 135 136 // State that is set before the preprocessor begins. 137 bool KeepComments : 1; 138 bool KeepMacroComments : 1; 139 bool SuppressIncludeNotFoundError : 1; 140 141 // State that changes while the preprocessor runs: 142 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 143 144 /// Whether the preprocessor owns the header search object. 145 bool OwnsHeaderSearch : 1; 146 147 /// DisableMacroExpansion - True if macro expansion is disabled. 148 bool DisableMacroExpansion : 1; 149 150 /// MacroExpansionInDirectivesOverride - Temporarily disables 151 /// DisableMacroExpansion (i.e. enables expansion) when parsing preprocessor 152 /// directives. 153 bool MacroExpansionInDirectivesOverride : 1; 154 155 class ResetMacroExpansionHelper; 156 157 /// \brief Whether we have already loaded macros from the external source. 158 mutable bool ReadMacrosFromExternalSource : 1; 159 160 /// \brief True if pragmas are enabled. 161 bool PragmasEnabled : 1; 162 163 /// \brief True if we are pre-expanding macro arguments. 164 bool InMacroArgPreExpansion; 165 166 /// Identifiers - This is mapping/lookup information for all identifiers in 167 /// the program, including program keywords. 168 mutable IdentifierTable Identifiers; 169 170 /// Selectors - This table contains all the selectors in the program. Unlike 171 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 172 /// It is declared/expanded here because it's role/lifetime is 173 /// conceptually similar the IdentifierTable. In addition, the current control 174 /// flow (in clang::ParseAST()), make it convenient to put here. 175 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 176 /// the lifetime of the preprocessor. 177 SelectorTable Selectors; 178 179 /// BuiltinInfo - Information about builtins. 180 Builtin::Context BuiltinInfo; 181 182 /// PragmaHandlers - This tracks all of the pragmas that the client registered 183 /// with this preprocessor. 184 PragmaNamespace *PragmaHandlers; 185 186 /// \brief Tracks all of the comment handlers that the client registered 187 /// with this preprocessor. 188 std::vector<CommentHandler *> CommentHandlers; 189 190 /// \brief True if we want to ignore EOF token and continue later on (thus 191 /// avoid tearing the Lexer and etc. down). 192 bool IncrementalProcessing; 193 194 /// \brief The code-completion handler. 195 CodeCompletionHandler *CodeComplete; 196 197 /// \brief The file that we're performing code-completion for, if any. 198 const FileEntry *CodeCompletionFile; 199 200 /// \brief The offset in file for the code-completion point. 201 unsigned CodeCompletionOffset; 202 203 /// \brief The location for the code-completion point. This gets instantiated 204 /// when the CodeCompletionFile gets \#include'ed for preprocessing. 205 SourceLocation CodeCompletionLoc; 206 207 /// \brief The start location for the file of the code-completion point. 208 /// 209 /// This gets instantiated when the CodeCompletionFile gets \#include'ed 210 /// for preprocessing. 211 SourceLocation CodeCompletionFileLoc; 212 213 /// \brief The source location of the 'import' contextual keyword we just 214 /// lexed, if any. 215 SourceLocation ModuleImportLoc; 216 217 /// \brief The module import path that we're currently processing. 218 llvm::SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> 219 ModuleImportPath; 220 221 /// \brief Whether the module import expectes an identifier next. Otherwise, 222 /// it expects a '.' or ';'. 223 bool ModuleImportExpectsIdentifier; 224 225 /// \brief The source location of the currently-active 226 /// #pragma clang arc_cf_code_audited begin. 227 SourceLocation PragmaARCCFCodeAuditedLoc; 228 229 /// \brief True if we hit the code-completion point. 230 bool CodeCompletionReached; 231 232 /// \brief The number of bytes that we will initially skip when entering the 233 /// main file, which is used when loading a precompiled preamble, along 234 /// with a flag that indicates whether skipping this number of bytes will 235 /// place the lexer at the start of a line. 236 std::pair<unsigned, bool> SkipMainFilePreamble; 237 238 /// CurLexer - This is the current top of the stack that we're lexing from if 239 /// not expanding a macro and we are lexing directly from source code. 240 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 241 OwningPtr<Lexer> CurLexer; 242 243 /// CurPTHLexer - This is the current top of stack that we're lexing from if 244 /// not expanding from a macro and we are lexing from a PTH cache. 245 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 246 OwningPtr<PTHLexer> CurPTHLexer; 247 248 /// CurPPLexer - This is the current top of the stack what we're lexing from 249 /// if not expanding a macro. This is an alias for either CurLexer or 250 /// CurPTHLexer. 251 PreprocessorLexer *CurPPLexer; 252 253 /// CurLookup - The DirectoryLookup structure used to find the current 254 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 255 /// implement \#include_next and find directory-specific properties. 256 const DirectoryLookup *CurDirLookup; 257 258 /// CurTokenLexer - This is the current macro we are expanding, if we are 259 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 260 OwningPtr<TokenLexer> CurTokenLexer; 261 262 /// \brief The kind of lexer we're currently working with. 263 enum CurLexerKind { 264 CLK_Lexer, 265 CLK_PTHLexer, 266 CLK_TokenLexer, 267 CLK_CachingLexer, 268 CLK_LexAfterModuleImport 269 } CurLexerKind; 270 271 /// IncludeMacroStack - This keeps track of the stack of files currently 272 /// \#included, and macros currently being expanded from, not counting 273 /// CurLexer/CurTokenLexer. 274 struct IncludeStackInfo { 275 enum CurLexerKind CurLexerKind; 276 Lexer *TheLexer; 277 PTHLexer *ThePTHLexer; 278 PreprocessorLexer *ThePPLexer; 279 TokenLexer *TheTokenLexer; 280 const DirectoryLookup *TheDirLookup; 281 282 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 283 PreprocessorLexer* PPL, 284 TokenLexer* TL, const DirectoryLookup *D) 285 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 286 TheTokenLexer(TL), TheDirLookup(D) {} 287 }; 288 std::vector<IncludeStackInfo> IncludeMacroStack; 289 290 /// Callbacks - These are actions invoked when some preprocessor activity is 291 /// encountered (e.g. a file is \#included, etc). 292 PPCallbacks *Callbacks; 293 294 /// \brief Listener whose actions are invoked when an entity in the 295 /// preprocessor (e.g., a macro) that was loaded from an AST file is 296 /// later mutated. 