Preprocessor.h revision 1952354bd376062c3ab3d328c0fc6c36530c9309
1//===--- Preprocessor.h - C Language Family Preprocessor --------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file is distributed under the University of Illinois Open Source 6// License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// This file defines the Preprocessor interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_LEX_PREPROCESSOR_H 15#define LLVM_CLANG_LEX_PREPROCESSOR_H 16 17#include "clang/Lex/MacroInfo.h" 18#include "clang/Lex/Lexer.h" 19#include "clang/Lex/PTHLexer.h" 20#include "clang/Lex/PPCallbacks.h" 21#include "clang/Lex/TokenLexer.h" 22#include "clang/Lex/PTHManager.h" 23#include "clang/Basic/Builtins.h" 24#include "clang/Basic/Diagnostic.h" 25#include "clang/Basic/IdentifierTable.h" 26#include "clang/Basic/SourceLocation.h" 27#include "llvm/ADT/DenseMap.h" 28#include "llvm/ADT/IntrusiveRefCntPtr.h" 29#include "llvm/ADT/SmallPtrSet.h" 30#include "llvm/ADT/OwningPtr.h" 31#include "llvm/ADT/SmallVector.h" 32#include "llvm/ADT/ArrayRef.h" 33#include "llvm/Support/Allocator.h" 34#include <vector> 35 36namespace llvm { 37 template<unsigned InternalLen> class SmallString; 38} 39 40namespace clang { 41 42class SourceManager; 43class ExternalPreprocessorSource; 44class FileManager; 45class FileEntry; 46class HeaderSearch; 47class PragmaNamespace; 48class PragmaHandler; 49class CommentHandler; 50class ScratchBuffer; 51class TargetInfo; 52class PPCallbacks; 53class CodeCompletionHandler; 54class DirectoryLookup; 55class PreprocessingRecord; 56class ModuleLoader; 57 58/// \brief Stores token information for comparing actual tokens with 59/// predefined values. Only handles simple tokens and identifiers. 60class TokenValue { 61 tok::TokenKind Kind; 62 IdentifierInfo *II; 63 64public: 65 TokenValue(tok::TokenKind Kind) : Kind(Kind), II(0) { 66 assert(Kind != tok::raw_identifier && "Raw identifiers are not supported."); 67 assert(Kind != tok::identifier && 68 "Identifiers should be created by TokenValue(IdentifierInfo *)"); 69 assert(!tok::isLiteral(Kind) && "Literals are not supported."); 70 assert(!tok::isAnnotation(Kind) && "Annotations are not supported."); 71 } 72 TokenValue(IdentifierInfo *II) : Kind(tok::identifier), II(II) {} 73 bool operator==(const Token &Tok) const { 74 return Tok.getKind() == Kind && 75 (!II || II == Tok.getIdentifierInfo()); 76 } 77}; 78 79/// Preprocessor - This object engages in a tight little dance with the lexer to 80/// efficiently preprocess tokens. Lexers know only about tokens within a 81/// single source file, and don't know anything about preprocessor-level issues 82/// like the \#include stack, token expansion, etc. 83/// 84class Preprocessor : public RefCountedBase<Preprocessor> { 85 DiagnosticsEngine *Diags; 86 LangOptions &LangOpts; 87 const TargetInfo *Target; 88 FileManager &FileMgr; 89 SourceManager &SourceMgr; 90 ScratchBuffer *ScratchBuf; 91 HeaderSearch &HeaderInfo; 92 ModuleLoader &TheModuleLoader; 93 94 /// \brief External source of macros. 95 ExternalPreprocessorSource *ExternalSource; 96 97 98 /// PTH - An optional PTHManager object used for getting tokens from 99 /// a token cache rather than lexing the original source file. 100 OwningPtr<PTHManager> PTH; 101 102 /// BP - A BumpPtrAllocator object used to quickly allocate and release 103 /// objects internal to the Preprocessor. 104 llvm::BumpPtrAllocator BP; 105 106 /// Identifiers for builtin macros and other builtins. 107 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 108 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 109 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 110 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 111 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 112 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 113 IdentifierInfo *Ident_Pragma, *Ident__pragma; // _Pragma, __pragma 114 IdentifierInfo *Ident__VA_ARGS__; // __VA_ARGS__ 115 IdentifierInfo *Ident__has_feature; // __has_feature 116 IdentifierInfo *Ident__has_extension; // __has_extension 117 IdentifierInfo *Ident__has_builtin; // __has_builtin 118 IdentifierInfo *Ident__has_attribute; // __has_attribute 119 IdentifierInfo *Ident__has_include; // __has_include 120 IdentifierInfo *Ident__has_include_next; // __has_include_next 121 IdentifierInfo *Ident__has_warning; // __has_warning 122 IdentifierInfo *Ident__building_module; // __building_module 123 IdentifierInfo *Ident__MODULE__; // __MODULE__ 124 125 SourceLocation DATELoc, TIMELoc; 126 unsigned CounterValue; // Next __COUNTER__ value. 