Preprocessor.h revision f953276b6165dc7e8f4679cce4d0e7c649cd4232
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 currently preprocessing a #if or #elif directive 164 bool ParsingIfOrElifDirective; 165 166 /// \brief True if we are pre-expanding macro arguments. 167 bool InMacroArgPreExpansion; 168 169 /// Identifiers - This is mapping/lookup information for all identifiers in 170 /// the program, including program keywords. 171 mutable IdentifierTable Identifiers; 172 173 /// Selectors - This table contains all the selectors in the program. Unlike 174 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 175 /// It is declared/expanded here because it's role/lifetime is 176 /// conceptually similar the IdentifierTable. In addition, the current control 177 /// flow (in clang::ParseAST()), make it convenient to put here. 178 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 179 /// the lifetime of the preprocessor. 180 SelectorTable Selectors; 181 182 /// BuiltinInfo - Information about builtins. 183 Builtin::Context BuiltinInfo; 184 185 /// PragmaHandlers - This tracks all of the pragmas that the client registered 186 /// with this preprocessor. 187 PragmaNamespace *PragmaHandlers; 188 189 /// \brief Tracks all of the comment handlers that the client registered 190 /// with this preprocessor. 191 std::vector<CommentHandler *> CommentHandlers; 192 193 /// \brief True if we want to ignore EOF token and continue later on (thus 194 /// avoid tearing the Lexer and etc. down). 195 bool IncrementalProcessing; 196 197 /// \brief The code-completion handler. 198 CodeCompletionHandler *CodeComplete; 199 200 /// \brief The file that we're performing code-completion for, if any. 201 const FileEntry *CodeCompletionFile; 202 203 /// \brief The offset in file for the code-completion point. 204 unsigned CodeCompletionOffset; 205 206 /// \brief The location for the code-completion point. This gets instantiated 207 /// when the CodeCompletionFile gets \#include'ed for preprocessing. 208 SourceLocation CodeCompletionLoc; 209 210 /// \brief The start location for the file of the code-completion point. 211 /// 212 /// This gets instantiated when the CodeCompletionFile gets \#include'ed 213 /// for preprocessing. 214 SourceLocation CodeCompletionFileLoc; 215 216 /// \brief The source location of the 'import' contextual keyword we just 217 /// lexed, if any. 218 SourceLocation ModuleImportLoc; 219 220 /// \brief The module import path that we're currently processing. 221 SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> ModuleImportPath; 222 223 /// \brief Whether the module import expectes an identifier next. Otherwise, 224 /// it expects a '.' or ';'. 225 bool ModuleImportExpectsIdentifier; 226 227 /// \brief The source location of the currently-active 228 /// #pragma clang arc_cf_code_audited begin. 229 SourceLocation PragmaARCCFCodeAuditedLoc; 230 231 /// \brief True if we hit the code-completion point. 232 bool CodeCompletionReached; 233 234 /// \brief The number of bytes that we will initially skip when entering the 235 /// main file, which is used when loading a precompiled preamble, along 236 /// with a flag that indicates whether skipping this number of bytes will 237 /// place the lexer at the start of a line. 238 std::pair<unsigned, bool> SkipMainFilePreamble; 239 240 /// CurLexer - This is the current top of the stack that we're lexing from if 241 /// not expanding a macro and we are lexing directly from source code. 242 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 243 OwningPtr<Lexer> CurLexer; 244 245 /// CurPTHLexer - This is the current top of stack that we're lexing from if 246 /// not expanding from a macro and we are lexing from a PTH cache. 247 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 248 OwningPtr<PTHLexer> CurPTHLexer; 249 250 /// CurPPLexer - This is the current top of the stack what we're lexing from 251 /// if not expanding a macro. This is an alias for either CurLexer or 252 /// CurPTHLexer. 253 PreprocessorLexer *CurPPLexer; 254 255 /// CurLookup - The DirectoryLookup structure used to find the current 256 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 257 /// implement \#include_next and find directory-specific properties. 258 const DirectoryLookup *CurDirLookup; 259 260 /// CurTokenLexer - This is the current macro we are expanding, if we are 261 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 262 OwningPtr<TokenLexer> CurTokenLexer; 263 264 /// \brief The kind of lexer we're currently working with. 265 enum CurLexerKind { 266 CLK_Lexer, 267 CLK_PTHLexer, 268 CLK_TokenLexer, 269 CLK_CachingLexer, 270 CLK_LexAfterModuleImport 271 } CurLexerKind; 272 273 /// IncludeMacroStack - This keeps track of the stack of files currently 274 /// \#included, and macros currently being expanded from, not counting 275 /// CurLexer/CurTokenLexer. 276 struct IncludeStackInfo { 277 enum CurLexerKind CurLexerKind; 278 Lexer *TheLexer; 279 PTHLexer *ThePTHLexer; 280 PreprocessorLexer *ThePPLexer; 281 TokenLexer *TheTokenLexer; 282 const DirectoryLookup *TheDirLookup; 283 284 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 285 PreprocessorLexer* PPL, 286 TokenLexer* TL, const DirectoryLookup *D) 287 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 288 TheTokenLexer(TL), TheDirLookup(D) {} 289 }; 290 std::vector<IncludeStackInfo> IncludeMacroStack; 291 292 /// Callbacks - These are actions invoked when some preprocessor activity is 293 /// encountered (e.g. a file is \#included, etc). 294 PPCallbacks *Callbacks; 295 296 /// \brief Listener whose actions are invoked when an entity in the 297 /// preprocessor (e.g., a macro) that was loaded from an AST file is 298 /// later mutated. 299 PPMutationListener *Listener; 300 301 struct MacroExpandsInfo { 302 Token Tok; 303 MacroInfo *MI; 304 SourceRange Range; 305 MacroExpandsInfo(Token Tok, MacroInfo *MI, SourceRange Range) 306 : Tok(Tok), MI(MI), Range(Range) { } 307 }; 308 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 309 310 /// Macros - For each IdentifierInfo that was associated with a macro, we 311 /// keep a mapping to the history of all macro definitions and #undefs in 312 /// the reverse order (the latest one is in the head of the list). 313 llvm::DenseMap<const IdentifierInfo*, MacroInfo*> Macros; 314 friend class ASTReader; 315 316 /// \brief Macros that we want to warn because they are not used at the end 317 /// of the translation unit; we store just their SourceLocations instead 318 /// something like MacroInfo*. The benefit of this is that when we are 319 /// deserializing from PCH, we don't need to deserialize identifier & macros 320 /// just so that we can report that they are unused, we just warn using 321 /// the SourceLocations of this set (that will be filled by the ASTReader). 