297 PPMutationListener *Listener; 298 299 struct MacroExpandsInfo { 300 Token Tok; 301 MacroInfo *MI; 302 SourceRange Range; 303 MacroExpandsInfo(Token Tok, MacroInfo *MI, SourceRange Range) 304 : Tok(Tok), MI(MI), Range(Range) { } 305 }; 306 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 307 308 /// Macros - For each IdentifierInfo that was associated with a macro, we 309 /// keep a mapping to the history of all macro definitions and #undefs in 310 /// the reverse order (the latest one is in the head of the list). 311 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 312 friend class ASTReader; 313 314 /// \brief Macros that we want to warn because they are not used at the end 315 /// of the translation unit; we store just their SourceLocations instead 316 /// something like MacroInfo*. The benefit of this is that when we are 317 /// deserializing from PCH, we don't need to deserialize identifier & macros 318 /// just so that we can report that they are unused, we just warn using 319 /// the SourceLocations of this set (that will be filled by the ASTReader). 320 /// We are using SmallPtrSet instead of a vector for faster removal. 321 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 322 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 323 324 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 325 /// reused for quick allocation. 326 MacroArgs *MacroArgCache; 327 friend class MacroArgs; 328 329 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 330 /// push_macro directive, we keep a MacroInfo stack used to restore 331 /// previous macro value. 332 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 333 334 // Various statistics we track for performance analysis. 335 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 336 unsigned NumIf, NumElse, NumEndif; 337 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 338 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 339 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 340 unsigned NumSkipped; 341 342 /// Predefines - This string is the predefined macros that preprocessor 343 /// should use from the command line etc. 344 std::string Predefines; 345 346 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 347 enum { TokenLexerCacheSize = 8 }; 348 unsigned NumCachedTokenLexers; 349 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 350 351 /// \brief Keeps macro expanded tokens for TokenLexers. 352 // 353 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 354 /// going to lex in the cache and when it finishes the tokens are removed 355 /// from the end of the cache. 356 SmallVector<Token, 16> MacroExpandedTokens; 357 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 358 359 /// \brief A record of the macro definitions and expansions that 360 /// occurred during preprocessing. 361 /// 362 /// This is an optional side structure that can be enabled with 363 /// \c createPreprocessingRecord() prior to preprocessing. 364 PreprocessingRecord *Record; 365 366private: // Cached tokens state. 367 typedef SmallVector<Token, 1> CachedTokensTy; 368 369 /// CachedTokens - Cached tokens are stored here when we do backtracking or 370 /// lookahead. They are "lexed" by the CachingLex() method. 371 CachedTokensTy CachedTokens; 372 373 /// CachedLexPos - The position of the cached token that CachingLex() should 374 /// "lex" next. If it points beyond the CachedTokens vector, it means that 375 /// a normal Lex() should be invoked. 376 CachedTokensTy::size_type CachedLexPos; 377 378 /// BacktrackPositions - Stack of backtrack positions, allowing nested 379 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 380 /// indicate where CachedLexPos should be set when the BackTrack() method is 381 /// invoked (at which point the last position is popped). 382 std::vector<CachedTokensTy::size_type> BacktrackPositions; 383 384 struct MacroInfoChain { 385 MacroInfo MI; 386 MacroInfoChain *Next; 387 MacroInfoChain *Prev; 388 }; 389 390 /// MacroInfos are managed as a chain for easy disposal. This is the head 391 /// of that list. 392 MacroInfoChain *MIChainHead; 393 394 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 395 /// allocation. 396 MacroInfoChain *MICache; 397 398public: 399 Preprocessor(llvm::IntrusiveRefCntPtr<PreprocessorOptions> PPOpts, 400 DiagnosticsEngine &diags, LangOptions &opts, 401 const TargetInfo *target, 402 SourceManager &SM, HeaderSearch &Headers, 403 ModuleLoader &TheModuleLoader, 404 IdentifierInfoLookup *IILookup = 0, 405 bool OwnsHeaderSearch = false, 406 bool DelayInitialization = false, 407 bool IncrProcessing = false); 408 409 ~Preprocessor(); 410 411 /// \brief Initialize the preprocessor, if the constructor did not already 412 /// perform the initialization. 413 /// 414 /// \param Target Information about the target. 415 void Initialize(const TargetInfo &Target); 416 417 /// \brief Retrieve the preprocessor options used to initialize this 418 /// preprocessor. 419 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; } 420 421 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 422 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 423 424 const LangOptions &getLangOpts() const { return LangOpts; } 425 const TargetInfo &getTargetInfo() const { return *Target; } 426 FileManager &getFileManager() const { return FileMgr; } 427 SourceManager &getSourceManager() const { return SourceMgr; } 428 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 429 430 IdentifierTable &getIdentifierTable() { return Identifiers; } 431 SelectorTable &getSelectorTable() { return Selectors; } 432 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 433 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 434 435 void setPTHManager(PTHManager* pm); 436 437 PTHManager *getPTHManager() { return PTH.get(); } 438 439 void setExternalSource(ExternalPreprocessorSource *Source) { 440 ExternalSource = Source; 441 } 442 443 ExternalPreprocessorSource *getExternalSource() const { 444 return ExternalSource; 445 } 446 447 /// \brief Retrieve the module loader associated with this preprocessor. 448 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 449 450 /// SetCommentRetentionState - Control whether or not the preprocessor retains 451 /// comments in output. 452 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 453 this->KeepComments = KeepComments | KeepMacroComments; 454 this->KeepMacroComments = KeepMacroComments; 455 } 456 457 bool getCommentRetentionState() const { return KeepComments; } 458 459 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 460 bool getPragmasEnabled() const { return PragmasEnabled; } 461 462 void SetSuppressIncludeNotFoundError(bool Suppress) { 463 SuppressIncludeNotFoundError = Suppress; 464 } 465 466 bool GetSuppressIncludeNotFoundError() { 467 return SuppressIncludeNotFoundError; 468 } 469 470 /// isCurrentLexer - Return true if we are lexing directly from the specified 471 /// lexer. 