127 128 enum { 129 /// MaxIncludeStackDepth - Maximum depth of \#includes. 130 MaxAllowedIncludeStackDepth = 200 131 }; 132 133 // State that is set before the preprocessor begins. 134 bool KeepComments : 1; 135 bool KeepMacroComments : 1; 136 bool SuppressIncludeNotFoundError : 1; 137 138 // State that changes while the preprocessor runs: 139 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 140 141 /// Whether the preprocessor owns the header search object. 142 bool OwnsHeaderSearch : 1; 143 144 /// DisableMacroExpansion - True if macro expansion is disabled. 145 bool DisableMacroExpansion : 1; 146 147 /// MacroExpansionInDirectivesOverride - Temporarily disables 148 /// DisableMacroExpansion (i.e. enables expansion) when parsing preprocessor 149 /// directives. 150 bool MacroExpansionInDirectivesOverride : 1; 151 152 class ResetMacroExpansionHelper; 153 154 /// \brief Whether we have already loaded macros from the external source. 155 mutable bool ReadMacrosFromExternalSource : 1; 156 157 /// \brief True if pragmas are enabled. 158 bool PragmasEnabled : 1; 159 160 /// \brief True if we are pre-expanding macro arguments. 161 bool InMacroArgPreExpansion; 162 163 /// Identifiers - This is mapping/lookup information for all identifiers in 164 /// the program, including program keywords. 165 mutable IdentifierTable Identifiers; 166 167 /// Selectors - This table contains all the selectors in the program. Unlike 168 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 169 /// It is declared/expanded here because it's role/lifetime is 170 /// conceptually similar the IdentifierTable. In addition, the current control 171 /// flow (in clang::ParseAST()), make it convenient to put here. 172 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 173 /// the lifetime of the preprocessor. 174 SelectorTable Selectors; 175 176 /// BuiltinInfo - Information about builtins. 177 Builtin::Context BuiltinInfo; 178 179 /// PragmaHandlers - This tracks all of the pragmas that the client registered 180 /// with this preprocessor. 181 PragmaNamespace *PragmaHandlers; 182 183 /// \brief Tracks all of the comment handlers that the client registered 184 /// with this preprocessor. 185 std::vector<CommentHandler *> CommentHandlers; 186 187 /// \brief True if we want to ignore EOF token and continue later on (thus 188 /// avoid tearing the Lexer and etc. down). 189 bool IncrementalProcessing; 190 191 /// \brief The code-completion handler. 192 CodeCompletionHandler *CodeComplete; 193 194 /// \brief The file that we're performing code-completion for, if any. 195 const FileEntry *CodeCompletionFile; 196 197 /// \brief The offset in file for the code-completion point. 198 unsigned CodeCompletionOffset; 199 200 /// \brief The location for the code-completion point. This gets instantiated 201 /// when the CodeCompletionFile gets \#include'ed for preprocessing. 202 SourceLocation CodeCompletionLoc; 203 204 /// \brief The start location for the file of the code-completion point. 205 /// 206 /// This gets instantiated when the CodeCompletionFile gets \#include'ed 207 /// for preprocessing. 208 SourceLocation CodeCompletionFileLoc; 209 210 /// \brief The source location of the 'import' contextual keyword we just 211 /// lexed, if any. 212 SourceLocation ModuleImportLoc; 213 214 /// \brief The module import path that we're currently processing. 215 llvm::SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> 216 ModuleImportPath; 217 218 /// \brief Whether the module import expectes an identifier next. Otherwise, 219 /// it expects a '.' or ';'. 220 bool ModuleImportExpectsIdentifier; 221 222 /// \brief The source location of the currently-active 223 /// #pragma clang arc_cf_code_audited begin. 