Preprocessor.h revision 37ed12720a35b7bfa1c4de73ad6f1c6c1c88ee17
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 /// Return information about the specified preprocessor 559 /// identifier token. 560 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 561 return &Identifiers.get(Name); 562 } 563 564 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 565 /// If 'Namespace' is non-null, then it is a token required to exist on the 566 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 567 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 568 void AddPragmaHandler(PragmaHandler *Handler) { 569 AddPragmaHandler(StringRef(), Handler); 570 } 571 572 /// RemovePragmaHandler - Remove the specific pragma handler from 573 /// the preprocessor. If \p Namespace is non-null, then it should 574 /// be the namespace that \p Handler was added to. It is an error 575 /// to remove a handler that has not been registered. 576 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 577 void RemovePragmaHandler(PragmaHandler *Handler) { 578 RemovePragmaHandler(StringRef(), Handler); 579 } 580 581 /// \brief Add the specified comment handler to the preprocessor. 582 void addCommentHandler(CommentHandler *Handler); 583 584 /// \brief Remove the specified comment handler. 585 /// 586 /// It is an error to remove a handler that has not been registered. 587 void removeCommentHandler(CommentHandler *Handler); 588 589 /// \brief Set the code completion handler to the given object. 590 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 591 CodeComplete = &Handler; 592 } 593 594 /// \brief Retrieve the current code-completion handler. 595 CodeCompletionHandler *getCodeCompletionHandler() const { 596 return CodeComplete; 597 } 598 599 /// \brief Clear out the code completion handler. 600 void clearCodeCompletionHandler() { 601 CodeComplete = 0; 602 } 603 604 /// \brief Hook used by the lexer to invoke the "natural language" code 605 /// completion point. 606 void CodeCompleteNaturalLanguage(); 607 608 /// \brief Retrieve the preprocessing record, or NULL if there is no 609 /// preprocessing record. 610 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 611 612 /// \brief Create a new preprocessing record, which will keep track of 613 /// all macro expansions, macro definitions, etc. 614 void createPreprocessingRecord(); 615 616 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 617 /// which implicitly adds the builtin defines etc. 618 void EnterMainSourceFile(); 619 620 /// EndSourceFile - Inform the preprocessor callbacks that processing is 621 /// complete. 622 void EndSourceFile(); 623 624 /// EnterSourceFile - Add a source file to the top of the include stack and 625 /// start lexing tokens from it instead of the current buffer. Emit an error 626 /// and don't enter the file on error. 627 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 628 SourceLocation Loc); 629 630 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 631 /// tokens from it instead of the current buffer. Args specifies the 632 /// tokens input to a function-like macro. 633 /// 634 /// ILEnd specifies the location of the ')' for a function-like macro or the 635 /// identifier for an object-like macro. 636 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 637 MacroArgs *Args); 638 639 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 640 /// which will cause the lexer to start returning the specified tokens. 641 /// 642 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 643 /// not be subject to further macro expansion. Otherwise, these tokens will 644 /// be re-macro-expanded when/if expansion is enabled. 645 /// 646 /// If OwnsTokens is false, this method assumes that the specified stream of 647 /// tokens has a permanent owner somewhere, so they do not need to be copied. 648 /// If it is true, it assumes the array of tokens is allocated with new[] and 649 /// must be freed. 650 /// 651 void EnterTokenStream(const Token *Toks, unsigned NumToks, 652 bool DisableMacroExpansion, bool OwnsTokens); 653 654 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 655 /// lexer stack. This should only be used in situations where the current 656 /// state of the top-of-stack lexer is known. 657 void RemoveTopOfLexerStack(); 658 659 /// EnableBacktrackAtThisPos - From the point that this method is called, and 660 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 661 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 662 /// make the Preprocessor re-lex the same tokens. 663 /// 664 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 665 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 666 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 667 /// 668 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 669 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 670 /// tokens will continue indefinitely. 671 /// 672 void EnableBacktrackAtThisPos(); 673 674 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 675 void CommitBacktrackedTokens(); 676 677 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 678 /// EnableBacktrackAtThisPos() was previously called. 