472 bool isCurrentLexer(const PreprocessorLexer *L) const { 473 return CurPPLexer == L; 474 } 475 476 /// getCurrentLexer - Return the current lexer being lexed from. Note 477 /// that this ignores any potentially active macro expansions and _Pragma 478 /// expansions going on at the time. 479 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 480 481 /// getCurrentFileLexer - Return the current file lexer being lexed from. 482 /// Note that this ignores any potentially active macro expansions and _Pragma 483 /// expansions going on at the time. 484 PreprocessorLexer *getCurrentFileLexer() const; 485 486 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 487 /// Note that this class takes ownership of any PPCallbacks object given to 488 /// it. 489 PPCallbacks *getPPCallbacks() const { return Callbacks; } 490 void addPPCallbacks(PPCallbacks *C) { 491 if (Callbacks) 492 C = new PPChainedCallbacks(C, Callbacks); 493 Callbacks = C; 494 } 495 496 /// \brief Attach an preprocessor mutation listener to the preprocessor. 497 /// 498 /// The preprocessor mutation listener provides the ability to track 499 /// modifications to the preprocessor entities committed after they were 500 /// initially created. 501 void setPPMutationListener(PPMutationListener *Listener) { 502 this->Listener = Listener; 503 } 504 505 /// \brief Retrieve a pointer to the preprocessor mutation listener 506 /// associated with this preprocessor, if any. 507 PPMutationListener *getPPMutationListener() const { return Listener; } 508 509 /// \brief Given an identifier, return the MacroInfo it is \#defined to 510 /// or null if it isn't \#define'd. 511 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 512 if (!II->hasMacroDefinition()) 513 return 0; 514 515 MacroInfo *MI = getMacroInfoHistory(II); 516 assert(MI->getUndefLoc().isInvalid() && "Macro is undefined!"); 517 return MI; 518 } 519 520 /// \brief Given an identifier, return the (probably #undef'd) MacroInfo 521 /// representing the most recent macro definition. One can iterate over all 522 /// previous macro definitions from it. This method should only be called for 523 /// identifiers that hadMacroDefinition(). 524 MacroInfo *getMacroInfoHistory(IdentifierInfo *II) const; 525 526 /// \brief Specify a macro for this identifier. 527 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 528 /// \brief Add a MacroInfo that was loaded from an AST file. 529 void addLoadedMacroInfo(IdentifierInfo *II, MacroInfo *MI, 530 MacroInfo *Hint = 0); 531 /// \brief Make the given MacroInfo, that was loaded from an AST file and 532 /// previously hidden, visible. 533 void makeLoadedMacroInfoVisible(IdentifierInfo *II, MacroInfo *MI); 534 /// \brief Undefine a macro for this identifier. 535 void clearMacroInfo(IdentifierInfo *II); 536 537 /// macro_iterator/macro_begin/macro_end - This allows you to walk the macro 538 /// history table. Currently defined macros have 539 /// IdentifierInfo::hasMacroDefinition() set and an empty 540 /// MacroInfo::getUndefLoc() at the head of the list. 541 typedef llvm::DenseMap<IdentifierInfo*, 542 MacroInfo*>::const_iterator macro_iterator; 543 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 544 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 545 546 /// \brief Return the name of the macro defined before \p Loc that has 547 /// spelling \p Tokens. If there are multiple macros with same spelling, 548 /// return the last one defined. 549 StringRef getLastMacroWithSpelling(SourceLocation Loc, 550 ArrayRef<TokenValue> Tokens) const; 551 552 const std::string &getPredefines() const { return Predefines; } 553 /// setPredefines - Set the predefines for this Preprocessor. These 554 /// predefines are automatically injected when parsing the main file. 555 void setPredefines(const char *P) { Predefines = P; } 556 void setPredefines(const std::string &P) { Predefines = P; } 557 558 /// getIdentifierInfo - Return information about the specified preprocessor 559 /// identifier token. The version of this method that takes two character 560 /// pointers is preferred unless the identifier is already available as a 561 /// string (this avoids allocation and copying of memory to construct an 562 /// std::string). 563 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 564 return &Identifiers.get(Name); 565 } 566 567 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 568 /// If 'Namespace' is non-null, then it is a token required to exist on the 569 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 570 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 571 void AddPragmaHandler(PragmaHandler *Handler) { 572 AddPragmaHandler(StringRef(), Handler); 573 } 574 575 /// RemovePragmaHandler - Remove the specific pragma handler from 576 /// the preprocessor. If \p Namespace is non-null, then it should 577 /// be the namespace that \p Handler was added to. It is an error 578 /// to remove a handler that has not been registered. 579 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 580 void RemovePragmaHandler(PragmaHandler *Handler) { 581 RemovePragmaHandler(StringRef(), Handler); 582 } 583 584 /// \brief Add the specified comment handler to the preprocessor. 585 void addCommentHandler(CommentHandler *Handler); 586 587 /// \brief Remove the specified comment handler. 588 /// 589 /// It is an error to remove a handler that has not been registered. 590 void removeCommentHandler(CommentHandler *Handler); 591 592 /// \brief Set the code completion handler to the given object. 593 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 594 CodeComplete = &Handler; 595 } 596 597 /// \brief Retrieve the current code-completion handler. 598 CodeCompletionHandler *getCodeCompletionHandler() const { 599 return CodeComplete; 600 } 601 602 /// \brief Clear out the code completion handler. 603 void clearCodeCompletionHandler() { 604 CodeComplete = 0; 605 } 606 607 /// \brief Hook used by the lexer to invoke the "natural language" code 608 /// completion point. 609 void CodeCompleteNaturalLanguage(); 610 611 /// \brief Retrieve the preprocessing record, or NULL if there is no 612 /// preprocessing record. 613 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 614 615 /// \brief Create a new preprocessing record, which will keep track of 616 /// all macro expansions, macro definitions, etc. 617 void createPreprocessingRecord(bool RecordConditionalDirectives); 618 619 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 620 /// which implicitly adds the builtin defines etc. 621 void EnterMainSourceFile(); 622 623 /// EndSourceFile - Inform the preprocessor callbacks that processing is 624 /// complete. 