224 SourceLocation PragmaARCCFCodeAuditedLoc; 225 226 /// \brief True if we hit the code-completion point. 227 bool CodeCompletionReached; 228 229 /// \brief The number of bytes that we will initially skip when entering the 230 /// main file, which is used when loading a precompiled preamble, along 231 /// with a flag that indicates whether skipping this number of bytes will 232 /// place the lexer at the start of a line. 233 std::pair<unsigned, bool> SkipMainFilePreamble; 234 235 /// CurLexer - This is the current top of the stack that we're lexing from if 236 /// not expanding a macro and we are lexing directly from source code. 237 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 238 OwningPtr<Lexer> CurLexer; 239 240 /// CurPTHLexer - This is the current top of stack that we're lexing from if 241 /// not expanding from a macro and we are lexing from a PTH cache. 242 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 243 OwningPtr<PTHLexer> CurPTHLexer; 244 245 /// CurPPLexer - This is the current top of the stack what we're lexing from 246 /// if not expanding a macro. This is an alias for either CurLexer or 247 /// CurPTHLexer. 248 PreprocessorLexer *CurPPLexer; 249 250 /// CurLookup - The DirectoryLookup structure used to find the current 251 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 252 /// implement \#include_next and find directory-specific properties. 253 const DirectoryLookup *CurDirLookup; 254 255 /// CurTokenLexer - This is the current macro we are expanding, if we are 256 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 257 OwningPtr<TokenLexer> CurTokenLexer; 258 259 /// \brief The kind of lexer we're currently working with. 260 enum CurLexerKind { 261 CLK_Lexer, 262 CLK_PTHLexer, 263 CLK_TokenLexer, 264 CLK_CachingLexer, 265 CLK_LexAfterModuleImport 266 } CurLexerKind; 267 268 /// IncludeMacroStack - This keeps track of the stack of files currently 269 /// \#included, and macros currently being expanded from, not counting 270 /// CurLexer/CurTokenLexer. 271 struct IncludeStackInfo { 272 enum CurLexerKind CurLexerKind; 273 Lexer *TheLexer; 274 PTHLexer *ThePTHLexer; 275 PreprocessorLexer *ThePPLexer; 276 TokenLexer *TheTokenLexer; 277 const DirectoryLookup *TheDirLookup; 278 279 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 280 PreprocessorLexer* PPL, 281 TokenLexer* TL, const DirectoryLookup *D) 282 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 283 TheTokenLexer(TL), TheDirLookup(D) {} 284 }; 285 std::vector<IncludeStackInfo> IncludeMacroStack; 286 287 /// Callbacks - These are actions invoked when some preprocessor activity is 288 /// encountered (e.g. a file is \#included, etc). 289 PPCallbacks *Callbacks; 290 291 struct MacroExpandsInfo { 292 Token Tok; 293 MacroInfo *MI; 294 SourceRange Range; 295 MacroExpandsInfo(Token Tok, MacroInfo *MI, SourceRange Range) 296 : Tok(Tok), MI(MI), Range(Range) { } 297 }; 298 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 299 300 /// Macros - For each IdentifierInfo that was associated with a macro, we 301 /// keep a mapping to the history of all macro definitions and #undefs in 302 /// the reverse order (the latest one is in the head of the list). 303 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 304 friend class ASTReader; 305 306 /// \brief Macros that we want to warn because they are not used at the end 307 /// of the translation unit; we store just their SourceLocations instead 308 /// something like MacroInfo*. The benefit of this is that when we are 309 /// deserializing from PCH, we don't need to deserialize identifier & macros 310 /// just so that we can report that they are unused, we just warn using 311 /// the SourceLocations of this set (that will be filled by the ASTReader). 312 /// We are using SmallPtrSet instead of a vector for faster removal. 313 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 314 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 315 316 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 317 /// reused for quick allocation. 318 MacroArgs *MacroArgCache; 319 friend class MacroArgs; 320 321 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 322 /// push_macro directive, we keep a MacroInfo stack used to restore 323 /// previous macro value. 