322 /// We are using SmallPtrSet instead of a vector for faster removal. 323 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 324 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 325 326 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 327 /// reused for quick allocation. 328 MacroArgs *MacroArgCache; 329 friend class MacroArgs; 330 331 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 332 /// push_macro directive, we keep a MacroInfo stack used to restore 333 /// previous macro value. 334 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 335 336 // Various statistics we track for performance analysis. 337 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 338 unsigned NumIf, NumElse, NumEndif; 339 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 340 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 341 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 342 unsigned NumSkipped; 343 344 /// Predefines - This string is the predefined macros that preprocessor 345 /// should use from the command line etc. 346 std::string Predefines; 347 348 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 349 enum { TokenLexerCacheSize = 8 }; 350 unsigned NumCachedTokenLexers; 351 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 352 353 /// \brief Keeps macro expanded tokens for TokenLexers. 354 // 355 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 356 /// going to lex in the cache and when it finishes the tokens are removed 357 /// from the end of the cache. 358 SmallVector<Token, 16> MacroExpandedTokens; 359 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 360 361 /// \brief A record of the macro definitions and expansions that 362 /// occurred during preprocessing. 363 /// 364 /// This is an optional side structure that can be enabled with 365 /// \c createPreprocessingRecord() prior to preprocessing. 366 PreprocessingRecord *Record; 367 368private: // Cached tokens state. 369 typedef SmallVector<Token, 1> CachedTokensTy; 370 371 /// CachedTokens - Cached tokens are stored here when we do backtracking or 372 /// lookahead. They are "lexed" by the CachingLex() method. 373 CachedTokensTy CachedTokens; 374 375 /// CachedLexPos - The position of the cached token that CachingLex() should 376 /// "lex" next. If it points beyond the CachedTokens vector, it means that 377 /// a normal Lex() should be invoked. 378 CachedTokensTy::size_type CachedLexPos; 379 380 /// BacktrackPositions - Stack of backtrack positions, allowing nested 381 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 382 /// indicate where CachedLexPos should be set when the BackTrack() method is 383 /// invoked (at which point the last position is popped). 384 std::vector<CachedTokensTy::size_type> BacktrackPositions; 385 386 struct MacroInfoChain { 387 MacroInfo MI; 388 MacroInfoChain *Next; 389 MacroInfoChain *Prev; 390 }; 391 392 /// MacroInfos are managed as a chain for easy disposal. This is the head 393 /// of that list. 394 MacroInfoChain *MIChainHead; 395 396 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 397 /// allocation. 398 MacroInfoChain *MICache; 399 400public: 401 Preprocessor(IntrusiveRefCntPtr<PreprocessorOptions> PPOpts, 402 DiagnosticsEngine &diags, LangOptions &opts, 403 const TargetInfo *target, 404 SourceManager &SM, HeaderSearch &Headers, 405 ModuleLoader &TheModuleLoader, 406 IdentifierInfoLookup *IILookup = 0, 407 bool OwnsHeaderSearch = false, 408 bool DelayInitialization = false, 409 bool IncrProcessing = false); 410 411 ~Preprocessor(); 412 413 /// \brief Initialize the preprocessor, if the constructor did not already 414 /// perform the initialization. 415 /// 416 /// \param Target Information about the target. 417 void Initialize(const TargetInfo &Target); 418 419 /// \brief Retrieve the preprocessor options used to initialize this 420 /// preprocessor. 421 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; } 422 423 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 424 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 425 426 const LangOptions &getLangOpts() const { return LangOpts; } 427 const TargetInfo &getTargetInfo() const { return *Target; } 428 FileManager &getFileManager() const { return FileMgr; } 429 SourceManager &getSourceManager() const { return SourceMgr; } 430 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 431 432 IdentifierTable &getIdentifierTable() { return Identifiers; } 433 SelectorTable &getSelectorTable() { return Selectors; } 434 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 435 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 436 437 void setPTHManager(PTHManager* pm); 438 439 PTHManager *getPTHManager() { return PTH.get(); } 440 441 void setExternalSource(ExternalPreprocessorSource *Source) { 442 ExternalSource = Source; 443 } 444 445 ExternalPreprocessorSource *getExternalSource() const { 446 return ExternalSource; 447 } 448 449 /// \brief Retrieve the module loader associated with this preprocessor. 450 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 451 452 /// \brief True if we are currently preprocessing a #if or #elif directive 453 bool isParsingIfOrElifDirective() const { 454 return ParsingIfOrElifDirective; 455 } 456 457 /// SetCommentRetentionState - Control whether or not the preprocessor retains 458 /// comments in output. 459 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 460 this->KeepComments = KeepComments | KeepMacroComments; 461 this->KeepMacroComments = KeepMacroComments; 462 } 463 464 bool getCommentRetentionState() const { return KeepComments; } 465 466 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 467 bool getPragmasEnabled() const { return PragmasEnabled; } 468 469 void SetSuppressIncludeNotFoundError(bool Suppress) { 470 SuppressIncludeNotFoundError = Suppress; 471 } 472 473 bool GetSuppressIncludeNotFoundError() { 474 return SuppressIncludeNotFoundError; 475 } 476 477 /// isCurrentLexer - Return true if we are lexing directly from the specified 478 /// lexer. 479 bool isCurrentLexer(const PreprocessorLexer *L) const { 480 return CurPPLexer == L; 481 } 482 483 /// getCurrentLexer - Return the current lexer being lexed from. Note 484 /// that this ignores any potentially active macro expansions and _Pragma 485 /// expansions going on at the time. 486 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 487 488 /// getCurrentFileLexer - Return the current file lexer being lexed from. 489 /// Note that this ignores any potentially active macro expansions and _Pragma 490 /// expansions going on at the time. 491 PreprocessorLexer *getCurrentFileLexer() const; 492 493 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 494 /// Note that this class takes ownership of any PPCallbacks object given to 495 /// it. 496 PPCallbacks *getPPCallbacks() const { return Callbacks; } 497 void addPPCallbacks(PPCallbacks *C) { 498 if (Callbacks) 499 C = new PPChainedCallbacks(C, Callbacks); 500 Callbacks = C; 501 } 502 503 /// \brief Attach an preprocessor mutation listener to the preprocessor. 