Preprocessor.h revision cfa88f893915ceb8ae4ce2f17c46c24a4d67502f
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/Basic/Builtins.h" 18#include "clang/Basic/Diagnostic.h" 19#include "clang/Basic/IdentifierTable.h" 20#include "clang/Basic/SourceLocation.h" 21#include "clang/Lex/Lexer.h" 22#include "clang/Lex/MacroInfo.h" 23#include "clang/Lex/PPCallbacks.h" 24#include "clang/Lex/PPMutationListener.h" 25#include "clang/Lex/PTHLexer.h" 26#include "clang/Lex/PTHManager.h" 27#include "clang/Lex/TokenLexer.h" 28#include "llvm/ADT/ArrayRef.h" 29#include "llvm/ADT/DenseMap.h" 30#include "llvm/ADT/IntrusiveRefCntPtr.h" 31#include "llvm/ADT/OwningPtr.h" 32#include "llvm/ADT/SmallPtrSet.h" 33#include "llvm/ADT/SmallVector.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 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 SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> ModuleImportPath; 219 220 /// \brief Whether the module import expectes an identifier next. Otherwise, 221 /// it expects a '.' or ';'. 222 bool ModuleImportExpectsIdentifier; 223 224 /// \brief The source location of the currently-active 225 /// #pragma clang arc_cf_code_audited begin. 226 SourceLocation PragmaARCCFCodeAuditedLoc; 227 228 /// \brief True if we hit the code-completion point. 229 bool CodeCompletionReached; 230 231 /// \brief The number of bytes that we will initially skip when entering the 232 /// main file, which is used when loading a precompiled preamble, along 233 /// with a flag that indicates whether skipping this number of bytes will 234 /// place the lexer at the start of a line. 235 std::pair<unsigned, bool> SkipMainFilePreamble; 236 237 /// CurLexer - This is the current top of the stack that we're lexing from if 238 /// not expanding a macro and we are lexing directly from source code. 239 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 240 OwningPtr<Lexer> CurLexer; 241 242 /// CurPTHLexer - This is the current top of stack that we're lexing from if 243 /// not expanding from a macro and we are lexing from a PTH cache. 244 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 245 OwningPtr<PTHLexer> CurPTHLexer; 246 247 /// CurPPLexer - This is the current top of the stack what we're lexing from 248 /// if not expanding a macro. This is an alias for either CurLexer or 249 /// CurPTHLexer. 250 PreprocessorLexer *CurPPLexer; 251 252 /// CurLookup - The DirectoryLookup structure used to find the current 253 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 254 /// implement \#include_next and find directory-specific properties. 255 const DirectoryLookup *CurDirLookup; 256 257 /// CurTokenLexer - This is the current macro we are expanding, if we are 258 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 259 OwningPtr<TokenLexer> CurTokenLexer; 260 261 /// \brief The kind of lexer we're currently working with. 262 enum CurLexerKind { 263 CLK_Lexer, 264 CLK_PTHLexer, 265 CLK_TokenLexer, 266 CLK_CachingLexer, 267 CLK_LexAfterModuleImport 268 } CurLexerKind; 269 270 /// IncludeMacroStack - This keeps track of the stack of files currently 271 /// \#included, and macros currently being expanded from, not counting 272 /// CurLexer/CurTokenLexer. 273 struct IncludeStackInfo { 274 enum CurLexerKind CurLexerKind; 275 Lexer *TheLexer; 276 PTHLexer *ThePTHLexer; 277 PreprocessorLexer *ThePPLexer; 278 TokenLexer *TheTokenLexer; 279 const DirectoryLookup *TheDirLookup; 280 281 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 282 PreprocessorLexer* PPL, 283 TokenLexer* TL, const DirectoryLookup *D) 284 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 285 TheTokenLexer(TL), TheDirLookup(D) {} 286 }; 287 std::vector<IncludeStackInfo> IncludeMacroStack; 288 289 /// Callbacks - These are actions invoked when some preprocessor activity is 290 /// encountered (e.g. a file is \#included, etc). 291 PPCallbacks *Callbacks; 292 293 /// \brief Listener whose actions are invoked when an entity in the 294 /// preprocessor (e.g., a macro) that was loaded from an AST file is 295 /// later mutated. 296 PPMutationListener *Listener; 297 298 struct MacroExpandsInfo { 299 Token Tok; 300 MacroInfo *MI; 301 SourceRange Range; 302 MacroExpandsInfo(Token Tok, MacroInfo *MI, SourceRange Range) 303 : Tok(Tok), MI(MI), Range(Range) { } 304 }; 305 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 306 307 /// Macros - For each IdentifierInfo that was associated with a macro, we 308 /// keep a mapping to the history of all macro definitions and #undefs in 309 /// the reverse order (the latest one is in the head of the list). 310 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 311 friend class ASTReader; 312 313 /// \brief Macros that we want to warn because they are not used at the end 314 /// of the translation unit; we store just their SourceLocations instead 315 /// something like MacroInfo*. The benefit of this is that when we are 316 /// deserializing from PCH, we don't need to deserialize identifier & macros 317 /// just so that we can report that they are unused, we just warn using 318 /// the SourceLocations of this set (that will be filled by the ASTReader). 319 /// We are using SmallPtrSet instead of a vector for faster removal. 320 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 321 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 322 323 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 324 /// reused for quick allocation. 325 MacroArgs *MacroArgCache; 326 friend class MacroArgs; 327 328 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 329 /// push_macro directive, we keep a MacroInfo stack used to restore 330 /// previous macro value. 331 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 332 333 // Various statistics we track for performance analysis. 334 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 335 unsigned NumIf, NumElse, NumEndif; 336 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 337 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 338 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 339 unsigned NumSkipped; 340 341 /// Predefines - This string is the predefined macros that preprocessor 342 /// should use from the command line etc. 343 std::string Predefines; 344 345 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 346 enum { TokenLexerCacheSize = 8 }; 347 unsigned NumCachedTokenLexers; 348 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 349 350 /// \brief Keeps macro expanded tokens for TokenLexers. 