324 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 325 326 // Various statistics we track for performance analysis. 327 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 328 unsigned NumIf, NumElse, NumEndif; 329 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 330 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 331 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 332 unsigned NumSkipped; 333 334 /// Predefines - This string is the predefined macros that preprocessor 335 /// should use from the command line etc. 336 std::string Predefines; 337 338 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 339 enum { TokenLexerCacheSize = 8 }; 340 unsigned NumCachedTokenLexers; 341 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 342 343 /// \brief Keeps macro expanded tokens for TokenLexers. 344 // 345 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 346 /// going to lex in the cache and when it finishes the tokens are removed 347 /// from the end of the cache. 348 SmallVector<Token, 16> MacroExpandedTokens; 349 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 350 351 /// \brief A record of the macro definitions and expansions that 352 /// occurred during preprocessing. 353 /// 354 /// This is an optional side structure that can be enabled with 355 /// \c createPreprocessingRecord() prior to preprocessing. 356 PreprocessingRecord *Record; 357 358private: // Cached tokens state. 359 typedef SmallVector<Token, 1> CachedTokensTy; 360 361 /// CachedTokens - Cached tokens are stored here when we do backtracking or 362 /// lookahead. They are "lexed" by the CachingLex() method. 363 CachedTokensTy CachedTokens; 364 365 /// CachedLexPos - The position of the cached token that CachingLex() should 366 /// "lex" next. If it points beyond the CachedTokens vector, it means that 367 /// a normal Lex() should be invoked. 368 CachedTokensTy::size_type CachedLexPos; 369 370 /// BacktrackPositions - Stack of backtrack positions, allowing nested 371 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 372 /// indicate where CachedLexPos should be set when the BackTrack() method is 373 /// invoked (at which point the last position is popped). 374 std::vector<CachedTokensTy::size_type> BacktrackPositions; 375 376 struct MacroInfoChain { 377 MacroInfo MI; 378 MacroInfoChain *Next; 379 MacroInfoChain *Prev; 380 }; 381 382 /// MacroInfos are managed as a chain for easy disposal. This is the head 383 /// of that list. 384 MacroInfoChain *MIChainHead; 385 386 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 387 /// allocation. 388 MacroInfoChain *MICache; 389 390public: 391 Preprocessor(DiagnosticsEngine &diags, LangOptions &opts, 392 const TargetInfo *target, 393 SourceManager &SM, HeaderSearch &Headers, 394 ModuleLoader &TheModuleLoader, 395 IdentifierInfoLookup *IILookup = 0, 396 bool OwnsHeaderSearch = false, 397 bool DelayInitialization = false, 398 bool IncrProcessing = false); 399 400 ~Preprocessor(); 401 402 /// \brief Initialize the preprocessor, if the constructor did not already 403 /// perform the initialization. 404 /// 405 /// \param Target Information about the target. 406 void Initialize(const TargetInfo &Target); 407 408 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 409 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 410 411 const LangOptions &getLangOpts() const { return LangOpts; } 412 const TargetInfo &getTargetInfo() const { return *Target; } 413 FileManager &getFileManager() const { return FileMgr; } 414 SourceManager &getSourceManager() const { return SourceMgr; } 415 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 416 417 IdentifierTable &getIdentifierTable() { return Identifiers; } 418 SelectorTable &getSelectorTable() { return Selectors; } 419 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 420 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 421 422 void setPTHManager(PTHManager* pm); 423 424 PTHManager *getPTHManager() { return PTH.get(); } 425 426 void setExternalSource(ExternalPreprocessorSource *Source) { 427 ExternalSource = Source; 428 } 429 430 ExternalPreprocessorSource *getExternalSource() const { 431 return ExternalSource; 432 } 433 434 /// \brief Retrieve the module loader associated with this preprocessor. 