504 /// 505 /// The preprocessor mutation listener provides the ability to track 506 /// modifications to the preprocessor entities committed after they were 507 /// initially created. 508 void setPPMutationListener(PPMutationListener *Listener) { 509 this->Listener = Listener; 510 } 511 512 /// \brief Retrieve a pointer to the preprocessor mutation listener 513 /// associated with this preprocessor, if any. 514 PPMutationListener *getPPMutationListener() const { return Listener; } 515 516 /// \brief Given an identifier, return the MacroInfo it is \#defined to 517 /// or null if it isn't \#define'd. 518 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 519 if (!II->hasMacroDefinition()) 520 return 0; 521 522 MacroInfo *MI = getMacroInfoHistory(II); 523 assert(MI->getUndefLoc().isInvalid() && "Macro is undefined!"); 524 return MI; 525 } 526 527 /// \brief Given an identifier, return the (probably #undef'd) MacroInfo 528 /// representing the most recent macro definition. One can iterate over all 529 /// previous macro definitions from it. This method should only be called for 530 /// identifiers that hadMacroDefinition(). 531 MacroInfo *getMacroInfoHistory(const IdentifierInfo *II) const; 532 533 /// \brief Specify a macro for this identifier. 534 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 535 /// \brief Add a MacroInfo that was loaded from an AST file. 536 void addLoadedMacroInfo(IdentifierInfo *II, MacroInfo *MI); 537 /// \brief Make the given MacroInfo, that was loaded from an AST file and 538 /// previously hidden, visible. 539 void makeLoadedMacroInfoVisible(IdentifierInfo *II, MacroInfo *MI); 540 /// \brief Undefine a macro for this identifier. 541 void clearMacroInfo(IdentifierInfo *II); 542 543 /// macro_iterator/macro_begin/macro_end - This allows you to walk the macro 544 /// history table. Currently defined macros have 545 /// IdentifierInfo::hasMacroDefinition() set and an empty 546 /// MacroInfo::getUndefLoc() at the head of the list. 547 typedef llvm::DenseMap<const IdentifierInfo *, 548 MacroInfo*>::const_iterator macro_iterator; 549 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 550 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 551 552 /// \brief Return the name of the macro defined before \p Loc that has 553 /// spelling \p Tokens. If there are multiple macros with same spelling, 554 /// return the last one defined. 555 StringRef getLastMacroWithSpelling(SourceLocation Loc, 556 ArrayRef<TokenValue> Tokens) const; 557 558 const std::string &getPredefines() const { return Predefines; } 559 /// setPredefines - Set the predefines for this Preprocessor. These 560 /// predefines are automatically injected when parsing the main file. 561 void setPredefines(const char *P) { Predefines = P; } 562 void setPredefines(const std::string &P) { Predefines = P; } 563 564 /// Return information about the specified preprocessor 565 /// identifier token. 566 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 567 return &Identifiers.get(Name); 568 } 569 570 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 571 /// If 'Namespace' is non-null, then it is a token required to exist on the 572 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 573 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 574 void AddPragmaHandler(PragmaHandler *Handler) { 575 AddPragmaHandler(StringRef(), Handler); 576 } 577 578 /// RemovePragmaHandler - Remove the specific pragma handler from 579 /// the preprocessor. If \p Namespace is non-null, then it should 580 /// be the namespace that \p Handler was added to. It is an error 581 /// to remove a handler that has not been registered. 582 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 583 void RemovePragmaHandler(PragmaHandler *Handler) { 584 RemovePragmaHandler(StringRef(), Handler); 585 } 586 587 /// \brief Add the specified comment handler to the preprocessor. 588 void addCommentHandler(CommentHandler *Handler); 589 590 /// \brief Remove the specified comment handler. 591 /// 592 /// It is an error to remove a handler that has not been registered. 593 void removeCommentHandler(CommentHandler *Handler); 594 595 /// \brief Set the code completion handler to the given object. 596 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 597 CodeComplete = &Handler; 598 } 599 600 /// \brief Retrieve the current code-completion handler. 601 CodeCompletionHandler *getCodeCompletionHandler() const { 602 return CodeComplete; 603 } 604 605 /// \brief Clear out the code completion handler. 606 void clearCodeCompletionHandler() { 607 CodeComplete = 0; 608 } 609 610 /// \brief Hook used by the lexer to invoke the "natural language" code 611 /// completion point. 612 void CodeCompleteNaturalLanguage(); 613 614 /// \brief Retrieve the preprocessing record, or NULL if there is no 615 /// preprocessing record. 616 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 617 618 /// \brief Create a new preprocessing record, which will keep track of 619 /// all macro expansions, macro definitions, etc. 620 void createPreprocessingRecord(); 621 622 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 623 /// which implicitly adds the builtin defines etc. 624 void EnterMainSourceFile(); 625 626 /// EndSourceFile - Inform the preprocessor callbacks that processing is 627 /// complete. 628 void EndSourceFile(); 629 630 /// EnterSourceFile - Add a source file to the top of the include stack and 631 /// start lexing tokens from it instead of the current buffer. Emit an error 632 /// and don't enter the file on error. 633 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 634 SourceLocation Loc); 635 636 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 637 /// tokens from it instead of the current buffer. Args specifies the 638 /// tokens input to a function-like macro. 639 /// 640 /// ILEnd specifies the location of the ')' for a function-like macro or the 641 /// identifier for an object-like macro. 642 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 643 MacroArgs *Args); 644 645 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 646 /// which will cause the lexer to start returning the specified tokens. 647 /// 648 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 649 /// not be subject to further macro expansion. Otherwise, these tokens will 650 /// be re-macro-expanded when/if expansion is enabled. 