351 // 352 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 353 /// going to lex in the cache and when it finishes the tokens are removed 354 /// from the end of the cache. 355 SmallVector<Token, 16> MacroExpandedTokens; 356 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 357 358 /// \brief A record of the macro definitions and expansions that 359 /// occurred during preprocessing. 360 /// 361 /// This is an optional side structure that can be enabled with 362 /// \c createPreprocessingRecord() prior to preprocessing. 363 PreprocessingRecord *Record; 364 365private: // Cached tokens state. 366 typedef SmallVector<Token, 1> CachedTokensTy; 367 368 /// CachedTokens - Cached tokens are stored here when we do backtracking or 369 /// lookahead. They are "lexed" by the CachingLex() method. 370 CachedTokensTy CachedTokens; 371 372 /// CachedLexPos - The position of the cached token that CachingLex() should 373 /// "lex" next. If it points beyond the CachedTokens vector, it means that 374 /// a normal Lex() should be invoked. 375 CachedTokensTy::size_type CachedLexPos; 376 377 /// BacktrackPositions - Stack of backtrack positions, allowing nested 378 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 379 /// indicate where CachedLexPos should be set when the BackTrack() method is 380 /// invoked (at which point the last position is popped). 381 std::vector<CachedTokensTy::size_type> BacktrackPositions; 382 383 struct MacroInfoChain { 384 MacroInfo MI; 385 MacroInfoChain *Next; 386 MacroInfoChain *Prev; 387 }; 388 389 /// MacroInfos are managed as a chain for easy disposal. This is the head 390 /// of that list. 391 MacroInfoChain *MIChainHead; 392 393 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 394 /// allocation. 395 MacroInfoChain *MICache; 396 397public: 398 Preprocessor(IntrusiveRefCntPtr<PreprocessorOptions> PPOpts, 399 DiagnosticsEngine &diags, LangOptions &opts, 400 const TargetInfo *target, 401 SourceManager &SM, HeaderSearch &Headers, 402 ModuleLoader &TheModuleLoader, 403 IdentifierInfoLookup *IILookup = 0, 404 bool OwnsHeaderSearch = false, 405 bool DelayInitialization = false, 406 bool IncrProcessing = false); 407 408 ~Preprocessor(); 409 410 /// \brief Initialize the preprocessor, if the constructor did not already 411 /// perform the initialization. 412 /// 413 /// \param Target Information about the target. 414 void Initialize(const TargetInfo &Target); 415 416 /// \brief Retrieve the preprocessor options used to initialize this 417 /// preprocessor. 418 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; } 419 420 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 421 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 422 423 const LangOptions &getLangOpts() const { return LangOpts; } 424 const TargetInfo &getTargetInfo() const { return *Target; } 425 FileManager &getFileManager() const { return FileMgr; } 426 SourceManager &getSourceManager() const { return SourceMgr; } 427 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 428 429 IdentifierTable &getIdentifierTable() { return Identifiers; } 430 SelectorTable &getSelectorTable() { return Selectors; } 431 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 432 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 433 434 void setPTHManager(PTHManager* pm); 435 436 PTHManager *getPTHManager() { return PTH.get(); } 437 438 void setExternalSource(ExternalPreprocessorSource *Source) { 439 ExternalSource = Source; 440 } 441 442 ExternalPreprocessorSource *getExternalSource() const { 443 return ExternalSource; 444 } 445 446 /// \brief Retrieve the module loader associated with this preprocessor. 447 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 448 449 /// SetCommentRetentionState - Control whether or not the preprocessor retains 450 /// comments in output. 451 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 452 this->KeepComments = KeepComments | KeepMacroComments; 453 this->KeepMacroComments = KeepMacroComments; 454 } 455 456 bool getCommentRetentionState() const { return KeepComments; } 457 458 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 459 bool getPragmasEnabled() const { return PragmasEnabled; } 460 461 void SetSuppressIncludeNotFoundError(bool Suppress) { 462 SuppressIncludeNotFoundError = Suppress; 463 } 464 465 bool GetSuppressIncludeNotFoundError() { 466 return SuppressIncludeNotFoundError; 467 } 468 469 /// isCurrentLexer - Return true if we are lexing directly from the specified 470 /// lexer. 471 bool isCurrentLexer(const PreprocessorLexer *L) const { 472 return CurPPLexer == L; 473 } 474 475 /// getCurrentLexer - Return the current lexer being lexed from. Note 476 /// that this ignores any potentially active macro expansions and _Pragma 477 /// expansions going on at the time. 478 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 479 480 /// getCurrentFileLexer - Return the current file lexer being lexed from. 481 /// Note that this ignores any potentially active macro expansions and _Pragma 482 /// expansions going on at the time. 483 PreprocessorLexer *getCurrentFileLexer() const; 484 485 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 486 /// Note that this class takes ownership of any PPCallbacks object given to 487 /// it. 488 PPCallbacks *getPPCallbacks() const { return Callbacks; } 489 void addPPCallbacks(PPCallbacks *C) { 490 if (Callbacks) 491 C = new PPChainedCallbacks(C, Callbacks); 492 Callbacks = C; 493 } 494 495 /// \brief Attach an preprocessor mutation listener to the preprocessor. 496 /// 497 /// The preprocessor mutation listener provides the ability to track 498 /// modifications to the preprocessor entities committed after they were 499 /// initially created. 500 void setPPMutationListener(PPMutationListener *Listener) { 501 this->Listener = Listener; 502 } 503 504 /// \brief Retrieve a pointer to the preprocessor mutation listener 505 /// associated with this preprocessor, if any. 506 PPMutationListener *getPPMutationListener() const { return Listener; } 507 508 /// \brief Given an identifier, return the MacroInfo it is \#defined to 509 /// or null if it isn't \#define'd. 510 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 511 if (!II->hasMacroDefinition()) 512 return 0; 513 514 MacroInfo *MI = getMacroInfoHistory(II); 515 assert(MI->getUndefLoc().isInvalid() && "Macro is undefined!"); 516 return MI; 517 } 518 519 /// \brief Given an identifier, return the (probably #undef'd) MacroInfo 520 /// representing the most recent macro definition. One can iterate over all 521 /// previous macro definitions from it. This method should only be called for 522 /// identifiers that hadMacroDefinition(). 