435 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 436 437 /// SetCommentRetentionState - Control whether or not the preprocessor retains 438 /// comments in output. 439 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 440 this->KeepComments = KeepComments | KeepMacroComments; 441 this->KeepMacroComments = KeepMacroComments; 442 } 443 444 bool getCommentRetentionState() const { return KeepComments; } 445 446 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 447 bool getPragmasEnabled() const { return PragmasEnabled; } 448 449 void SetSuppressIncludeNotFoundError(bool Suppress) { 450 SuppressIncludeNotFoundError = Suppress; 451 } 452 453 bool GetSuppressIncludeNotFoundError() { 454 return SuppressIncludeNotFoundError; 455 } 456 457 /// isCurrentLexer - Return true if we are lexing directly from the specified 458 /// lexer. 459 bool isCurrentLexer(const PreprocessorLexer *L) const { 460 return CurPPLexer == L; 461 } 462 463 /// getCurrentLexer - Return the current lexer being lexed from. Note 464 /// that this ignores any potentially active macro expansions and _Pragma 465 /// expansions going on at the time. 466 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 467 468 /// getCurrentFileLexer - Return the current file lexer being lexed from. 469 /// Note that this ignores any potentially active macro expansions and _Pragma 470 /// expansions going on at the time. 471 PreprocessorLexer *getCurrentFileLexer() const; 472 473 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 474 /// Note that this class takes ownership of any PPCallbacks object given to 475 /// it. 476 PPCallbacks *getPPCallbacks() const { return Callbacks; } 477 void addPPCallbacks(PPCallbacks *C) { 478 if (Callbacks) 479 C = new PPChainedCallbacks(C, Callbacks); 480 Callbacks = C; 481 } 482 483 /// \brief Given an identifier, return the MacroInfo it is \#defined to 484 /// or null if it isn't \#define'd. 485 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 486 if (!II->hasMacroDefinition()) 487 return 0; 488 489 MacroInfo *MI = getMacroInfoHistory(II); 490 assert(MI->getUndefLoc().isInvalid() && "Macro is undefined!"); 491 return MI; 492 } 493 494 /// \brief Given an identifier, return the (probably #undef'd) MacroInfo 495 /// representing the most recent macro definition. One can iterate over all 496 /// previous macro definitions from it. This method should only be called for 497 /// identifiers that hadMacroDefinition(). 498 MacroInfo *getMacroInfoHistory(IdentifierInfo *II) const; 499 500 /// \brief Specify a macro for this identifier. 501 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI, 502 bool LoadedFromAST = false); 503 /// \brief Undefine a macro for this identifier. 504 void clearMacroInfo(IdentifierInfo *II); 505 506 /// macro_iterator/macro_begin/macro_end - This allows you to walk the macro 507 /// history table. Currently defined macros have 508 /// IdentifierInfo::hasMacroDefinition() set and an empty 509 /// MacroInfo::getUndefLoc() at the head of the list. 510 typedef llvm::DenseMap<IdentifierInfo*, 511 MacroInfo*>::const_iterator macro_iterator; 512 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 513 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 514 515 /// \brief Return the name of the macro defined before \p Loc that has 516 /// spelling \p Tokens. If there are multiple macros with same spelling, 517 /// return the last one defined. 518 StringRef getLastMacroWithSpelling(SourceLocation Loc, 519 ArrayRef<TokenValue> Tokens) const; 520 521 const std::string &getPredefines() const { return Predefines; } 522 /// setPredefines - Set the predefines for this Preprocessor. These 523 /// predefines are automatically injected when parsing the main file. 524 void setPredefines(const char *P) { Predefines = P; } 525 void setPredefines(const std::string &P) { Predefines = P; } 526 527 /// getIdentifierInfo - Return information about the specified preprocessor 528 /// identifier token. The version of this method that takes two character 529 /// pointers is preferred unless the identifier is already available as a 530 /// string (this avoids allocation and copying of memory to construct an 531 /// std::string). 