679 void Backtrack(); 680 681 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 682 /// caching of tokens is on. 683 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 684 685 /// Lex - To lex a token from the preprocessor, just pull a token from the 686 /// current lexer or macro object. 687 void Lex(Token &Result) { 688 switch (CurLexerKind) { 689 case CLK_Lexer: CurLexer->Lex(Result); break; 690 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 691 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 692 case CLK_CachingLexer: CachingLex(Result); break; 693 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 694 } 695 } 696 697 void LexAfterModuleImport(Token &Result); 698 699 /// \brief Lex a string literal, which may be the concatenation of multiple 700 /// string literals and may even come from macro expansion. 701 /// \returns true on success, false if a error diagnostic has been generated. 702 bool LexStringLiteral(Token &Result, std::string &String, 703 const char *DiagnosticTag, bool AllowMacroExpansion) { 704 if (AllowMacroExpansion) 705 Lex(Result); 706 else 707 LexUnexpandedToken(Result); 708 return FinishLexStringLiteral(Result, String, DiagnosticTag, 709 AllowMacroExpansion); 710 } 711 712 /// \brief Complete the lexing of a string literal where the first token has 713 /// already been lexed (see LexStringLiteral). 714 bool FinishLexStringLiteral(Token &Result, std::string &String, 715 const char *DiagnosticTag, 716 bool AllowMacroExpansion); 717 718 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 719 /// something not a comment. This is useful in -E -C mode where comments 720 /// would foul up preprocessor directive handling. 721 void LexNonComment(Token &Result) { 722 do 723 Lex(Result); 724 while (Result.getKind() == tok::comment); 725 } 726 727 /// LexUnexpandedToken - This is just like Lex, but this disables macro 728 /// expansion of identifier tokens. 729 void LexUnexpandedToken(Token &Result) { 730 // Disable macro expansion. 731 bool OldVal = DisableMacroExpansion; 732 DisableMacroExpansion = true; 733 // Lex the token. 734 Lex(Result); 735 736 // Reenable it. 737 DisableMacroExpansion = OldVal; 738 } 739 740 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 741 /// expansion of identifier tokens. 742 void LexUnexpandedNonComment(Token &Result) { 743 do 744 LexUnexpandedToken(Result); 745 while (Result.getKind() == tok::comment); 746 } 747 748 /// Disables macro expansion everywhere except for preprocessor directives. 749 void SetMacroExpansionOnlyInDirectives() { 750 DisableMacroExpansion = true; 751 MacroExpansionInDirectivesOverride = true; 752 } 753 754 /// LookAhead - This peeks ahead N tokens and returns that token without 755 /// consuming any tokens. LookAhead(0) returns the next token that would be 756 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 757 /// returns normal tokens after phase 5. As such, it is equivalent to using 758 /// 'Lex', not 'LexUnexpandedToken'. 759 const Token &LookAhead(unsigned N) { 760 if (CachedLexPos + N < CachedTokens.size()) 761 return CachedTokens[CachedLexPos+N]; 762 else 763 return PeekAhead(N+1); 764 } 765 766 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 767 /// this allows to revert a specific number of tokens. 768 /// Note that the number of tokens being reverted should be up to the last 769 /// backtrack position, not more. 770 void RevertCachedTokens(unsigned N) { 771 assert(isBacktrackEnabled() && 772 "Should only be called when tokens are cached for backtracking"); 773 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 774 && "Should revert tokens up to the last backtrack position, not more"); 775 assert(signed(CachedLexPos) - signed(N) >= 0 && 776 "Corrupted backtrack positions ?"); 777 CachedLexPos -= N; 778 } 779 780 /// EnterToken - Enters a token in the token stream to be lexed next. If 781 /// BackTrack() is called afterwards, the token will remain at the insertion 782 /// point. 783 void EnterToken(const Token &Tok) { 784 EnterCachingLexMode(); 785 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 786 } 787 788 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 789 /// tokens (because backtrack is enabled) it should replace the most recent 790 /// cached tokens with the given annotation token. This function has no effect 791 /// if backtracking is not enabled. 792 /// 793 /// Note that the use of this function is just for optimization; so that the 794 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 795 /// invoked. 796 void AnnotateCachedTokens(const Token &Tok) { 797 assert(Tok.isAnnotation() && "Expected annotation token"); 798 if (CachedLexPos != 0 && isBacktrackEnabled()) 799 AnnotatePreviousCachedTokens(Tok); 800 } 801 802 /// \brief Replace the last token with an annotation token. 803 /// 804 /// Like AnnotateCachedTokens(), this routine replaces an 805 /// already-parsed (and resolved) token with an annotation 806 /// token. However, this routine only replaces the last token with 807 /// the annotation token; it does not affect any other cached 808 /// tokens. This function has no effect if backtracking is not 809 /// enabled. 810 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 811 assert(Tok.isAnnotation() && "Expected annotation token"); 812 if (CachedLexPos != 0 && isBacktrackEnabled()) 813 CachedTokens[CachedLexPos-1] = Tok; 814 } 815 816 /// TypoCorrectToken - Update the current token to represent the provided 817 /// identifier, in order to cache an action performed by typo correction. 