523 MacroInfo *getMacroInfoHistory(IdentifierInfo *II) const; 524 525 /// \brief Specify a macro for this identifier. 526 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 527 /// \brief Add a MacroInfo that was loaded from an AST file. 528 void addLoadedMacroInfo(IdentifierInfo *II, MacroInfo *MI, 529 MacroInfo *Hint = 0); 530 /// \brief Make the given MacroInfo, that was loaded from an AST file and 531 /// previously hidden, visible. 532 void makeLoadedMacroInfoVisible(IdentifierInfo *II, MacroInfo *MI); 533 /// \brief Undefine a macro for this identifier. 534 void clearMacroInfo(IdentifierInfo *II); 535 536 /// macro_iterator/macro_begin/macro_end - This allows you to walk the macro 537 /// history table. Currently defined macros have 538 /// IdentifierInfo::hasMacroDefinition() set and an empty 539 /// MacroInfo::getUndefLoc() at the head of the list. 540 typedef llvm::DenseMap<IdentifierInfo*, 541 MacroInfo*>::const_iterator macro_iterator; 542 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 543 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 544 545 /// \brief Return the name of the macro defined before \p Loc that has 546 /// spelling \p Tokens. If there are multiple macros with same spelling, 547 /// return the last one defined. 548 StringRef getLastMacroWithSpelling(SourceLocation Loc, 549 ArrayRef<TokenValue> Tokens) const; 550 551 const std::string &getPredefines() const { return Predefines; } 552 /// setPredefines - Set the predefines for this Preprocessor. These 553 /// predefines are automatically injected when parsing the main file. 554 void setPredefines(const char *P) { Predefines = P; } 555 void setPredefines(const std::string &P) { Predefines = P; } 556 557 /// Return information about the specified preprocessor 558 /// identifier token. 559 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 560 return &Identifiers.get(Name); 561 } 562 563 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 564 /// If 'Namespace' is non-null, then it is a token required to exist on the 565 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 566 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 567 void AddPragmaHandler(PragmaHandler *Handler) { 568 AddPragmaHandler(StringRef(), Handler); 569 } 570 571 /// RemovePragmaHandler - Remove the specific pragma handler from 572 /// the preprocessor. If \p Namespace is non-null, then it should 573 /// be the namespace that \p Handler was added to. It is an error 574 /// to remove a handler that has not been registered. 575 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 576 void RemovePragmaHandler(PragmaHandler *Handler) { 577 RemovePragmaHandler(StringRef(), Handler); 578 } 579 580 /// \brief Add the specified comment handler to the preprocessor. 581 void addCommentHandler(CommentHandler *Handler); 582 583 /// \brief Remove the specified comment handler. 584 /// 585 /// It is an error to remove a handler that has not been registered. 586 void removeCommentHandler(CommentHandler *Handler); 587 588 /// \brief Set the code completion handler to the given object. 589 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 590 CodeComplete = &Handler; 591 } 592 593 /// \brief Retrieve the current code-completion handler. 594 CodeCompletionHandler *getCodeCompletionHandler() const { 595 return CodeComplete; 596 } 597 598 /// \brief Clear out the code completion handler. 599 void clearCodeCompletionHandler() { 600 CodeComplete = 0; 601 } 602 603 /// \brief Hook used by the lexer to invoke the "natural language" code 604 /// completion point. 605 void CodeCompleteNaturalLanguage(); 606 607 /// \brief Retrieve the preprocessing record, or NULL if there is no 608 /// preprocessing record. 609 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 610 611 /// \brief Create a new preprocessing record, which will keep track of 612 /// all macro expansions, macro definitions, etc. 613 void createPreprocessingRecord(); 614 615 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 616 /// which implicitly adds the builtin defines etc. 617 void EnterMainSourceFile(); 618 619 /// EndSourceFile - Inform the preprocessor callbacks that processing is 620 /// complete. 621 void EndSourceFile(); 622 623 /// EnterSourceFile - Add a source file to the top of the include stack and 624 /// start lexing tokens from it instead of the current buffer. Emit an error 625 /// and don't enter the file on error. 626 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 627 SourceLocation Loc); 628 629 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 630 /// tokens from it instead of the current buffer. Args specifies the 631 /// tokens input to a function-like macro. 632 /// 633 /// ILEnd specifies the location of the ')' for a function-like macro or the 634 /// identifier for an object-like macro. 635 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 636 MacroArgs *Args); 637 638 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 639 /// which will cause the lexer to start returning the specified tokens. 640 /// 641 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 642 /// not be subject to further macro expansion. Otherwise, these tokens will 643 /// be re-macro-expanded when/if expansion is enabled. 644 /// 645 /// If OwnsTokens is false, this method assumes that the specified stream of 646 /// tokens has a permanent owner somewhere, so they do not need to be copied. 647 /// If it is true, it assumes the array of tokens is allocated with new[] and 648 /// must be freed. 649 /// 650 void EnterTokenStream(const Token *Toks, unsigned NumToks, 651 bool DisableMacroExpansion, bool OwnsTokens); 652 653 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 654 /// lexer stack. This should only be used in situations where the current 655 /// state of the top-of-stack lexer is known. 656 void RemoveTopOfLexerStack(); 657 658 /// EnableBacktrackAtThisPos - From the point that this method is called, and 659 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 660 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 661 /// make the Preprocessor re-lex the same tokens. 662 /// 663 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 664 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 665 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 666 /// 667 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 668 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 669 /// tokens will continue indefinitely. 