625 void EndSourceFile(); 626 627 /// EnterSourceFile - Add a source file to the top of the include stack and 628 /// start lexing tokens from it instead of the current buffer. Emit an error 629 /// and don't enter the file on error. 630 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 631 SourceLocation Loc); 632 633 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 634 /// tokens from it instead of the current buffer. Args specifies the 635 /// tokens input to a function-like macro. 636 /// 637 /// ILEnd specifies the location of the ')' for a function-like macro or the 638 /// identifier for an object-like macro. 639 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 640 MacroArgs *Args); 641 642 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 643 /// which will cause the lexer to start returning the specified tokens. 644 /// 645 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 646 /// not be subject to further macro expansion. Otherwise, these tokens will 647 /// be re-macro-expanded when/if expansion is enabled. 648 /// 649 /// If OwnsTokens is false, this method assumes that the specified stream of 650 /// tokens has a permanent owner somewhere, so they do not need to be copied. 651 /// If it is true, it assumes the array of tokens is allocated with new[] and 652 /// must be freed. 653 /// 654 void EnterTokenStream(const Token *Toks, unsigned NumToks, 655 bool DisableMacroExpansion, bool OwnsTokens); 656 657 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 658 /// lexer stack. This should only be used in situations where the current 659 /// state of the top-of-stack lexer is known. 660 void RemoveTopOfLexerStack(); 661 662 /// EnableBacktrackAtThisPos - From the point that this method is called, and 663 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 664 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 665 /// make the Preprocessor re-lex the same tokens. 666 /// 667 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 668 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 669 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 670 /// 671 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 672 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 673 /// tokens will continue indefinitely. 674 /// 675 void EnableBacktrackAtThisPos(); 676 677 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 678 void CommitBacktrackedTokens(); 679 680 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 681 /// EnableBacktrackAtThisPos() was previously called. 682 void Backtrack(); 683 684 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 685 /// caching of tokens is on. 686 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 687 688 /// Lex - To lex a token from the preprocessor, just pull a token from the 689 /// current lexer or macro object. 690 void Lex(Token &Result) { 691 switch (CurLexerKind) { 692 case CLK_Lexer: CurLexer->Lex(Result); break; 693 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 694 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 695 case CLK_CachingLexer: CachingLex(Result); break; 696 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 697 } 698 } 699 700 void LexAfterModuleImport(Token &Result); 701 702 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 703 /// something not a comment. This is useful in -E -C mode where comments 704 /// would foul up preprocessor directive handling. 705 void LexNonComment(Token &Result) { 706 do 707 Lex(Result); 708 while (Result.getKind() == tok::comment); 709 } 710 711 /// LexUnexpandedToken - This is just like Lex, but this disables macro 712 /// expansion of identifier tokens. 713 void LexUnexpandedToken(Token &Result) { 714 // Disable macro expansion. 715 bool OldVal = DisableMacroExpansion; 716 DisableMacroExpansion = true; 717 // Lex the token. 718 Lex(Result); 719 720 // Reenable it. 721 DisableMacroExpansion = OldVal; 722 } 723 724 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 725 /// expansion of identifier tokens. 726 void LexUnexpandedNonComment(Token &Result) { 727 do 728 LexUnexpandedToken(Result); 729 while (Result.getKind() == tok::comment); 730 } 731 732 /// Disables macro expansion everywhere except for preprocessor directives. 733 void SetMacroExpansionOnlyInDirectives() { 734 DisableMacroExpansion = true; 735 MacroExpansionInDirectivesOverride = true; 736 } 737 738 /// LookAhead - This peeks ahead N tokens and returns that token without 739 /// consuming any tokens. LookAhead(0) returns the next token that would be 740 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 741 /// returns normal tokens after phase 5. As such, it is equivalent to using 742 /// 'Lex', not 'LexUnexpandedToken'. 743 const Token &LookAhead(unsigned N) { 744 if (CachedLexPos + N < CachedTokens.size()) 745 return CachedTokens[CachedLexPos+N]; 746 else 747 return PeekAhead(N+1); 748 } 749 750 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 751 /// this allows to revert a specific number of tokens. 752 /// Note that the number of tokens being reverted should be up to the last 753 /// backtrack position, not more. 754 void RevertCachedTokens(unsigned N) { 755 assert(isBacktrackEnabled() && 756 "Should only be called when tokens are cached for backtracking"); 757 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 758 && "Should revert tokens up to the last backtrack position, not more"); 759 assert(signed(CachedLexPos) - signed(N) >= 0 && 760 "Corrupted backtrack positions ?"); 761 CachedLexPos -= N; 762 } 763 764 /// EnterToken - Enters a token in the token stream to be lexed next. If 765 /// BackTrack() is called afterwards, the token will remain at the insertion 766 /// point. 767 void EnterToken(const Token &Tok) { 768 EnterCachingLexMode(); 769 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 770 } 771 772 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 773 /// tokens (because backtrack is enabled) it should replace the most recent 774 /// cached tokens with the given annotation token. This function has no effect 775 /// if backtracking is not enabled. 776 /// 777 /// Note that the use of this function is just for optimization; so that the 778 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 779 /// invoked. 780 void AnnotateCachedTokens(const Token &Tok) { 781 assert(Tok.isAnnotation() && "Expected annotation token"); 782 if (CachedLexPos != 0 && isBacktrackEnabled()) 783 AnnotatePreviousCachedTokens(Tok); 784 } 785 786 /// \brief Replace the last token with an annotation token. 