818 void TypoCorrectToken(const Token &Tok) { 819 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 820 if (CachedLexPos != 0 && isBacktrackEnabled()) 821 CachedTokens[CachedLexPos-1] = Tok; 822 } 823 824 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 825 /// CurTokenLexer pointers. 826 void recomputeCurLexerKind(); 827 828 /// \brief Returns true if incremental processing is enabled 829 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 830 831 /// \brief Enables the incremental processing 832 void enableIncrementalProcessing(bool value = true) { 833 IncrementalProcessing = value; 834 } 835 836 /// \brief Specify the point at which code-completion will be performed. 837 /// 838 /// \param File the file in which code completion should occur. If 839 /// this file is included multiple times, code-completion will 840 /// perform completion the first time it is included. If NULL, this 841 /// function clears out the code-completion point. 842 /// 843 /// \param Line the line at which code completion should occur 844 /// (1-based). 845 /// 846 /// \param Column the column at which code completion should occur 847 /// (1-based). 848 /// 849 /// \returns true if an error occurred, false otherwise. 850 bool SetCodeCompletionPoint(const FileEntry *File, 851 unsigned Line, unsigned Column); 852 853 /// \brief Determine if we are performing code completion. 854 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 855 856 /// \brief Returns the location of the code-completion point. 857 /// Returns an invalid location if code-completion is not enabled or the file 858 /// containing the code-completion point has not been lexed yet. 859 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 860 861 /// \brief Returns the start location of the file of code-completion point. 862 /// Returns an invalid location if code-completion is not enabled or the file 863 /// containing the code-completion point has not been lexed yet. 864 SourceLocation getCodeCompletionFileLoc() const { 865 return CodeCompletionFileLoc; 866 } 867 868 /// \brief Returns true if code-completion is enabled and we have hit the 869 /// code-completion point. 870 bool isCodeCompletionReached() const { return CodeCompletionReached; } 871 872 /// \brief Note that we hit the code-completion point. 873 void setCodeCompletionReached() { 874 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 875 CodeCompletionReached = true; 876 // Silence any diagnostics that occur after we hit the code-completion. 877 getDiagnostics().setSuppressAllDiagnostics(true); 878 } 879 880 /// \brief The location of the currently-active \#pragma clang 881 /// arc_cf_code_audited begin. Returns an invalid location if there 882 /// is no such pragma active. 883 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 884 return PragmaARCCFCodeAuditedLoc; 885 } 886 887 /// \brief Set the location of the currently-active \#pragma clang 888 /// arc_cf_code_audited begin. An invalid location ends the pragma. 889 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 890 PragmaARCCFCodeAuditedLoc = Loc; 891 } 892 893 /// \brief Instruct the preprocessor to skip part of the main source file. 894 /// 895 /// \param Bytes The number of bytes in the preamble to skip. 896 /// 897 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 898 /// start of a line. 899 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 900 SkipMainFilePreamble.first = Bytes; 901 SkipMainFilePreamble.second = StartOfLine; 902 } 903 904 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 905 /// the specified Token's location, translating the token's start 906 /// position in the current buffer into a SourcePosition object for rendering. 907 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 908 return Diags->Report(Loc, DiagID); 909 } 910 911 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 912 return Diags->Report(Tok.getLocation(), DiagID); 913 } 914 915 /// getSpelling() - Return the 'spelling' of the token at the given 916 /// location; does not go up to the spelling location or down to the 917 /// expansion location. 918 /// 919 /// \param buffer A buffer which will be used only if the token requires 920 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 921 /// \param invalid If non-null, will be set \c true if an error occurs. 922 StringRef getSpelling(SourceLocation loc, 923 SmallVectorImpl<char> &buffer, 924 bool *invalid = 0) const { 925 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 926 } 927 928 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 929 /// token is the characters used to represent the token in the source file 930 /// after trigraph expansion and escaped-newline folding. In particular, this 931 /// wants to get the true, uncanonicalized, spelling of things like digraphs 932 /// UCNs, etc. 933 /// 934 /// \param Invalid If non-null, will be set \c true if an error occurs. 935 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 936 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 937 } 938 939 /// getSpelling - This method is used to get the spelling of a token into a 940 /// preallocated buffer, instead of as an std::string. The caller is required 941 /// to allocate enough space for the token, which is guaranteed to be at least 942 /// Tok.getLength() bytes long. The length of the actual result is returned. 