670 /// 671 void EnableBacktrackAtThisPos(); 672 673 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 674 void CommitBacktrackedTokens(); 675 676 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 677 /// EnableBacktrackAtThisPos() was previously called. 678 void Backtrack(); 679 680 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 681 /// caching of tokens is on. 682 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 683 684 /// Lex - To lex a token from the preprocessor, just pull a token from the 685 /// current lexer or macro object. 686 void Lex(Token &Result) { 687 switch (CurLexerKind) { 688 case CLK_Lexer: CurLexer->Lex(Result); break; 689 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 690 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 691 case CLK_CachingLexer: CachingLex(Result); break; 692 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 693 } 694 } 695 696 void LexAfterModuleImport(Token &Result); 697 698 /// \brief Lex a string literal, which may be the concatenation of multiple 699 /// string literals and may even come from macro expansion. 700 /// \returns true on success, false if a error diagnostic has been generated. 701 bool LexStringLiteral(Token &Result, std::string &String, 702 const char *DiagnosticTag, bool AllowMacroExpansion) { 703 if (AllowMacroExpansion) 704 Lex(Result); 705 else 706 LexUnexpandedToken(Result); 707 return FinishLexStringLiteral(Result, String, DiagnosticTag, 708 AllowMacroExpansion); 709 } 710 711 /// \brief Complete the lexing of a string literal where the first token has 712 /// already been lexed (see LexStringLiteral). 713 bool FinishLexStringLiteral(Token &Result, std::string &String, 714 const char *DiagnosticTag, 715 bool AllowMacroExpansion); 716 717 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 718 /// something not a comment. This is useful in -E -C mode where comments 719 /// would foul up preprocessor directive handling. 720 void LexNonComment(Token &Result) { 721 do 722 Lex(Result); 723 while (Result.getKind() == tok::comment); 724 } 725 726 /// LexUnexpandedToken - This is just like Lex, but this disables macro 727 /// expansion of identifier tokens. 728 void LexUnexpandedToken(Token &Result) { 729 // Disable macro expansion. 730 bool OldVal = DisableMacroExpansion; 731 DisableMacroExpansion = true; 732 // Lex the token. 733 Lex(Result); 734 735 // Reenable it. 736 DisableMacroExpansion = OldVal; 737 } 738 739 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 740 /// expansion of identifier tokens. 741 void LexUnexpandedNonComment(Token &Result) { 742 do 743 LexUnexpandedToken(Result); 744 while (Result.getKind() == tok::comment); 745 } 746 747 /// Disables macro expansion everywhere except for preprocessor directives. 748 void SetMacroExpansionOnlyInDirectives() { 749 DisableMacroExpansion = true; 750 MacroExpansionInDirectivesOverride = true; 751 } 752 753 /// LookAhead - This peeks ahead N tokens and returns that token without 754 /// consuming any tokens. LookAhead(0) returns the next token that would be 755 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 756 /// returns normal tokens after phase 5. As such, it is equivalent to using 757 /// 'Lex', not 'LexUnexpandedToken'. 758 const Token &LookAhead(unsigned N) { 759 if (CachedLexPos + N < CachedTokens.size()) 760 return CachedTokens[CachedLexPos+N]; 761 else 762 return PeekAhead(N+1); 763 } 764 765 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 766 /// this allows to revert a specific number of tokens. 767 /// Note that the number of tokens being reverted should be up to the last 768 /// backtrack position, not more. 769 void RevertCachedTokens(unsigned N) { 770 assert(isBacktrackEnabled() && 771 "Should only be called when tokens are cached for backtracking"); 772 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 773 && "Should revert tokens up to the last backtrack position, not more"); 774 assert(signed(CachedLexPos) - signed(N) >= 0 && 775 "Corrupted backtrack positions ?"); 776 CachedLexPos -= N; 777 } 778 779 /// EnterToken - Enters a token in the token stream to be lexed next. If 780 /// BackTrack() is called afterwards, the token will remain at the insertion 781 /// point. 782 void EnterToken(const Token &Tok) { 783 EnterCachingLexMode(); 784 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 785 } 786 787 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 788 /// tokens (because backtrack is enabled) it should replace the most recent 789 /// cached tokens with the given annotation token. This function has no effect 790 /// if backtracking is not enabled. 791 /// 792 /// Note that the use of this function is just for optimization; so that the 793 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 794 /// invoked. 795 void AnnotateCachedTokens(const Token &Tok) { 796 assert(Tok.isAnnotation() && "Expected annotation token"); 797 if (CachedLexPos != 0 && isBacktrackEnabled()) 798 AnnotatePreviousCachedTokens(Tok); 799 } 800 801 /// \brief Replace the last token with an annotation token. 802 /// 803 /// Like AnnotateCachedTokens(), this routine replaces an 804 /// already-parsed (and resolved) token with an annotation 805 /// token. However, this routine only replaces the last token with 806 /// the annotation token; it does not affect any other cached 807 /// tokens. This function has no effect if backtracking is not 808 /// enabled. 809 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 810 assert(Tok.isAnnotation() && "Expected annotation token"); 811 if (CachedLexPos != 0 && isBacktrackEnabled()) 812 CachedTokens[CachedLexPos-1] = Tok; 813 } 814 815 /// TypoCorrectToken - Update the current token to represent the provided 816 /// identifier, in order to cache an action performed by typo correction. 817 void TypoCorrectToken(const Token &Tok) { 818 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 819 if (CachedLexPos != 0 && isBacktrackEnabled()) 820 CachedTokens[CachedLexPos-1] = Tok; 821 } 822 823 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 824 /// CurTokenLexer pointers. 825 void recomputeCurLexerKind(); 826 827 /// \brief Returns true if incremental processing is enabled 828 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 829 830 /// \brief Enables the incremental processing 831 void enableIncrementalProcessing(bool value = true) { 832 IncrementalProcessing = value; 833 } 834 835 /// \brief Specify the point at which code-completion will be performed. 