787 /// 788 /// Like AnnotateCachedTokens(), this routine replaces an 789 /// already-parsed (and resolved) token with an annotation 790 /// token. However, this routine only replaces the last token with 791 /// the annotation token; it does not affect any other cached 792 /// tokens. This function has no effect if backtracking is not 793 /// enabled. 794 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 795 assert(Tok.isAnnotation() && "Expected annotation token"); 796 if (CachedLexPos != 0 && isBacktrackEnabled()) 797 CachedTokens[CachedLexPos-1] = Tok; 798 } 799 800 /// TypoCorrectToken - Update the current token to represent the provided 801 /// identifier, in order to cache an action performed by typo correction. 802 void TypoCorrectToken(const Token &Tok) { 803 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 804 if (CachedLexPos != 0 && isBacktrackEnabled()) 805 CachedTokens[CachedLexPos-1] = Tok; 806 } 807 808 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 809 /// CurTokenLexer pointers. 810 void recomputeCurLexerKind(); 811 812 /// \brief Returns true if incremental processing is enabled 813 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 814 815 /// \brief Enables the incremental processing 816 void enableIncrementalProcessing(bool value = true) { 817 IncrementalProcessing = value; 818 } 819 820 /// \brief Specify the point at which code-completion will be performed. 821 /// 822 /// \param File the file in which code completion should occur. If 823 /// this file is included multiple times, code-completion will 824 /// perform completion the first time it is included. If NULL, this 825 /// function clears out the code-completion point. 826 /// 827 /// \param Line the line at which code completion should occur 828 /// (1-based). 829 /// 830 /// \param Column the column at which code completion should occur 831 /// (1-based). 832 /// 833 /// \returns true if an error occurred, false otherwise. 834 bool SetCodeCompletionPoint(const FileEntry *File, 835 unsigned Line, unsigned Column); 836 837 /// \brief Determine if we are performing code completion. 838 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 839 840 /// \brief Returns the location of the code-completion point. 841 /// Returns an invalid location if code-completion is not enabled or the file 842 /// containing the code-completion point has not been lexed yet. 843 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 844 845 /// \brief Returns the start location of the file of code-completion point. 846 /// Returns an invalid location if code-completion is not enabled or the file 847 /// containing the code-completion point has not been lexed yet. 848 SourceLocation getCodeCompletionFileLoc() const { 849 return CodeCompletionFileLoc; 850 } 851 852 /// \brief Returns true if code-completion is enabled and we have hit the 853 /// code-completion point. 854 bool isCodeCompletionReached() const { return CodeCompletionReached; } 855 856 /// \brief Note that we hit the code-completion point. 857 void setCodeCompletionReached() { 858 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 859 CodeCompletionReached = true; 860 // Silence any diagnostics that occur after we hit the code-completion. 861 getDiagnostics().setSuppressAllDiagnostics(true); 862 } 863 864 /// \brief The location of the currently-active \#pragma clang 865 /// arc_cf_code_audited begin. Returns an invalid location if there 866 /// is no such pragma active. 867 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 868 return PragmaARCCFCodeAuditedLoc; 869 } 870 871 /// \brief Set the location of the currently-active \#pragma clang 872 /// arc_cf_code_audited begin. An invalid location ends the pragma. 873 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 874 PragmaARCCFCodeAuditedLoc = Loc; 875 } 876 877 /// \brief Instruct the preprocessor to skip part of the main source file. 878 /// 879 /// \param Bytes The number of bytes in the preamble to skip. 880 /// 881 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 882 /// start of a line. 883 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 884 SkipMainFilePreamble.first = Bytes; 885 SkipMainFilePreamble.second = StartOfLine; 886 } 887 888 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 889 /// the specified Token's location, translating the token's start 890 /// position in the current buffer into a SourcePosition object for rendering. 891 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 892 return Diags->Report(Loc, DiagID); 893 } 894 895 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 896 return Diags->Report(Tok.getLocation(), DiagID); 897 } 898 899 /// getSpelling() - Return the 'spelling' of the token at the given 900 /// location; does not go up to the spelling location or down to the 901 /// expansion location. 902 /// 903 /// \param buffer A buffer which will be used only if the token requires 904 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 905 /// \param invalid If non-null, will be set \c true if an error occurs. 906 StringRef getSpelling(SourceLocation loc, 907 SmallVectorImpl<char> &buffer, 908 bool *invalid = 0) const { 909 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 910 } 911 912 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 913 /// token is the characters used to represent the token in the source file 914 /// after trigraph expansion and escaped-newline folding. In particular, this 915 /// wants to get the true, uncanonicalized, spelling of things like digraphs 916 /// UCNs, etc. 917 /// 918 /// \param Invalid If non-null, will be set \c true if an error occurs. 919 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 920 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 921 } 922 923 /// getSpelling - This method is used to get the spelling of a token into a 924 /// preallocated buffer, instead of as an std::string. The caller is required 925 /// to allocate enough space for the token, which is guaranteed to be at least 926 /// Tok.getLength() bytes long. The length of the actual result is returned. 927 /// 928 /// Note that this method may do two possible things: it may either fill in 929 /// the buffer specified with characters, or it may *change the input pointer* 930 /// to point to a constant buffer with the data already in it (avoiding a 931 /// copy). The caller is not allowed to modify the returned buffer pointer 932 /// if an internal buffer is returned. 933 unsigned getSpelling(const Token &Tok, const char *&Buffer, 934 bool *Invalid = 0) const { 935 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 936 } 937 938 /// getSpelling - This method is used to get the spelling of a token into a 939 /// SmallVector. Note that the returned StringRef may not point to the 940 /// supplied buffer if a copy can be avoided. 