943 /// 944 /// Note that this method may do two possible things: it may either fill in 945 /// the buffer specified with characters, or it may *change the input pointer* 946 /// to point to a constant buffer with the data already in it (avoiding a 947 /// copy). The caller is not allowed to modify the returned buffer pointer 948 /// if an internal buffer is returned. 949 unsigned getSpelling(const Token &Tok, const char *&Buffer, 950 bool *Invalid = 0) const { 951 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 952 } 953 954 /// getSpelling - This method is used to get the spelling of a token into a 955 /// SmallVector. Note that the returned StringRef may not point to the 956 /// supplied buffer if a copy can be avoided. 957 StringRef getSpelling(const Token &Tok, 958 SmallVectorImpl<char> &Buffer, 959 bool *Invalid = 0) const; 960 961 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 962 /// with length 1, return the character. 963 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 964 bool *Invalid = 0) const { 965 assert(Tok.is(tok::numeric_constant) && 966 Tok.getLength() == 1 && "Called on unsupported token"); 967 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 968 969 // If the token is carrying a literal data pointer, just use it. 970 if (const char *D = Tok.getLiteralData()) 971 return *D; 972 973 // Otherwise, fall back on getCharacterData, which is slower, but always 974 // works. 975 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 976 } 977 978 /// \brief Retrieve the name of the immediate macro expansion. 979 /// 980 /// This routine starts from a source location, and finds the name of the macro 981 /// responsible for its immediate expansion. It looks through any intervening 982 /// macro argument expansions to compute this. It returns a StringRef which 983 /// refers to the SourceManager-owned buffer of the source where that macro 984 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 985 StringRef getImmediateMacroName(SourceLocation Loc) { 986 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 987 } 988 989 /// CreateString - Plop the specified string into a scratch buffer and set the 990 /// specified token's location and length to it. If specified, the source 991 /// location provides a location of the expansion point of the token. 992 void CreateString(StringRef Str, Token &Tok, 993 SourceLocation ExpansionLocStart = SourceLocation(), 994 SourceLocation ExpansionLocEnd = SourceLocation()); 995 996 /// \brief Computes the source location just past the end of the 997 /// token at this source location. 998 /// 999 /// This routine can be used to produce a source location that 1000 /// points just past the end of the token referenced by \p Loc, and 1001 /// is generally used when a diagnostic needs to point just after a 1002 /// token where it expected something different that it received. If 1003 /// the returned source location would not be meaningful (e.g., if 1004 /// it points into a macro), this routine returns an invalid 1005 /// source location. 1006 /// 1007 /// \param Offset an offset from the end of the token, where the source 1008 /// location should refer to. The default offset (0) produces a source 1009 /// location pointing just past the end of the token; an offset of 1 produces 1010 /// a source location pointing to the last character in the token, etc. 1011 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 1012 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 1013 } 1014 1015 /// \brief Returns true if the given MacroID location points at the first 1016 /// token of the macro expansion. 1017 /// 1018 /// \param MacroBegin If non-null and function returns true, it is set to 1019 /// begin location of the macro. 1020 bool isAtStartOfMacroExpansion(SourceLocation loc, 1021 SourceLocation *MacroBegin = 0) const { 1022 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 1023 MacroBegin); 1024 } 1025 1026 /// \brief Returns true if the given MacroID location points at the last 1027 /// token of the macro expansion. 1028 /// 1029 /// \param MacroEnd If non-null and function returns true, it is set to 1030 /// end location of the macro. 1031 bool isAtEndOfMacroExpansion(SourceLocation loc, 1032 SourceLocation *MacroEnd = 0) const { 1033 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 1034 } 1035 1036 /// DumpToken - Print the token to stderr, used for debugging. 1037 /// 1038 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 1039 void DumpLocation(SourceLocation Loc) const; 1040 void DumpMacro(const MacroInfo &MI) const; 1041 1042 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 1043 /// token, return a new location that specifies a character within the token. 1044 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 1045 unsigned Char) const { 1046 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 1047 } 1048 1049 /// IncrementPasteCounter - Increment the counters for the number of token 1050 /// paste operations performed. If fast was specified, this is a 'fast paste' 1051 /// case we handled. 