836 /// 837 /// \param File the file in which code completion should occur. If 838 /// this file is included multiple times, code-completion will 839 /// perform completion the first time it is included. If NULL, this 840 /// function clears out the code-completion point. 841 /// 842 /// \param Line the line at which code completion should occur 843 /// (1-based). 844 /// 845 /// \param Column the column at which code completion should occur 846 /// (1-based). 847 /// 848 /// \returns true if an error occurred, false otherwise. 849 bool SetCodeCompletionPoint(const FileEntry *File, 850 unsigned Line, unsigned Column); 851 852 /// \brief Determine if we are performing code completion. 853 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 854 855 /// \brief Returns the location of the code-completion point. 856 /// Returns an invalid location if code-completion is not enabled or the file 857 /// containing the code-completion point has not been lexed yet. 858 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 859 860 /// \brief Returns the start location of the file of code-completion point. 861 /// Returns an invalid location if code-completion is not enabled or the file 862 /// containing the code-completion point has not been lexed yet. 863 SourceLocation getCodeCompletionFileLoc() const { 864 return CodeCompletionFileLoc; 865 } 866 867 /// \brief Returns true if code-completion is enabled and we have hit the 868 /// code-completion point. 869 bool isCodeCompletionReached() const { return CodeCompletionReached; } 870 871 /// \brief Note that we hit the code-completion point. 872 void setCodeCompletionReached() { 873 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 874 CodeCompletionReached = true; 875 // Silence any diagnostics that occur after we hit the code-completion. 876 getDiagnostics().setSuppressAllDiagnostics(true); 877 } 878 879 /// \brief The location of the currently-active \#pragma clang 880 /// arc_cf_code_audited begin. Returns an invalid location if there 881 /// is no such pragma active. 882 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 883 return PragmaARCCFCodeAuditedLoc; 884 } 885 886 /// \brief Set the location of the currently-active \#pragma clang 887 /// arc_cf_code_audited begin. An invalid location ends the pragma. 888 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 889 PragmaARCCFCodeAuditedLoc = Loc; 890 } 891 892 /// \brief Instruct the preprocessor to skip part of the main source file. 893 /// 894 /// \param Bytes The number of bytes in the preamble to skip. 895 /// 896 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 897 /// start of a line. 898 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 899 SkipMainFilePreamble.first = Bytes; 900 SkipMainFilePreamble.second = StartOfLine; 901 } 902 903 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 904 /// the specified Token's location, translating the token's start 905 /// position in the current buffer into a SourcePosition object for rendering. 906 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 907 return Diags->Report(Loc, DiagID); 908 } 909 910 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 911 return Diags->Report(Tok.getLocation(), DiagID); 912 } 913 914 /// getSpelling() - Return the 'spelling' of the token at the given 915 /// location; does not go up to the spelling location or down to the 916 /// expansion location. 917 /// 918 /// \param buffer A buffer which will be used only if the token requires 919 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 920 /// \param invalid If non-null, will be set \c true if an error occurs. 921 StringRef getSpelling(SourceLocation loc, 922 SmallVectorImpl<char> &buffer, 923 bool *invalid = 0) const { 924 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 925 } 926 927 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 928 /// token is the characters used to represent the token in the source file 929 /// after trigraph expansion and escaped-newline folding. In particular, this 930 /// wants to get the true, uncanonicalized, spelling of things like digraphs 931 /// UCNs, etc. 932 /// 933 /// \param Invalid If non-null, will be set \c true if an error occurs. 934 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 935 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 936 } 937 938 /// getSpelling - This method is used to get the spelling of a token into a 939 /// preallocated buffer, instead of as an std::string. The caller is required 940 /// to allocate enough space for the token, which is guaranteed to be at least 941 /// Tok.getLength() bytes long. The length of the actual result is returned. 942 /// 943 /// Note that this method may do two possible things: it may either fill in 944 /// the buffer specified with characters, or it may *change the input pointer* 945 /// to point to a constant buffer with the data already in it (avoiding a 946 /// copy). The caller is not allowed to modify the returned buffer pointer 947 /// if an internal buffer is returned. 948 unsigned getSpelling(const Token &Tok, const char *&Buffer, 949 bool *Invalid = 0) const { 950 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 951 } 952 953 /// getSpelling - This method is used to get the spelling of a token into a 954 /// SmallVector. Note that the returned StringRef may not point to the 955 /// supplied buffer if a copy can be avoided. 956 StringRef getSpelling(const Token &Tok, 957 SmallVectorImpl<char> &Buffer, 958 bool *Invalid = 0) const; 959 960 /// \brief Relex the token at the specified location. 961 /// \returns true if there was a failure, false on success. 962 bool getRawToken(SourceLocation Loc, Token &Result) { 963 return Lexer::getRawToken(Loc, Result, SourceMgr, LangOpts); 964 } 965 966 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 967 /// with length 1, return the character. 968 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 969 bool *Invalid = 0) const { 970 assert(Tok.is(tok::numeric_constant) && 971 Tok.getLength() == 1 && "Called on unsupported token"); 972 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 973 974 // If the token is carrying a literal data pointer, just use it. 975 if (const char *D = Tok.getLiteralData()) 976 return *D; 977 978 // Otherwise, fall back on getCharacterData, which is slower, but always 979 // works. 