651 /// 652 /// If OwnsTokens is false, this method assumes that the specified stream of 653 /// tokens has a permanent owner somewhere, so they do not need to be copied. 654 /// If it is true, it assumes the array of tokens is allocated with new[] and 655 /// must be freed. 656 /// 657 void EnterTokenStream(const Token *Toks, unsigned NumToks, 658 bool DisableMacroExpansion, bool OwnsTokens); 659 660 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 661 /// lexer stack. This should only be used in situations where the current 662 /// state of the top-of-stack lexer is known. 663 void RemoveTopOfLexerStack(); 664 665 /// EnableBacktrackAtThisPos - From the point that this method is called, and 666 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 667 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 668 /// make the Preprocessor re-lex the same tokens. 669 /// 670 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 671 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 672 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 673 /// 674 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 675 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 676 /// tokens will continue indefinitely. 677 /// 678 void EnableBacktrackAtThisPos(); 679 680 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 681 void CommitBacktrackedTokens(); 682 683 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 684 /// EnableBacktrackAtThisPos() was previously called. 685 void Backtrack(); 686 687 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 688 /// caching of tokens is on. 689 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 690 691 /// Lex - To lex a token from the preprocessor, just pull a token from the 692 /// current lexer or macro object. 693 void Lex(Token &Result) { 694 switch (CurLexerKind) { 695 case CLK_Lexer: CurLexer->Lex(Result); break; 696 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 697 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 698 case CLK_CachingLexer: CachingLex(Result); break; 699 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 700 } 701 } 702 703 void LexAfterModuleImport(Token &Result); 704 705 /// \brief Lex a string literal, which may be the concatenation of multiple 706 /// string literals and may even come from macro expansion. 707 /// \returns true on success, false if a error diagnostic has been generated. 708 bool LexStringLiteral(Token &Result, std::string &String, 709 const char *DiagnosticTag, bool AllowMacroExpansion) { 710 if (AllowMacroExpansion) 711 Lex(Result); 712 else 713 LexUnexpandedToken(Result); 714 return FinishLexStringLiteral(Result, String, DiagnosticTag, 715 AllowMacroExpansion); 716 } 717 718 /// \brief Complete the lexing of a string literal where the first token has 719 /// already been lexed (see LexStringLiteral). 720 bool FinishLexStringLiteral(Token &Result, std::string &String, 721 const char *DiagnosticTag, 722 bool AllowMacroExpansion); 723 724 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 725 /// something not a comment. This is useful in -E -C mode where comments 726 /// would foul up preprocessor directive handling. 727 void LexNonComment(Token &Result) { 728 do 729 Lex(Result); 730 while (Result.getKind() == tok::comment); 731 } 732 733 /// LexUnexpandedToken - This is just like Lex, but this disables macro 734 /// expansion of identifier tokens. 735 void LexUnexpandedToken(Token &Result) { 736 // Disable macro expansion. 737 bool OldVal = DisableMacroExpansion; 738 DisableMacroExpansion = true; 739 // Lex the token. 740 Lex(Result); 741 742 // Reenable it. 743 DisableMacroExpansion = OldVal; 744 } 745 746 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 747 /// expansion of identifier tokens. 748 void LexUnexpandedNonComment(Token &Result) { 749 do 750 LexUnexpandedToken(Result); 751 while (Result.getKind() == tok::comment); 752 } 753 754 /// Disables macro expansion everywhere except for preprocessor directives. 755 void SetMacroExpansionOnlyInDirectives() { 756 DisableMacroExpansion = true; 757 MacroExpansionInDirectivesOverride = true; 758 } 759 760 /// LookAhead - This peeks ahead N tokens and returns that token without 761 /// consuming any tokens. LookAhead(0) returns the next token that would be 762 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 763 /// returns normal tokens after phase 5. As such, it is equivalent to using 764 /// 'Lex', not 'LexUnexpandedToken'. 765 const Token &LookAhead(unsigned N) { 766 if (CachedLexPos + N < CachedTokens.size()) 767 return CachedTokens[CachedLexPos+N]; 768 else 769 return PeekAhead(N+1); 770 } 771 772 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 773 /// this allows to revert a specific number of tokens. 774 /// Note that the number of tokens being reverted should be up to the last 775 /// backtrack position, not more. 776 void RevertCachedTokens(unsigned N) { 777 assert(isBacktrackEnabled() && 778 "Should only be called when tokens are cached for backtracking"); 779 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 780 && "Should revert tokens up to the last backtrack position, not more"); 781 assert(signed(CachedLexPos) - signed(N) >= 0 && 782 "Corrupted backtrack positions ?"); 783 CachedLexPos -= N; 784 } 785 786 /// EnterToken - Enters a token in the token stream to be lexed next. If 787 /// BackTrack() is called afterwards, the token will remain at the insertion 788 /// point. 789 void EnterToken(const Token &Tok) { 790 EnterCachingLexMode(); 791 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 792 } 793 794 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 795 /// tokens (because backtrack is enabled) it should replace the most recent 796 /// cached tokens with the given annotation token. This function has no effect 797 /// if backtracking is not enabled. 798 /// 799 /// Note that the use of this function is just for optimization; so that the 800 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 801 /// invoked. 802 void AnnotateCachedTokens(const Token &Tok) { 803 assert(Tok.isAnnotation() && "Expected annotation token"); 804 if (CachedLexPos != 0 && isBacktrackEnabled()) 805 AnnotatePreviousCachedTokens(Tok); 806 } 807 808 /// \brief Replace the last token with an annotation token. 809 /// 810 /// Like AnnotateCachedTokens(), this routine replaces an 811 /// already-parsed (and resolved) token with an annotation 812 /// token. However, this routine only replaces the last token with 813 /// the annotation token; it does not affect any other cached 814 /// tokens. This function has no effect if backtracking is not 815 /// enabled. 