941 StringRef getSpelling(const Token &Tok, 942 SmallVectorImpl<char> &Buffer, 943 bool *Invalid = 0) const; 944 945 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 946 /// with length 1, return the character. 947 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 948 bool *Invalid = 0) const { 949 assert(Tok.is(tok::numeric_constant) && 950 Tok.getLength() == 1 && "Called on unsupported token"); 951 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 952 953 // If the token is carrying a literal data pointer, just use it. 954 if (const char *D = Tok.getLiteralData()) 955 return *D; 956 957 // Otherwise, fall back on getCharacterData, which is slower, but always 958 // works. 959 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 960 } 961 962 /// \brief Retrieve the name of the immediate macro expansion. 963 /// 964 /// This routine starts from a source location, and finds the name of the macro 965 /// responsible for its immediate expansion. It looks through any intervening 966 /// macro argument expansions to compute this. It returns a StringRef which 967 /// refers to the SourceManager-owned buffer of the source where that macro 968 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 969 StringRef getImmediateMacroName(SourceLocation Loc) { 970 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 971 } 972 973 /// CreateString - Plop the specified string into a scratch buffer and set the 974 /// specified token's location and length to it. If specified, the source 975 /// location provides a location of the expansion point of the token. 976 void CreateString(StringRef Str, Token &Tok, 977 SourceLocation ExpansionLocStart = SourceLocation(), 978 SourceLocation ExpansionLocEnd = SourceLocation()); 979 980 /// \brief Computes the source location just past the end of the 981 /// token at this source location. 982 /// 983 /// This routine can be used to produce a source location that 984 /// points just past the end of the token referenced by \p Loc, and 985 /// is generally used when a diagnostic needs to point just after a 986 /// token where it expected something different that it received. If 987 /// the returned source location would not be meaningful (e.g., if 988 /// it points into a macro), this routine returns an invalid 989 /// source location. 990 /// 991 /// \param Offset an offset from the end of the token, where the source 992 /// location should refer to. The default offset (0) produces a source 993 /// location pointing just past the end of the token; an offset of 1 produces 994 /// a source location pointing to the last character in the token, etc. 995 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 996 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 997 } 998 999 /// \brief Returns true if the given MacroID location points at the first 1000 /// token of the macro expansion. 1001 /// 1002 /// \param MacroBegin If non-null and function returns true, it is set to 1003 /// begin location of the macro. 1004 bool isAtStartOfMacroExpansion(SourceLocation loc, 1005 SourceLocation *MacroBegin = 0) const { 1006 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 1007 MacroBegin); 1008 } 1009 1010 /// \brief Returns true if the given MacroID location points at the last 1011 /// token of the macro expansion. 1012 /// 1013 /// \param MacroEnd If non-null and function returns true, it is set to 1014 /// end location of the macro. 1015 bool isAtEndOfMacroExpansion(SourceLocation loc, 1016 SourceLocation *MacroEnd = 0) const { 1017 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 1018 } 1019 1020 /// DumpToken - Print the token to stderr, used for debugging. 1021 /// 1022 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 1023 void DumpLocation(SourceLocation Loc) const; 1024 void DumpMacro(const MacroInfo &MI) const; 1025 1026 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 1027 /// token, return a new location that specifies a character within the token. 1028 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 1029 unsigned Char) const { 1030 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 1031 } 1032 1033 /// IncrementPasteCounter - Increment the counters for the number of token 1034 /// paste operations performed. If fast was specified, this is a 'fast paste' 1035 /// case we handled. 1036 /// 1037 void IncrementPasteCounter(bool isFast) { 1038 if (isFast) 1039 ++NumFastTokenPaste; 1040 else 1041 ++NumTokenPaste; 1042 } 1043 1044 void PrintStats(); 1045 1046 size_t getTotalMemory() const; 1047 1048 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 1049 /// comment (/##/) in microsoft mode, this method handles updating the current 1050 /// state, returning the token on the next source line. 1051 void HandleMicrosoftCommentPaste(Token &Tok); 1052 1053 //===--------------------------------------------------------------------===// 1054 // Preprocessor callback methods. These are invoked by a lexer as various 1055 // directives and events are found. 1056 1057 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1058 /// identifier information for the token and install it into the token, 1059 /// updating the token kind accordingly. 1060 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1061 1062private: 1063 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1064 1065public: 1066 1067 // SetPoisonReason - Call this function to indicate the reason for 1068 // poisoning an identifier. If that identifier is accessed while 1069 // poisoned, then this reason will be used instead of the default 1070 // "poisoned" diagnostic. 1071 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1072 1073 // HandlePoisonedIdentifier - Display reason for poisoned 1074 // identifier. 1075 void HandlePoisonedIdentifier(Token & Tok); 1076 1077 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1078 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1079 if(II->isPoisoned()) { 1080 HandlePoisonedIdentifier(Identifier); 1081 } 1082 } 1083 } 1084 1085private: 1086 /// Identifiers used for SEH handling in Borland. These are only 1087 /// allowed in particular circumstances 1088 // __except block 1089 IdentifierInfo *Ident__exception_code, 1090 *Ident___exception_code, 1091 *Ident_GetExceptionCode; 1092 // __except filter expression 1093 IdentifierInfo *Ident__exception_info, 1094 *Ident___exception_info, 1095 *Ident_GetExceptionInfo; 1096 // __finally 1097 IdentifierInfo *Ident__abnormal_termination, 1098 *Ident___abnormal_termination, 1099 *Ident_AbnormalTermination; 1100public: 1101 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1102 1103 /// HandleIdentifier - This callback is invoked when the lexer reads an 1104 /// identifier and has filled in the tokens IdentifierInfo member. This 1105 /// callback potentially macro expands it or turns it into a named token (like 1106 /// 'for'). 