532 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 533 return &Identifiers.get(Name); 534 } 535 536 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 537 /// If 'Namespace' is non-null, then it is a token required to exist on the 538 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 539 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 540 void AddPragmaHandler(PragmaHandler *Handler) { 541 AddPragmaHandler(StringRef(), Handler); 542 } 543 544 /// RemovePragmaHandler - Remove the specific pragma handler from 545 /// the preprocessor. If \p Namespace is non-null, then it should 546 /// be the namespace that \p Handler was added to. It is an error 547 /// to remove a handler that has not been registered. 548 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 549 void RemovePragmaHandler(PragmaHandler *Handler) { 550 RemovePragmaHandler(StringRef(), Handler); 551 } 552 553 /// \brief Add the specified comment handler to the preprocessor. 554 void addCommentHandler(CommentHandler *Handler); 555 556 /// \brief Remove the specified comment handler. 557 /// 558 /// It is an error to remove a handler that has not been registered. 559 void removeCommentHandler(CommentHandler *Handler); 560 561 /// \brief Set the code completion handler to the given object. 562 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 563 CodeComplete = &Handler; 564 } 565 566 /// \brief Retrieve the current code-completion handler. 567 CodeCompletionHandler *getCodeCompletionHandler() const { 568 return CodeComplete; 569 } 570 571 /// \brief Clear out the code completion handler. 572 void clearCodeCompletionHandler() { 573 CodeComplete = 0; 574 } 575 576 /// \brief Hook used by the lexer to invoke the "natural language" code 577 /// completion point. 578 void CodeCompleteNaturalLanguage(); 579 580 /// \brief Retrieve the preprocessing record, or NULL if there is no 581 /// preprocessing record. 582 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 583 584 /// \brief Create a new preprocessing record, which will keep track of 585 /// all macro expansions, macro definitions, etc. 586 void createPreprocessingRecord(bool RecordConditionalDirectives); 587 588 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 589 /// which implicitly adds the builtin defines etc. 590 void EnterMainSourceFile(); 591 592 /// EndSourceFile - Inform the preprocessor callbacks that processing is 593 /// complete. 594 void EndSourceFile(); 595 596 /// EnterSourceFile - Add a source file to the top of the include stack and 597 /// start lexing tokens from it instead of the current buffer. Emit an error 598 /// and don't enter the file on error. 599 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 600 SourceLocation Loc); 601 602 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 603 /// tokens from it instead of the current buffer. Args specifies the 604 /// tokens input to a function-like macro. 605 /// 606 /// ILEnd specifies the location of the ')' for a function-like macro or the 607 /// identifier for an object-like macro. 608 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 609 MacroArgs *Args); 610 611 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 612 /// which will cause the lexer to start returning the specified tokens. 613 /// 614 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 615 /// not be subject to further macro expansion. Otherwise, these tokens will 616 /// be re-macro-expanded when/if expansion is enabled. 617 /// 618 /// If OwnsTokens is false, this method assumes that the specified stream of 619 /// tokens has a permanent owner somewhere, so they do not need to be copied. 620 /// If it is true, it assumes the array of tokens is allocated with new[] and 621 /// must be freed. 622 /// 623 void EnterTokenStream(const Token *Toks, unsigned NumToks, 624 bool DisableMacroExpansion, bool OwnsTokens); 625 626 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 627 /// lexer stack. This should only be used in situations where the current 628 /// state of the top-of-stack lexer is known. 