980 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 981 } 982 983 /// \brief Retrieve the name of the immediate macro expansion. 984 /// 985 /// This routine starts from a source location, and finds the name of the macro 986 /// responsible for its immediate expansion. It looks through any intervening 987 /// macro argument expansions to compute this. It returns a StringRef which 988 /// refers to the SourceManager-owned buffer of the source where that macro 989 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 990 StringRef getImmediateMacroName(SourceLocation Loc) { 991 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 992 } 993 994 /// CreateString - Plop the specified string into a scratch buffer and set the 995 /// specified token's location and length to it. If specified, the source 996 /// location provides a location of the expansion point of the token. 997 void CreateString(StringRef Str, Token &Tok, 998 SourceLocation ExpansionLocStart = SourceLocation(), 999 SourceLocation ExpansionLocEnd = SourceLocation()); 1000 1001 /// \brief Computes the source location just past the end of the 1002 /// token at this source location. 1003 /// 1004 /// This routine can be used to produce a source location that 1005 /// points just past the end of the token referenced by \p Loc, and 1006 /// is generally used when a diagnostic needs to point just after a 1007 /// token where it expected something different that it received. If 1008 /// the returned source location would not be meaningful (e.g., if 1009 /// it points into a macro), this routine returns an invalid 1010 /// source location. 1011 /// 1012 /// \param Offset an offset from the end of the token, where the source 1013 /// location should refer to. The default offset (0) produces a source 1014 /// location pointing just past the end of the token; an offset of 1 produces 1015 /// a source location pointing to the last character in the token, etc. 1016 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 1017 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 1018 } 1019 1020 /// \brief Returns true if the given MacroID location points at the first 1021 /// token of the macro expansion. 1022 /// 1023 /// \param MacroBegin If non-null and function returns true, it is set to 1024 /// begin location of the macro. 1025 bool isAtStartOfMacroExpansion(SourceLocation loc, 1026 SourceLocation *MacroBegin = 0) const { 1027 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 1028 MacroBegin); 1029 } 1030 1031 /// \brief Returns true if the given MacroID location points at the last 1032 /// token of the macro expansion. 1033 /// 1034 /// \param MacroEnd If non-null and function returns true, it is set to 1035 /// end location of the macro. 1036 bool isAtEndOfMacroExpansion(SourceLocation loc, 1037 SourceLocation *MacroEnd = 0) const { 1038 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 1039 } 1040 1041 /// DumpToken - Print the token to stderr, used for debugging. 1042 /// 1043 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 1044 void DumpLocation(SourceLocation Loc) const; 1045 void DumpMacro(const MacroInfo &MI) const; 1046 1047 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 1048 /// token, return a new location that specifies a character within the token. 1049 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 1050 unsigned Char) const { 1051 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 1052 } 1053 1054 /// IncrementPasteCounter - Increment the counters for the number of token 1055 /// paste operations performed. If fast was specified, this is a 'fast paste' 1056 /// case we handled. 1057 /// 1058 void IncrementPasteCounter(bool isFast) { 1059 if (isFast) 1060 ++NumFastTokenPaste; 1061 else 1062 ++NumTokenPaste; 1063 } 1064 1065 void PrintStats(); 1066 1067 size_t getTotalMemory() const; 1068 1069 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 1070 /// comment (/##/) in microsoft mode, this method handles updating the current 1071 /// state, returning the token on the next source line. 1072 void HandleMicrosoftCommentPaste(Token &Tok); 1073 1074 //===--------------------------------------------------------------------===// 1075 // Preprocessor callback methods. These are invoked by a lexer as various 1076 // directives and events are found. 1077 1078 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1079 /// identifier information for the token and install it into the token, 1080 /// updating the token kind accordingly. 1081 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1082 1083private: 1084 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1085 1086public: 1087 1088 // SetPoisonReason - Call this function to indicate the reason for 1089 // poisoning an identifier. If that identifier is accessed while 1090 // poisoned, then this reason will be used instead of the default 1091 // "poisoned" diagnostic. 1092 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1093 1094 // HandlePoisonedIdentifier - Display reason for poisoned 1095 // identifier. 1096 void HandlePoisonedIdentifier(Token & Tok); 1097 1098 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1099 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1100 if(II->isPoisoned()) { 1101 HandlePoisonedIdentifier(Identifier); 1102 } 1103 } 1104 } 1105 1106private: 1107 /// Identifiers used for SEH handling in Borland. These are only 1108 /// allowed in particular circumstances 1109 // __except block 1110 IdentifierInfo *Ident__exception_code, 1111 *Ident___exception_code, 1112 *Ident_GetExceptionCode; 1113 // __except filter expression 1114 IdentifierInfo *Ident__exception_info, 1115 *Ident___exception_info, 1116 *Ident_GetExceptionInfo; 1117 // __finally 1118 IdentifierInfo *Ident__abnormal_termination, 1119 *Ident___abnormal_termination, 1120 *Ident_AbnormalTermination; 1121public: 1122 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1123 1124 /// HandleIdentifier - This callback is invoked when the lexer reads an 1125 /// identifier and has filled in the tokens IdentifierInfo member. This 1126 /// callback potentially macro expands it or turns it into a named token (like 1127 /// 'for'). 1128 void HandleIdentifier(Token &Identifier); 1129 1130 1131 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1132 /// the current file. This either returns the EOF token and returns true, or 1133 /// pops a level off the include stack and returns false, at which point the 1134 /// client should call lex again. 