1052 /// 1053 void IncrementPasteCounter(bool isFast) { 1054 if (isFast) 1055 ++NumFastTokenPaste; 1056 else 1057 ++NumTokenPaste; 1058 } 1059 1060 void PrintStats(); 1061 1062 size_t getTotalMemory() const; 1063 1064 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 1065 /// comment (/##/) in microsoft mode, this method handles updating the current 1066 /// state, returning the token on the next source line. 1067 void HandleMicrosoftCommentPaste(Token &Tok); 1068 1069 //===--------------------------------------------------------------------===// 1070 // Preprocessor callback methods. These are invoked by a lexer as various 1071 // directives and events are found. 1072 1073 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1074 /// identifier information for the token and install it into the token, 1075 /// updating the token kind accordingly. 1076 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1077 1078private: 1079 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1080 1081public: 1082 1083 // SetPoisonReason - Call this function to indicate the reason for 1084 // poisoning an identifier. If that identifier is accessed while 1085 // poisoned, then this reason will be used instead of the default 1086 // "poisoned" diagnostic. 1087 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1088 1089 // HandlePoisonedIdentifier - Display reason for poisoned 1090 // identifier. 1091 void HandlePoisonedIdentifier(Token & Tok); 1092 1093 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1094 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1095 if(II->isPoisoned()) { 1096 HandlePoisonedIdentifier(Identifier); 1097 } 1098 } 1099 } 1100 1101private: 1102 /// Identifiers used for SEH handling in Borland. These are only 1103 /// allowed in particular circumstances 1104 // __except block 1105 IdentifierInfo *Ident__exception_code, 1106 *Ident___exception_code, 1107 *Ident_GetExceptionCode; 1108 // __except filter expression 1109 IdentifierInfo *Ident__exception_info, 1110 *Ident___exception_info, 1111 *Ident_GetExceptionInfo; 1112 // __finally 1113 IdentifierInfo *Ident__abnormal_termination, 1114 *Ident___abnormal_termination, 1115 *Ident_AbnormalTermination; 1116public: 1117 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1118 1119 /// HandleIdentifier - This callback is invoked when the lexer reads an 1120 /// identifier and has filled in the tokens IdentifierInfo member. This 1121 /// callback potentially macro expands it or turns it into a named token (like 1122 /// 'for'). 1123 void HandleIdentifier(Token &Identifier); 1124 1125 1126 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1127 /// the current file. This either returns the EOF token and returns true, or 1128 /// pops a level off the include stack and returns false, at which point the 1129 /// client should call lex again. 1130 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1131 1132 /// HandleEndOfTokenLexer - This callback is invoked when the current 1133 /// TokenLexer hits the end of its token stream. 1134 bool HandleEndOfTokenLexer(Token &Result); 1135 1136 /// HandleDirective - This callback is invoked when the lexer sees a # token 1137 /// at the start of a line. This consumes the directive, modifies the 1138 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1139 /// read is the correct one. 1140 void HandleDirective(Token &Result); 1141 1142 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1143 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1144 /// true, then we consider macros that expand to zero tokens as being ok. 1145 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1146 1147 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1148 /// current line until the tok::eod token is found. 1149 void DiscardUntilEndOfDirective(); 1150 1151 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1152 /// __DATE__ or __TIME__ in the file so far. 1153 bool SawDateOrTime() const { 1154 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1155 } 1156 unsigned getCounterValue() const { return CounterValue; } 1157 void setCounterValue(unsigned V) { CounterValue = V; } 1158 1159 /// \brief Retrieves the module that we're currently building, if any. 1160 Module *getCurrentModule(); 1161 1162 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1163 MacroInfo *AllocateMacroInfo(SourceLocation L); 1164 1165 /// \brief Allocate a new MacroInfo object which is clone of \p MI. 1166 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1167 1168 /// \brief Turn the specified lexer token into a fully checked and spelled 1169 /// filename, e.g. as an operand of \#include. 1170 /// 1171 /// The caller is expected to provide a buffer that is large enough to hold 1172 /// the spelling of the filename, but is also expected to handle the case 1173 /// when this method decides to use a different buffer. 