816 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 817 assert(Tok.isAnnotation() && "Expected annotation token"); 818 if (CachedLexPos != 0 && isBacktrackEnabled()) 819 CachedTokens[CachedLexPos-1] = Tok; 820 } 821 822 /// TypoCorrectToken - Update the current token to represent the provided 823 /// identifier, in order to cache an action performed by typo correction. 824 void TypoCorrectToken(const Token &Tok) { 825 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 826 if (CachedLexPos != 0 && isBacktrackEnabled()) 827 CachedTokens[CachedLexPos-1] = Tok; 828 } 829 830 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 831 /// CurTokenLexer pointers. 832 void recomputeCurLexerKind(); 833 834 /// \brief Returns true if incremental processing is enabled 835 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 836 837 /// \brief Enables the incremental processing 838 void enableIncrementalProcessing(bool value = true) { 839 IncrementalProcessing = value; 840 } 841 842 /// \brief Specify the point at which code-completion will be performed. 843 /// 844 /// \param File the file in which code completion should occur. If 845 /// this file is included multiple times, code-completion will 846 /// perform completion the first time it is included. If NULL, this 847 /// function clears out the code-completion point. 848 /// 849 /// \param Line the line at which code completion should occur 850 /// (1-based). 851 /// 852 /// \param Column the column at which code completion should occur 853 /// (1-based). 854 /// 855 /// \returns true if an error occurred, false otherwise. 856 bool SetCodeCompletionPoint(const FileEntry *File, 857 unsigned Line, unsigned Column); 858 859 /// \brief Determine if we are performing code completion. 860 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 861 862 /// \brief Returns the location of the code-completion point. 863 /// Returns an invalid location if code-completion is not enabled or the file 864 /// containing the code-completion point has not been lexed yet. 865 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 866 867 /// \brief Returns the start location of the file of code-completion point. 868 /// Returns an invalid location if code-completion is not enabled or the file 869 /// containing the code-completion point has not been lexed yet. 870 SourceLocation getCodeCompletionFileLoc() const { 871 return CodeCompletionFileLoc; 872 } 873 874 /// \brief Returns true if code-completion is enabled and we have hit the 875 /// code-completion point. 876 bool isCodeCompletionReached() const { return CodeCompletionReached; } 877 878 /// \brief Note that we hit the code-completion point. 879 void setCodeCompletionReached() { 880 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 881 CodeCompletionReached = true; 882 // Silence any diagnostics that occur after we hit the code-completion. 883 getDiagnostics().setSuppressAllDiagnostics(true); 884 } 885 886 /// \brief The location of the currently-active \#pragma clang 887 /// arc_cf_code_audited begin. Returns an invalid location if there 888 /// is no such pragma active. 889 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 890 return PragmaARCCFCodeAuditedLoc; 891 } 892 893 /// \brief Set the location of the currently-active \#pragma clang 894 /// arc_cf_code_audited begin. An invalid location ends the pragma. 895 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 896 PragmaARCCFCodeAuditedLoc = Loc; 897 } 898 899 /// \brief Instruct the preprocessor to skip part of the main source file. 900 /// 901 /// \param Bytes The number of bytes in the preamble to skip. 902 /// 903 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 904 /// start of a line. 905 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 906 SkipMainFilePreamble.first = Bytes; 907 SkipMainFilePreamble.second = StartOfLine; 908 } 909 910 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 911 /// the specified Token's location, translating the token's start 912 /// position in the current buffer into a SourcePosition object for rendering. 913 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 914 return Diags->Report(Loc, DiagID); 915 } 916 917 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 918 return Diags->Report(Tok.getLocation(), DiagID); 919 } 920 921 /// getSpelling() - Return the 'spelling' of the token at the given 922 /// location; does not go up to the spelling location or down to the 923 /// expansion location. 924 /// 925 /// \param buffer A buffer which will be used only if the token requires 926 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 927 /// \param invalid If non-null, will be set \c true if an error occurs. 928 StringRef getSpelling(SourceLocation loc, 929 SmallVectorImpl<char> &buffer, 930 bool *invalid = 0) const { 931 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 932 } 933 934 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 935 /// token is the characters used to represent the token in the source file 936 /// after trigraph expansion and escaped-newline folding. In particular, this 937 /// wants to get the true, uncanonicalized, spelling of things like digraphs 938 /// UCNs, etc. 939 /// 940 /// \param Invalid If non-null, will be set \c true if an error occurs. 941 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 942 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 943 } 944 945 /// getSpelling - This method is used to get the spelling of a token into a 946 /// preallocated buffer, instead of as an std::string. The caller is required 947 /// to allocate enough space for the token, which is guaranteed to be at least 948 /// Tok.getLength() bytes long. The length of the actual result is returned. 949 /// 950 /// Note that this method may do two possible things: it may either fill in 951 /// the buffer specified with characters, or it may *change the input pointer* 952 /// to point to a constant buffer with the data already in it (avoiding a 953 /// copy). The caller is not allowed to modify the returned buffer pointer 954 /// if an internal buffer is returned. 955 unsigned getSpelling(const Token &Tok, const char *&Buffer, 956 bool *Invalid = 0) const { 957 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 958 } 959 960 /// getSpelling - This method is used to get the spelling of a token into a 961 /// SmallVector. Note that the returned StringRef may not point to the 962 /// supplied buffer if a copy can be avoided. 963 StringRef getSpelling(const Token &Tok, 964 SmallVectorImpl<char> &Buffer, 965 bool *Invalid = 0) const; 966 967 /// \brief Relex the token at the specified location. 968 /// \returns true if there was a failure, false on success. 