1135 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1136 1137 /// HandleEndOfTokenLexer - This callback is invoked when the current 1138 /// TokenLexer hits the end of its token stream. 1139 bool HandleEndOfTokenLexer(Token &Result); 1140 1141 /// HandleDirective - This callback is invoked when the lexer sees a # token 1142 /// at the start of a line. This consumes the directive, modifies the 1143 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1144 /// read is the correct one. 1145 void HandleDirective(Token &Result); 1146 1147 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1148 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1149 /// true, then we consider macros that expand to zero tokens as being ok. 1150 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1151 1152 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1153 /// current line until the tok::eod token is found. 1154 void DiscardUntilEndOfDirective(); 1155 1156 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1157 /// __DATE__ or __TIME__ in the file so far. 1158 bool SawDateOrTime() const { 1159 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1160 } 1161 unsigned getCounterValue() const { return CounterValue; } 1162 void setCounterValue(unsigned V) { CounterValue = V; } 1163 1164 /// \brief Retrieves the module that we're currently building, if any. 1165 Module *getCurrentModule(); 1166 1167 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1168 MacroInfo *AllocateMacroInfo(SourceLocation L); 1169 1170 /// \brief Allocate a new MacroInfo object which is clone of \p MI. 1171 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1172 1173 /// \brief Turn the specified lexer token into a fully checked and spelled 1174 /// filename, e.g. as an operand of \#include. 1175 /// 1176 /// The caller is expected to provide a buffer that is large enough to hold 1177 /// the spelling of the filename, but is also expected to handle the case 1178 /// when this method decides to use a different buffer. 1179 /// 1180 /// \returns true if the input filename was in <>'s or false if it was 1181 /// in ""'s. 1182 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1183 1184 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1185 /// 1186 /// Returns null on failure. \p isAngled indicates whether the file 1187 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1188 const FileEntry *LookupFile(StringRef Filename, 1189 bool isAngled, const DirectoryLookup *FromDir, 1190 const DirectoryLookup *&CurDir, 1191 SmallVectorImpl<char> *SearchPath, 1192 SmallVectorImpl<char> *RelativePath, 1193 Module **SuggestedModule, 1194 bool SkipCache = false); 1195 1196 /// GetCurLookup - The DirectoryLookup structure used to find the current 1197 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1198 /// implement \#include_next and find directory-specific properties. 1199 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1200 1201 /// \brief Return true if we're in the top-level file, not in a \#include. 1202 bool isInPrimaryFile() const; 1203 1204 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1205 /// from a macro as multiple tokens, which need to be glued together. This 1206 /// occurs for code like: 1207 /// \code 1208 /// \#define FOO <x/y.h> 1209 /// \#include FOO 1210 /// \endcode 1211 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1212 /// 1213 /// This code concatenates and consumes tokens up to the '>' token. It 1214 /// returns false if the > was found, otherwise it returns true if it finds 1215 /// and consumes the EOD marker. 1216 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1217 SourceLocation &End); 1218 1219 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1220 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1221 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1222 1223private: 1224 1225 void PushIncludeMacroStack() { 1226 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1227 CurLexer.take(), 1228 CurPTHLexer.take(), 1229 CurPPLexer, 1230 CurTokenLexer.take(), 1231 CurDirLookup)); 1232 CurPPLexer = 0; 1233 } 1234 1235 void PopIncludeMacroStack() { 1236 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1237 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1238 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1239 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1240 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1241 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1242 IncludeMacroStack.pop_back(); 1243 } 1244 1245 /// \brief Allocate a new MacroInfo object. 1246 MacroInfo *AllocateMacroInfo(); 1247 1248 /// \brief Release the specified MacroInfo for re-use. 1249 /// 1250 /// This memory will be reused for allocating new MacroInfo objects. 1251 void ReleaseMacroInfo(MacroInfo* MI); 1252 1253 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1254 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1255 /// and discards the rest of the macro line if the macro name is invalid. 1256 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1257 1258 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1259 /// definition has just been read. Lex the rest of the arguments and the 1260 /// closing ), updating MI with what we learn and saving in LastTok the 1261 /// last token read. 1262 /// Return true if an error occurs parsing the arg list. 1263 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1264 1265 /// We just read a \#if or related directive and decided that the 1266 /// subsequent tokens are in the \#if'd out portion of the 1267 /// file. Lex the rest of the file, until we see an \#endif. If \p 1268 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1269 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1270 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1271 /// already seen one so a \#else directive is a duplicate. When this returns, 1272 /// the caller can lex the first valid token. 1273 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1274 bool FoundNonSkipPortion, bool FoundElse, 1275 SourceLocation ElseLoc = SourceLocation()); 1276 1277 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1278 void PTHSkipExcludedConditionalBlock(); 1279 1280 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1281 /// may occur after a #if or #elif directive and return it as a bool. If the 1282 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1283 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1284 1285 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1286 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1287 void RegisterBuiltinPragmas(); 1288 1289 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1290 void RegisterBuiltinMacros(); 1291 1292 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1293 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1294 /// the macro should not be expanded return true, otherwise return false. 