1174 /// 1175 /// \returns true if the input filename was in <>'s or false if it was 1176 /// in ""'s. 1177 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1178 1179 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1180 /// 1181 /// Returns null on failure. \p isAngled indicates whether the file 1182 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1183 const FileEntry *LookupFile(StringRef Filename, 1184 bool isAngled, const DirectoryLookup *FromDir, 1185 const DirectoryLookup *&CurDir, 1186 SmallVectorImpl<char> *SearchPath, 1187 SmallVectorImpl<char> *RelativePath, 1188 Module **SuggestedModule, 1189 bool SkipCache = false); 1190 1191 /// GetCurLookup - The DirectoryLookup structure used to find the current 1192 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1193 /// implement \#include_next and find directory-specific properties. 1194 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1195 1196 /// \brief Return true if we're in the top-level file, not in a \#include. 1197 bool isInPrimaryFile() const; 1198 1199 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1200 /// from a macro as multiple tokens, which need to be glued together. This 1201 /// occurs for code like: 1202 /// \code 1203 /// \#define FOO <x/y.h> 1204 /// \#include FOO 1205 /// \endcode 1206 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1207 /// 1208 /// This code concatenates and consumes tokens up to the '>' token. It 1209 /// returns false if the > was found, otherwise it returns true if it finds 1210 /// and consumes the EOD marker. 1211 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1212 SourceLocation &End); 1213 1214 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1215 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1216 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1217 1218private: 1219 1220 void PushIncludeMacroStack() { 1221 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1222 CurLexer.take(), 1223 CurPTHLexer.take(), 1224 CurPPLexer, 1225 CurTokenLexer.take(), 1226 CurDirLookup)); 1227 CurPPLexer = 0; 1228 } 1229 1230 void PopIncludeMacroStack() { 1231 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1232 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1233 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1234 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1235 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1236 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1237 IncludeMacroStack.pop_back(); 1238 } 1239 1240 /// \brief Allocate a new MacroInfo object. 1241 MacroInfo *AllocateMacroInfo(); 1242 1243 /// \brief Release the specified MacroInfo for re-use. 1244 /// 1245 /// This memory will be reused for allocating new MacroInfo objects. 1246 void ReleaseMacroInfo(MacroInfo* MI); 1247 1248 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1249 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1250 /// and discards the rest of the macro line if the macro name is invalid. 1251 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1252 1253 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1254 /// definition has just been read. Lex the rest of the arguments and the 1255 /// closing ), updating MI with what we learn and saving in LastTok the 1256 /// last token read. 1257 /// Return true if an error occurs parsing the arg list. 1258 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1259 1260 /// We just read a \#if or related directive and decided that the 1261 /// subsequent tokens are in the \#if'd out portion of the 1262 /// file. Lex the rest of the file, until we see an \#endif. If \p 1263 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1264 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1265 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1266 /// already seen one so a \#else directive is a duplicate. When this returns, 1267 /// the caller can lex the first valid token. 1268 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1269 bool FoundNonSkipPortion, bool FoundElse, 1270 SourceLocation ElseLoc = SourceLocation()); 1271 1272 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1273 void PTHSkipExcludedConditionalBlock(); 1274 1275 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1276 /// may occur after a #if or #elif directive and return it as a bool. If the 1277 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1278 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1279 1280 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1281 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1282 void RegisterBuiltinPragmas(); 1283 1284 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1285 void RegisterBuiltinMacros(); 1286 1287 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1288 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1289 /// the macro should not be expanded return true, otherwise return false. 1290 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1291 1292 /// \brief Cache macro expanded tokens for TokenLexers. 1293 // 1294 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1295 /// going to lex in the cache and when it finishes the tokens are removed 1296 /// from the end of the cache. 1297 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1298 ArrayRef<Token> tokens); 1299 void removeCachedMacroExpandedTokensOfLastLexer(); 1300 friend void TokenLexer::ExpandFunctionArguments(); 1301 1302 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1303 /// lexed is a '('. If so, consume the token and return true, if not, this 1304 /// method should have no observable side-effect on the lexed tokens. 