969 bool getRawToken(SourceLocation Loc, Token &Result) { 970 return Lexer::getRawToken(Loc, Result, SourceMgr, LangOpts); 971 } 972 973 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 974 /// with length 1, return the character. 975 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 976 bool *Invalid = 0) const { 977 assert(Tok.is(tok::numeric_constant) && 978 Tok.getLength() == 1 && "Called on unsupported token"); 979 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 980 981 // If the token is carrying a literal data pointer, just use it. 982 if (const char *D = Tok.getLiteralData()) 983 return *D; 984 985 // Otherwise, fall back on getCharacterData, which is slower, but always 986 // works. 987 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 988 } 989 990 /// \brief Retrieve the name of the immediate macro expansion. 991 /// 992 /// This routine starts from a source location, and finds the name of the macro 993 /// responsible for its immediate expansion. It looks through any intervening 994 /// macro argument expansions to compute this. It returns a StringRef which 995 /// refers to the SourceManager-owned buffer of the source where that macro 996 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 997 StringRef getImmediateMacroName(SourceLocation Loc) { 998 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 999 } 1000 1001 /// CreateString - Plop the specified string into a scratch buffer and set the 1002 /// specified token's location and length to it. If specified, the source 1003 /// location provides a location of the expansion point of the token. 1004 void CreateString(StringRef Str, Token &Tok, 1005 SourceLocation ExpansionLocStart = SourceLocation(), 1006 SourceLocation ExpansionLocEnd = SourceLocation()); 1007 1008 /// \brief Computes the source location just past the end of the 1009 /// token at this source location. 1010 /// 1011 /// This routine can be used to produce a source location that 1012 /// points just past the end of the token referenced by \p Loc, and 1013 /// is generally used when a diagnostic needs to point just after a 1014 /// token where it expected something different that it received. If 1015 /// the returned source location would not be meaningful (e.g., if 1016 /// it points into a macro), this routine returns an invalid 1017 /// source location. 1018 /// 1019 /// \param Offset an offset from the end of the token, where the source 1020 /// location should refer to. The default offset (0) produces a source 1021 /// location pointing just past the end of the token; an offset of 1 produces 1022 /// a source location pointing to the last character in the token, etc. 1023 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 1024 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 1025 } 1026 1027 /// \brief Returns true if the given MacroID location points at the first 1028 /// token of the macro expansion. 1029 /// 1030 /// \param MacroBegin If non-null and function returns true, it is set to 1031 /// begin location of the macro. 1032 bool isAtStartOfMacroExpansion(SourceLocation loc, 1033 SourceLocation *MacroBegin = 0) const { 1034 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 1035 MacroBegin); 1036 } 1037 1038 /// \brief Returns true if the given MacroID location points at the last 1039 /// token of the macro expansion. 1040 /// 1041 /// \param MacroEnd If non-null and function returns true, it is set to 1042 /// end location of the macro. 1043 bool isAtEndOfMacroExpansion(SourceLocation loc, 1044 SourceLocation *MacroEnd = 0) const { 1045 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 1046 } 1047 1048 /// DumpToken - Print the token to stderr, used for debugging. 1049 /// 1050 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 1051 void DumpLocation(SourceLocation Loc) const; 1052 void DumpMacro(const MacroInfo &MI) const; 1053 1054 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 1055 /// token, return a new location that specifies a character within the token. 1056 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 1057 unsigned Char) const { 1058 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 1059 } 1060 1061 /// IncrementPasteCounter - Increment the counters for the number of token 1062 /// paste operations performed. If fast was specified, this is a 'fast paste' 1063 /// case we handled. 1064 /// 1065 void IncrementPasteCounter(bool isFast) { 1066 if (isFast) 1067 ++NumFastTokenPaste; 1068 else 1069 ++NumTokenPaste; 1070 } 1071 1072 void PrintStats(); 1073 1074 size_t getTotalMemory() const; 1075 1076 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 1077 /// comment (/##/) in microsoft mode, this method handles updating the current 1078 /// state, returning the token on the next source line. 1079 void HandleMicrosoftCommentPaste(Token &Tok); 1080 1081 //===--------------------------------------------------------------------===// 1082 // Preprocessor callback methods. These are invoked by a lexer as various 1083 // directives and events are found. 1084 1085 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1086 /// identifier information for the token and install it into the token, 1087 /// updating the token kind accordingly. 1088 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1089 1090private: 1091 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1092 1093public: 1094 1095 // SetPoisonReason - Call this function to indicate the reason for 1096 // poisoning an identifier. If that identifier is accessed while 1097 // poisoned, then this reason will be used instead of the default 1098 // "poisoned" diagnostic. 1099 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1100 1101 // HandlePoisonedIdentifier - Display reason for poisoned 1102 // identifier. 1103 void HandlePoisonedIdentifier(Token & Tok); 1104 1105 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1106 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1107 if(II->isPoisoned()) { 1108 HandlePoisonedIdentifier(Identifier); 1109 } 1110 } 1111 } 1112 1113private: 1114 /// Identifiers used for SEH handling in Borland. These are only 1115 /// allowed in particular circumstances 1116 // __except block 1117 IdentifierInfo *Ident__exception_code, 1118 *Ident___exception_code, 1119 *Ident_GetExceptionCode; 1120 // __except filter expression 1121 IdentifierInfo *Ident__exception_info, 1122 *Ident___exception_info, 1123 *Ident_GetExceptionInfo; 1124 // __finally 1125 IdentifierInfo *Ident__abnormal_termination, 1126 *Ident___abnormal_termination, 1127 *Ident_AbnormalTermination; 1128public: 1129 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1130 1131 /// HandleIdentifier - This callback is invoked when the lexer reads an 1132 /// identifier and has filled in the tokens IdentifierInfo member. This 1133 /// callback potentially macro expands it or turns it into a named token (like 1134 /// 'for'). 