1107 void HandleIdentifier(Token &Identifier); 1108 1109 1110 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1111 /// the current file. This either returns the EOF token and returns true, or 1112 /// pops a level off the include stack and returns false, at which point the 1113 /// client should call lex again. 1114 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1115 1116 /// HandleEndOfTokenLexer - This callback is invoked when the current 1117 /// TokenLexer hits the end of its token stream. 1118 bool HandleEndOfTokenLexer(Token &Result); 1119 1120 /// HandleDirective - This callback is invoked when the lexer sees a # token 1121 /// at the start of a line. This consumes the directive, modifies the 1122 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1123 /// read is the correct one. 1124 void HandleDirective(Token &Result); 1125 1126 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1127 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1128 /// true, then we consider macros that expand to zero tokens as being ok. 1129 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1130 1131 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1132 /// current line until the tok::eod token is found. 1133 void DiscardUntilEndOfDirective(); 1134 1135 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1136 /// __DATE__ or __TIME__ in the file so far. 1137 bool SawDateOrTime() const { 1138 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1139 } 1140 unsigned getCounterValue() const { return CounterValue; } 1141 void setCounterValue(unsigned V) { CounterValue = V; } 1142 1143 /// \brief Retrieves the module that we're currently building, if any. 1144 Module *getCurrentModule(); 1145 1146 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1147 MacroInfo *AllocateMacroInfo(SourceLocation L); 1148 1149 /// \brief Allocate a new MacroInfo object which is clone of \p MI. 1150 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1151 1152 /// \brief Turn the specified lexer token into a fully checked and spelled 1153 /// filename, e.g. as an operand of \#include. 1154 /// 1155 /// The caller is expected to provide a buffer that is large enough to hold 1156 /// the spelling of the filename, but is also expected to handle the case 1157 /// when this method decides to use a different buffer. 1158 /// 1159 /// \returns true if the input filename was in <>'s or false if it was 1160 /// in ""'s. 1161 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1162 1163 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1164 /// 1165 /// Returns null on failure. \p isAngled indicates whether the file 1166 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1167 const FileEntry *LookupFile(StringRef Filename, 1168 bool isAngled, const DirectoryLookup *FromDir, 1169 const DirectoryLookup *&CurDir, 1170 SmallVectorImpl<char> *SearchPath, 1171 SmallVectorImpl<char> *RelativePath, 1172 Module **SuggestedModule, 1173 bool SkipCache = false); 1174 1175 /// GetCurLookup - The DirectoryLookup structure used to find the current 1176 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1177 /// implement \#include_next and find directory-specific properties. 1178 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1179 1180 /// \brief Return true if we're in the top-level file, not in a \#include. 1181 bool isInPrimaryFile() const; 1182 1183 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1184 /// from a macro as multiple tokens, which need to be glued together. This 1185 /// occurs for code like: 1186 /// \code 1187 /// \#define FOO <x/y.h> 1188 /// \#include FOO 1189 /// \endcode 1190 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1191 /// 1192 /// This code concatenates and consumes tokens up to the '>' token. It 1193 /// returns false if the > was found, otherwise it returns true if it finds 1194 /// and consumes the EOD marker. 1195 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1196 SourceLocation &End); 1197 1198 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1199 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1200 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1201 1202private: 1203 1204 void PushIncludeMacroStack() { 1205 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1206 CurLexer.take(), 1207 CurPTHLexer.take(), 1208 CurPPLexer, 1209 CurTokenLexer.take(), 1210 CurDirLookup)); 1211 CurPPLexer = 0; 1212 } 1213 1214 void PopIncludeMacroStack() { 1215 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1216 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1217 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1218 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1219 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1220 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1221 IncludeMacroStack.pop_back(); 1222 } 1223 1224 /// \brief Allocate a new MacroInfo object. 1225 MacroInfo *AllocateMacroInfo(); 1226 1227 /// \brief Release the specified MacroInfo for re-use. 1228 /// 1229 /// This memory will be reused for allocating new MacroInfo objects. 1230 void ReleaseMacroInfo(MacroInfo* MI); 1231 1232 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1233 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1234 /// and discards the rest of the macro line if the macro name is invalid. 1235 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1236 1237 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1238 /// definition has just been read. Lex the rest of the arguments and the 1239 /// closing ), updating MI with what we learn and saving in LastTok the 1240 /// last token read. 1241 /// Return true if an error occurs parsing the arg list. 1242 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1243 1244 /// We just read a \#if or related directive and decided that the 1245 /// subsequent tokens are in the \#if'd out portion of the 1246 /// file. Lex the rest of the file, until we see an \#endif. If \p 1247 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1248 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1249 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1250 /// already seen one so a \#else directive is a duplicate. When this returns, 1251 /// the caller can lex the first valid token. 