1295 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1296 1297 /// \brief Cache macro expanded tokens for TokenLexers. 1298 // 1299 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1300 /// going to lex in the cache and when it finishes the tokens are removed 1301 /// from the end of the cache. 1302 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1303 ArrayRef<Token> tokens); 1304 void removeCachedMacroExpandedTokensOfLastLexer(); 1305 friend void TokenLexer::ExpandFunctionArguments(); 1306 1307 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1308 /// lexed is a '('. If so, consume the token and return true, if not, this 1309 /// method should have no observable side-effect on the lexed tokens. 1310 bool isNextPPTokenLParen(); 1311 1312 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1313 /// invoked to read all of the formal arguments specified for the macro 1314 /// invocation. This returns null on error. 1315 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1316 SourceLocation &ExpansionEnd); 1317 1318 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1319 /// as a builtin macro, handle it and return the next token as 'Tok'. 1320 void ExpandBuiltinMacro(Token &Tok); 1321 1322 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1323 /// return the first token after the directive. The _Pragma token has just 1324 /// been read into 'Tok'. 1325 void Handle_Pragma(Token &Tok); 1326 1327 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1328 /// is not enclosed within a string literal. 1329 void HandleMicrosoft__pragma(Token &Tok); 1330 1331 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1332 /// start lexing tokens from it instead of the current buffer. 1333 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1334 1335 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1336 /// start getting tokens from it using the PTH cache. 1337 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1338 1339 /// IsFileLexer - Returns true if we are lexing from a file and not a 1340 /// pragma or a macro. 1341 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1342 return L ? !L->isPragmaLexer() : P != 0; 1343 } 1344 1345 static bool IsFileLexer(const IncludeStackInfo& I) { 1346 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1347 } 1348 1349 bool IsFileLexer() const { 1350 return IsFileLexer(CurLexer.get(), CurPPLexer); 1351 } 1352 1353 //===--------------------------------------------------------------------===// 1354 // Caching stuff. 1355 void CachingLex(Token &Result); 1356 bool InCachingLexMode() const { 1357 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1358 // that we are past EOF, not that we are in CachingLex mode. 1359 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1360 !IncludeMacroStack.empty(); 1361 } 1362 void EnterCachingLexMode(); 1363 void ExitCachingLexMode() { 1364 if (InCachingLexMode()) 1365 RemoveTopOfLexerStack(); 1366 } 1367 const Token &PeekAhead(unsigned N); 1368 void AnnotatePreviousCachedTokens(const Token &Tok); 1369 1370 //===--------------------------------------------------------------------===// 1371 /// Handle*Directive - implement the various preprocessor directives. These 1372 /// should side-effect the current preprocessor object so that the next call 1373 /// to Lex() will return the appropriate token next. 1374 void HandleLineDirective(Token &Tok); 1375 void HandleDigitDirective(Token &Tok); 1376 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1377 void HandleIdentSCCSDirective(Token &Tok); 1378 void HandleMacroPublicDirective(Token &Tok); 1379 void HandleMacroPrivateDirective(Token &Tok); 1380 1381 // File inclusion. 1382 void HandleIncludeDirective(SourceLocation HashLoc, 1383 Token &Tok, 1384 const DirectoryLookup *LookupFrom = 0, 1385 bool isImport = false); 1386 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1387 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1388 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1389 void HandleMicrosoftImportDirective(Token &Tok); 1390 1391 // Macro handling. 1392 void HandleDefineDirective(Token &Tok); 1393 void HandleUndefDirective(Token &Tok); 1394 void UndefineMacro(IdentifierInfo *II, MacroInfo *MI, 1395 SourceLocation UndefLoc); 1396 1397 // Conditional Inclusion. 1398 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1399 bool ReadAnyTokensBeforeDirective); 1400 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1401 void HandleEndifDirective(Token &Tok); 1402 void HandleElseDirective(Token &Tok); 1403 void HandleElifDirective(Token &Tok); 1404 1405 // Pragmas. 1406 void HandlePragmaDirective(unsigned Introducer); 1407public: 1408 void HandlePragmaOnce(Token &OnceTok); 1409 void HandlePragmaMark(); 1410 void HandlePragmaPoison(Token &PoisonTok); 1411 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1412 void HandlePragmaDependency(Token &DependencyTok); 1413 void HandlePragmaComment(Token &CommentTok); 1414 void HandlePragmaMessage(Token &MessageTok); 1415 void HandlePragmaPushMacro(Token &Tok); 1416 void HandlePragmaPopMacro(Token &Tok); 1417 void HandlePragmaIncludeAlias(Token &Tok); 1418 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1419 1420 // Return true and store the first token only if any CommentHandler 1421 // has inserted some tokens and getCommentRetentionState() is false. 1422 bool HandleComment(Token &Token, SourceRange Comment); 1423 1424 /// \brief A macro is used, update information about macros that need unused 1425 /// warnings. 1426 void markMacroAsUsed(MacroInfo *MI); 1427}; 1428 1429/// \brief Abstract base class that describes a handler that will receive 1430/// source ranges for each of the comments encountered in the source file. 1431class CommentHandler { 1432public: 1433 virtual ~CommentHandler(); 1434 1435 // The handler shall return true if it has pushed any tokens 1436 // to be read using e.g. EnterToken or EnterTokenStream. 1437 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1438}; 1439 1440} // end namespace clang 1441 1442#endif 1443