1135 void HandleIdentifier(Token &Identifier); 1136 1137 1138 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1139 /// the current file. This either returns the EOF token and returns true, or 1140 /// pops a level off the include stack and returns false, at which point the 1141 /// client should call lex again. 1142 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1143 1144 /// HandleEndOfTokenLexer - This callback is invoked when the current 1145 /// TokenLexer hits the end of its token stream. 1146 bool HandleEndOfTokenLexer(Token &Result); 1147 1148 /// HandleDirective - This callback is invoked when the lexer sees a # token 1149 /// at the start of a line. This consumes the directive, modifies the 1150 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1151 /// read is the correct one. 1152 void HandleDirective(Token &Result); 1153 1154 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1155 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1156 /// true, then we consider macros that expand to zero tokens as being ok. 1157 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1158 1159 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1160 /// current line until the tok::eod token is found. 1161 void DiscardUntilEndOfDirective(); 1162 1163 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1164 /// __DATE__ or __TIME__ in the file so far. 1165 bool SawDateOrTime() const { 1166 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1167 } 1168 unsigned getCounterValue() const { return CounterValue; } 1169 void setCounterValue(unsigned V) { CounterValue = V; } 1170 1171 /// \brief Retrieves the module that we're currently building, if any. 1172 Module *getCurrentModule(); 1173 1174 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1175 MacroInfo *AllocateMacroInfo(SourceLocation L); 1176 1177 /// \brief Allocate a new MacroInfo object which is clone of \p MI. 1178 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1179 1180 /// \brief Turn the specified lexer token into a fully checked and spelled 1181 /// filename, e.g. as an operand of \#include. 1182 /// 1183 /// The caller is expected to provide a buffer that is large enough to hold 1184 /// the spelling of the filename, but is also expected to handle the case 1185 /// when this method decides to use a different buffer. 1186 /// 1187 /// \returns true if the input filename was in <>'s or false if it was 1188 /// in ""'s. 1189 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1190 1191 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1192 /// 1193 /// Returns null on failure. \p isAngled indicates whether the file 1194 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1195 const FileEntry *LookupFile(StringRef Filename, 1196 bool isAngled, const DirectoryLookup *FromDir, 1197 const DirectoryLookup *&CurDir, 1198 SmallVectorImpl<char> *SearchPath, 1199 SmallVectorImpl<char> *RelativePath, 1200 Module **SuggestedModule, 1201 bool SkipCache = false); 1202 1203 /// GetCurLookup - The DirectoryLookup structure used to find the current 1204 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1205 /// implement \#include_next and find directory-specific properties. 1206 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1207 1208 /// \brief Return true if we're in the top-level file, not in a \#include. 1209 bool isInPrimaryFile() const; 1210 1211 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1212 /// from a macro as multiple tokens, which need to be glued together. This 1213 /// occurs for code like: 1214 /// \code 1215 /// \#define FOO <x/y.h> 1216 /// \#include FOO 1217 /// \endcode 1218 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1219 /// 1220 /// This code concatenates and consumes tokens up to the '>' token. It 1221 /// returns false if the > was found, otherwise it returns true if it finds 1222 /// and consumes the EOD marker. 1223 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1224 SourceLocation &End); 1225 1226 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1227 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1228 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1229 1230private: 1231 1232 void PushIncludeMacroStack() { 1233 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1234 CurLexer.take(), 1235 CurPTHLexer.take(), 1236 CurPPLexer, 1237 CurTokenLexer.take(), 1238 CurDirLookup)); 1239 CurPPLexer = 0; 1240 } 1241 1242 void PopIncludeMacroStack() { 1243 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1244 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1245 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1246 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1247 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1248 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1249 IncludeMacroStack.pop_back(); 1250 } 1251 1252 /// \brief Allocate a new MacroInfo object. 1253 MacroInfo *AllocateMacroInfo(); 1254 1255 /// \brief Release the specified MacroInfo for re-use. 1256 /// 1257 /// This memory will be reused for allocating new MacroInfo objects. 1258 void ReleaseMacroInfo(MacroInfo* MI); 1259 1260 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1261 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1262 /// and discards the rest of the macro line if the macro name is invalid. 1263 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1264 1265 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1266 /// definition has just been read. Lex the rest of the arguments and the 1267 /// closing ), updating MI with what we learn and saving in LastTok the 1268 /// last token read. 1269 /// Return true if an error occurs parsing the arg list. 1270 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1271 1272 /// We just read a \#if or related directive and decided that the 1273 /// subsequent tokens are in the \#if'd out portion of the 1274 /// file. Lex the rest of the file, until we see an \#endif. If \p 1275 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1276 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1277 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1278 /// already seen one so a \#else directive is a duplicate. When this returns, 1279 /// the caller can lex the first valid token. 