629 void RemoveTopOfLexerStack(); 630 631 /// EnableBacktrackAtThisPos - From the point that this method is called, and 632 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 633 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 634 /// make the Preprocessor re-lex the same tokens. 635 /// 636 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 637 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 638 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 639 /// 640 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 641 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 642 /// tokens will continue indefinitely. 643 /// 644 void EnableBacktrackAtThisPos(); 645 646 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 647 void CommitBacktrackedTokens(); 648 649 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 650 /// EnableBacktrackAtThisPos() was previously called. 651 void Backtrack(); 652 653 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 654 /// caching of tokens is on. 655 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 656 657 /// Lex - To lex a token from the preprocessor, just pull a token from the 658 /// current lexer or macro object. 659 void Lex(Token &Result) { 660 switch (CurLexerKind) { 661 case CLK_Lexer: CurLexer->Lex(Result); break; 662 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 663 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 664 case CLK_CachingLexer: CachingLex(Result); break; 665 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 666 } 667 } 668 669 void LexAfterModuleImport(Token &Result); 670 671 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 672 /// something not a comment. This is useful in -E -C mode where comments 673 /// would foul up preprocessor directive handling. 674 void LexNonComment(Token &Result) { 675 do 676 Lex(Result); 677 while (Result.getKind() == tok::comment); 678 } 679 680 /// LexUnexpandedToken - This is just like Lex, but this disables macro 681 /// expansion of identifier tokens. 682 void LexUnexpandedToken(Token &Result) { 683 // Disable macro expansion. 684 bool OldVal = DisableMacroExpansion; 685 DisableMacroExpansion = true; 686 // Lex the token. 687 Lex(Result); 688 689 // Reenable it. 690 DisableMacroExpansion = OldVal; 691 } 692 693 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 694 /// expansion of identifier tokens. 695 void LexUnexpandedNonComment(Token &Result) { 696 do 697 LexUnexpandedToken(Result); 698 while (Result.getKind() == tok::comment); 699 } 700 701 /// Disables macro expansion everywhere except for preprocessor directives. 702 void SetMacroExpansionOnlyInDirectives() { 703 DisableMacroExpansion = true; 704 MacroExpansionInDirectivesOverride = true; 705 } 706 707 /// LookAhead - This peeks ahead N tokens and returns that token without 708 /// consuming any tokens. LookAhead(0) returns the next token that would be 709 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 710 /// returns normal tokens after phase 5. As such, it is equivalent to using 711 /// 'Lex', not 'LexUnexpandedToken'. 712 const Token &LookAhead(unsigned N) { 713 if (CachedLexPos + N < CachedTokens.size()) 714 return CachedTokens[CachedLexPos+N]; 715 else 716 return PeekAhead(N+1); 717 } 718 719 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 720 /// this allows to revert a specific number of tokens. 721 /// Note that the number of tokens being reverted should be up to the last 722 /// backtrack position, not more. 723 void RevertCachedTokens(unsigned N) { 724 assert(isBacktrackEnabled() && 725 "Should only be called when tokens are cached for backtracking"); 726 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 727 && "Should revert tokens up to the last backtrack position, not more"); 728 assert(signed(CachedLexPos) - signed(N) >= 0 && 729 "Corrupted backtrack positions ?"); 730 CachedLexPos -= N; 731 } 732 733 /// EnterToken - Enters a token in the token stream to be lexed next. If 734 /// BackTrack() is called afterwards, the token will remain at the insertion 735 /// point. 