1305 bool isNextPPTokenLParen(); 1306 1307 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1308 /// invoked to read all of the formal arguments specified for the macro 1309 /// invocation. This returns null on error. 1310 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1311 SourceLocation &ExpansionEnd); 1312 1313 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1314 /// as a builtin macro, handle it and return the next token as 'Tok'. 1315 void ExpandBuiltinMacro(Token &Tok); 1316 1317 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1318 /// return the first token after the directive. The _Pragma token has just 1319 /// been read into 'Tok'. 1320 void Handle_Pragma(Token &Tok); 1321 1322 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1323 /// is not enclosed within a string literal. 1324 void HandleMicrosoft__pragma(Token &Tok); 1325 1326 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1327 /// start lexing tokens from it instead of the current buffer. 1328 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1329 1330 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1331 /// start getting tokens from it using the PTH cache. 1332 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1333 1334 /// IsFileLexer - Returns true if we are lexing from a file and not a 1335 /// pragma or a macro. 1336 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1337 return L ? !L->isPragmaLexer() : P != 0; 1338 } 1339 1340 static bool IsFileLexer(const IncludeStackInfo& I) { 1341 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1342 } 1343 1344 bool IsFileLexer() const { 1345 return IsFileLexer(CurLexer.get(), CurPPLexer); 1346 } 1347 1348 //===--------------------------------------------------------------------===// 1349 // Caching stuff. 1350 void CachingLex(Token &Result); 1351 bool InCachingLexMode() const { 1352 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1353 // that we are past EOF, not that we are in CachingLex mode. 1354 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1355 !IncludeMacroStack.empty(); 1356 } 1357 void EnterCachingLexMode(); 1358 void ExitCachingLexMode() { 1359 if (InCachingLexMode()) 1360 RemoveTopOfLexerStack(); 1361 } 1362 const Token &PeekAhead(unsigned N); 1363 void AnnotatePreviousCachedTokens(const Token &Tok); 1364 1365 //===--------------------------------------------------------------------===// 1366 /// Handle*Directive - implement the various preprocessor directives. These 1367 /// should side-effect the current preprocessor object so that the next call 1368 /// to Lex() will return the appropriate token next. 1369 void HandleLineDirective(Token &Tok); 1370 void HandleDigitDirective(Token &Tok); 1371 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1372 void HandleIdentSCCSDirective(Token &Tok); 1373 void HandleMacroPublicDirective(Token &Tok); 1374 void HandleMacroPrivateDirective(Token &Tok); 1375 1376 // File inclusion. 1377 void HandleIncludeDirective(SourceLocation HashLoc, 1378 Token &Tok, 1379 const DirectoryLookup *LookupFrom = 0, 1380 bool isImport = false); 1381 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1382 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1383 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1384 void HandleMicrosoftImportDirective(Token &Tok); 1385 1386 // Macro handling. 1387 void HandleDefineDirective(Token &Tok); 1388 void HandleUndefDirective(Token &Tok); 1389 void UndefineMacro(IdentifierInfo *II, MacroInfo *MI, 1390 SourceLocation UndefLoc); 1391 1392 // Conditional Inclusion. 1393 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1394 bool ReadAnyTokensBeforeDirective); 1395 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1396 void HandleEndifDirective(Token &Tok); 1397 void HandleElseDirective(Token &Tok); 1398 void HandleElifDirective(Token &Tok); 1399 1400 // Pragmas. 1401 void HandlePragmaDirective(unsigned Introducer); 1402public: 1403 void HandlePragmaOnce(Token &OnceTok); 1404 void HandlePragmaMark(); 1405 void HandlePragmaPoison(Token &PoisonTok); 1406 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1407 void HandlePragmaDependency(Token &DependencyTok); 1408 void HandlePragmaComment(Token &CommentTok); 1409 void HandlePragmaMessage(Token &MessageTok); 1410 void HandlePragmaPushMacro(Token &Tok); 1411 void HandlePragmaPopMacro(Token &Tok); 1412 void HandlePragmaIncludeAlias(Token &Tok); 1413 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1414 1415 // Return true and store the first token only if any CommentHandler 1416 // has inserted some tokens and getCommentRetentionState() is false. 1417 bool HandleComment(Token &Token, SourceRange Comment); 1418 1419 /// \brief A macro is used, update information about macros that need unused 1420 /// warnings. 1421 void markMacroAsUsed(MacroInfo *MI); 1422}; 1423 1424/// \brief Abstract base class that describes a handler that will receive 1425/// source ranges for each of the comments encountered in the source file. 1426class CommentHandler { 1427public: 1428 virtual ~CommentHandler(); 1429 1430 // The handler shall return true if it has pushed any tokens 1431 // to be read using e.g. EnterToken or EnterTokenStream. 1432 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1433}; 1434 1435} // end namespace clang 1436 1437#endif 1438