1280 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1281 bool FoundNonSkipPortion, bool FoundElse, 1282 SourceLocation ElseLoc = SourceLocation()); 1283 1284 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1285 void PTHSkipExcludedConditionalBlock(); 1286 1287 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1288 /// may occur after a #if or #elif directive and return it as a bool. If the 1289 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1290 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1291 1292 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1293 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1294 void RegisterBuiltinPragmas(); 1295 1296 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1297 void RegisterBuiltinMacros(); 1298 1299 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1300 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1301 /// the macro should not be expanded return true, otherwise return false. 1302 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1303 1304 /// \brief Cache macro expanded tokens for TokenLexers. 1305 // 1306 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1307 /// going to lex in the cache and when it finishes the tokens are removed 1308 /// from the end of the cache. 1309 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1310 ArrayRef<Token> tokens); 1311 void removeCachedMacroExpandedTokensOfLastLexer(); 1312 friend void TokenLexer::ExpandFunctionArguments(); 1313 1314 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1315 /// lexed is a '('. If so, consume the token and return true, if not, this 1316 /// method should have no observable side-effect on the lexed tokens. 1317 bool isNextPPTokenLParen(); 1318 1319 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1320 /// invoked to read all of the formal arguments specified for the macro 1321 /// invocation. This returns null on error. 1322 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1323 SourceLocation &ExpansionEnd); 1324 1325 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1326 /// as a builtin macro, handle it and return the next token as 'Tok'. 1327 void ExpandBuiltinMacro(Token &Tok); 1328 1329 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1330 /// return the first token after the directive. The _Pragma token has just 1331 /// been read into 'Tok'. 1332 void Handle_Pragma(Token &Tok); 1333 1334 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1335 /// is not enclosed within a string literal. 1336 void HandleMicrosoft__pragma(Token &Tok); 1337 1338 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1339 /// start lexing tokens from it instead of the current buffer. 1340 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1341 1342 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1343 /// start getting tokens from it using the PTH cache. 1344 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1345 1346 /// IsFileLexer - Returns true if we are lexing from a file and not a 1347 /// pragma or a macro. 1348 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1349 return L ? !L->isPragmaLexer() : P != 0; 1350 } 1351 1352 static bool IsFileLexer(const IncludeStackInfo& I) { 1353 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1354 } 1355 1356 bool IsFileLexer() const { 1357 return IsFileLexer(CurLexer.get(), CurPPLexer); 1358 } 1359 1360 //===--------------------------------------------------------------------===// 1361 // Caching stuff. 1362 void CachingLex(Token &Result); 1363 bool InCachingLexMode() const { 1364 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1365 // that we are past EOF, not that we are in CachingLex mode. 1366 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1367 !IncludeMacroStack.empty(); 1368 } 1369 void EnterCachingLexMode(); 1370 void ExitCachingLexMode() { 1371 if (InCachingLexMode()) 1372 RemoveTopOfLexerStack(); 1373 } 1374 const Token &PeekAhead(unsigned N); 1375 void AnnotatePreviousCachedTokens(const Token &Tok); 1376 1377 //===--------------------------------------------------------------------===// 1378 /// Handle*Directive - implement the various preprocessor directives. These 1379 /// should side-effect the current preprocessor object so that the next call 1380 /// to Lex() will return the appropriate token next. 1381 void HandleLineDirective(Token &Tok); 1382 void HandleDigitDirective(Token &Tok); 1383 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1384 void HandleIdentSCCSDirective(Token &Tok); 1385 void HandleMacroPublicDirective(Token &Tok); 1386 void HandleMacroPrivateDirective(Token &Tok); 1387 1388 // File inclusion. 1389 void HandleIncludeDirective(SourceLocation HashLoc, 1390 Token &Tok, 1391 const DirectoryLookup *LookupFrom = 0, 1392 bool isImport = false); 1393 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1394 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1395 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1396 void HandleMicrosoftImportDirective(Token &Tok); 1397 1398 // Macro handling. 1399 void HandleDefineDirective(Token &Tok); 1400 void HandleUndefDirective(Token &Tok); 1401 void UndefineMacro(IdentifierInfo *II, MacroInfo *MI, 1402 SourceLocation UndefLoc); 1403 1404 // Conditional Inclusion. 1405 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1406 bool ReadAnyTokensBeforeDirective); 1407 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1408 void HandleEndifDirective(Token &Tok); 1409 void HandleElseDirective(Token &Tok); 1410 void HandleElifDirective(Token &Tok); 1411 1412 // Pragmas. 1413 void HandlePragmaDirective(unsigned Introducer); 1414public: 1415 void HandlePragmaOnce(Token &OnceTok); 1416 void HandlePragmaMark(); 1417 void HandlePragmaPoison(Token &PoisonTok); 1418 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1419 void HandlePragmaDependency(Token &DependencyTok); 1420 void HandlePragmaComment(Token &CommentTok); 1421 void HandlePragmaMessage(Token &MessageTok); 1422 void HandlePragmaPushMacro(Token &Tok); 1423 void HandlePragmaPopMacro(Token &Tok); 1424 void HandlePragmaIncludeAlias(Token &Tok); 1425 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1426 1427 // Return true and store the first token only if any CommentHandler 1428 // has inserted some tokens and getCommentRetentionState() is false. 1429 bool HandleComment(Token &Token, SourceRange Comment); 1430 1431 /// \brief A macro is used, update information about macros that need unused 1432 /// warnings. 1433 void markMacroAsUsed(MacroInfo *MI); 1434}; 1435 1436/// \brief Abstract base class that describes a handler that will receive 1437/// source ranges for each of the comments encountered in the source file. 1438class CommentHandler { 1439public: 1440 virtual ~CommentHandler(); 1441 1442 // The handler shall return true if it has pushed any tokens 1443 // to be read using e.g. EnterToken or EnterTokenStream. 1444 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1445}; 1446 1447} // end namespace clang 1448 1449#endif 1450