1252 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1253 bool FoundNonSkipPortion, bool FoundElse, 1254 SourceLocation ElseLoc = SourceLocation()); 1255 1256 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1257 void PTHSkipExcludedConditionalBlock(); 1258 1259 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1260 /// may occur after a #if or #elif directive and return it as a bool. If the 1261 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1262 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1263 1264 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1265 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1266 void RegisterBuiltinPragmas(); 1267 1268 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1269 void RegisterBuiltinMacros(); 1270 1271 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1272 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1273 /// the macro should not be expanded return true, otherwise return false. 1274 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1275 1276 /// \brief Cache macro expanded tokens for TokenLexers. 1277 // 1278 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1279 /// going to lex in the cache and when it finishes the tokens are removed 1280 /// from the end of the cache. 1281 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1282 ArrayRef<Token> tokens); 1283 void removeCachedMacroExpandedTokensOfLastLexer(); 1284 friend void TokenLexer::ExpandFunctionArguments(); 1285 1286 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1287 /// lexed is a '('. If so, consume the token and return true, if not, this 1288 /// method should have no observable side-effect on the lexed tokens. 1289 bool isNextPPTokenLParen(); 1290 1291 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1292 /// invoked to read all of the formal arguments specified for the macro 1293 /// invocation. This returns null on error. 1294 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1295 SourceLocation &ExpansionEnd); 1296 1297 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1298 /// as a builtin macro, handle it and return the next token as 'Tok'. 1299 void ExpandBuiltinMacro(Token &Tok); 1300 1301 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1302 /// return the first token after the directive. The _Pragma token has just 1303 /// been read into 'Tok'. 1304 void Handle_Pragma(Token &Tok); 1305 1306 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1307 /// is not enclosed within a string literal. 1308 void HandleMicrosoft__pragma(Token &Tok); 1309 1310 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1311 /// start lexing tokens from it instead of the current buffer. 1312 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1313 1314 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1315 /// start getting tokens from it using the PTH cache. 1316 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1317 1318 /// IsFileLexer - Returns true if we are lexing from a file and not a 1319 /// pragma or a macro. 1320 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1321 return L ? !L->isPragmaLexer() : P != 0; 1322 } 1323 1324 static bool IsFileLexer(const IncludeStackInfo& I) { 1325 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1326 } 1327 1328 bool IsFileLexer() const { 1329 return IsFileLexer(CurLexer.get(), CurPPLexer); 1330 } 1331 1332 //===--------------------------------------------------------------------===// 1333 // Caching stuff. 1334 void CachingLex(Token &Result); 1335 bool InCachingLexMode() const { 1336 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1337 // that we are past EOF, not that we are in CachingLex mode. 1338 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1339 !IncludeMacroStack.empty(); 1340 } 1341 void EnterCachingLexMode(); 1342 void ExitCachingLexMode() { 1343 if (InCachingLexMode()) 1344 RemoveTopOfLexerStack(); 1345 } 1346 const Token &PeekAhead(unsigned N); 1347 void AnnotatePreviousCachedTokens(const Token &Tok); 1348 1349 //===--------------------------------------------------------------------===// 1350 /// Handle*Directive - implement the various preprocessor directives. These 1351 /// should side-effect the current preprocessor object so that the next call 1352 /// to Lex() will return the appropriate token next. 1353 void HandleLineDirective(Token &Tok); 1354 void HandleDigitDirective(Token &Tok); 1355 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1356 void HandleIdentSCCSDirective(Token &Tok); 1357 void HandleMacroPublicDirective(Token &Tok); 1358 void HandleMacroPrivateDirective(Token &Tok); 1359 1360 // File inclusion. 1361 void HandleIncludeDirective(SourceLocation HashLoc, 1362 Token &Tok, 1363 const DirectoryLookup *LookupFrom = 0, 1364 bool isImport = false); 1365 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1366 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1367 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1368 void HandleMicrosoftImportDirective(Token &Tok); 1369 1370 // Macro handling. 1371 void HandleDefineDirective(Token &Tok); 1372 void HandleUndefDirective(Token &Tok); 1373 void UndefineMacro(IdentifierInfo *II, MacroInfo *MI, 1374 SourceLocation UndefLoc); 1375 1376 // Conditional Inclusion. 1377 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1378 bool ReadAnyTokensBeforeDirective); 1379 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1380 void HandleEndifDirective(Token &Tok); 1381 void HandleElseDirective(Token &Tok); 1382 void HandleElifDirective(Token &Tok); 1383 1384 // Pragmas. 1385 void HandlePragmaDirective(unsigned Introducer); 1386public: 1387 void HandlePragmaOnce(Token &OnceTok); 1388 void HandlePragmaMark(); 1389 void HandlePragmaPoison(Token &PoisonTok); 1390 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1391 void HandlePragmaDependency(Token &DependencyTok); 1392 void HandlePragmaComment(Token &CommentTok); 1393 void HandlePragmaMessage(Token &MessageTok); 1394 void HandlePragmaPushMacro(Token &Tok); 1395 void HandlePragmaPopMacro(Token &Tok); 1396 void HandlePragmaIncludeAlias(Token &Tok); 1397 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1398 1399 // Return true and store the first token only if any CommentHandler 1400 // has inserted some tokens and getCommentRetentionState() is false. 1401 bool HandleComment(Token &Token, SourceRange Comment); 1402 1403 /// \brief A macro is used, update information about macros that need unused 1404 /// warnings. 1405 void markMacroAsUsed(MacroInfo *MI); 1406}; 1407 1408/// \brief Abstract base class that describes a handler that will receive 1409/// source ranges for each of the comments encountered in the source file. 1410class CommentHandler { 1411public: 1412 virtual ~CommentHandler(); 1413 1414 // The handler shall return true if it has pushed any tokens 1415 // to be read using e.g. EnterToken or EnterTokenStream. 1416 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1417}; 1418 1419} // end namespace clang 1420 1421#endif 1422