736 void EnterToken(const Token &Tok) { 737 EnterCachingLexMode(); 738 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 739 } 740 741 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 742 /// tokens (because backtrack is enabled) it should replace the most recent 743 /// cached tokens with the given annotation token. This function has no effect 744 /// if backtracking is not enabled. 745 /// 746 /// Note that the use of this function is just for optimization; so that the 747 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 748 /// invoked. 749 void AnnotateCachedTokens(const Token &Tok) { 750 assert(Tok.isAnnotation() && "Expected annotation token"); 751 if (CachedLexPos != 0 && isBacktrackEnabled()) 752 AnnotatePreviousCachedTokens(Tok); 753 } 754 755 /// \brief Replace the last token with an annotation token. 756 /// 757 /// Like AnnotateCachedTokens(), this routine replaces an 758 /// already-parsed (and resolved) token with an annotation 759 /// token. However, this routine only replaces the last token with 760 /// the annotation token; it does not affect any other cached 761 /// tokens. This function has no effect if backtracking is not 762 /// enabled. 763 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 764 assert(Tok.isAnnotation() && "Expected annotation token"); 765 if (CachedLexPos != 0 && isBacktrackEnabled()) 766 CachedTokens[CachedLexPos-1] = Tok; 767 } 768 769 /// TypoCorrectToken - Update the current token to represent the provided 770 /// identifier, in order to cache an action performed by typo correction. 771 void TypoCorrectToken(const Token &Tok) { 772 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 773 if (CachedLexPos != 0 && isBacktrackEnabled()) 774 CachedTokens[CachedLexPos-1] = Tok; 775 } 776 777 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 778 /// CurTokenLexer pointers. 779 void recomputeCurLexerKind(); 780 781 /// \brief Returns true if incremental processing is enabled 782 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 783 784 /// \brief Enables the incremental processing 785 void enableIncrementalProcessing(bool value = true) { 786 IncrementalProcessing = value; 787 } 788 789 /// \brief Specify the point at which code-completion will be performed. 790 /// 791 /// \param File the file in which code completion should occur. If 792 /// this file is included multiple times, code-completion will 793 /// perform completion the first time it is included. If NULL, this 794 /// function clears out the code-completion point. 795 /// 796 /// \param Line the line at which code completion should occur 797 /// (1-based). 798 /// 799 /// \param Column the column at which code completion should occur 800 /// (1-based). 801 /// 802 /// \returns true if an error occurred, false otherwise. 803 bool SetCodeCompletionPoint(const FileEntry *File, 804 unsigned Line, unsigned Column); 805 806 /// \brief Determine if we are performing code completion. 807 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 808 809 /// \brief Returns the location of the code-completion point. 810 /// Returns an invalid location if code-completion is not enabled or the file 811 /// containing the code-completion point has not been lexed yet. 812 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 813 814 /// \brief Returns the start location of the file of code-completion point. 815 /// Returns an invalid location if code-completion is not enabled or the file 816 /// containing the code-completion point has not been lexed yet. 817 SourceLocation getCodeCompletionFileLoc() const { 818 return CodeCompletionFileLoc; 819 } 820 821 /// \brief Returns true if code-completion is enabled and we have hit the 822 /// code-completion point. 823 bool isCodeCompletionReached() const { return CodeCompletionReached; } 824 825 /// \brief Note that we hit the code-completion point. 826 void setCodeCompletionReached() { 827 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 828 CodeCompletionReached = true; 829 // Silence any diagnostics that occur after we hit the code-completion. 830 getDiagnostics().setSuppressAllDiagnostics(true); 831 } 832 833 /// \brief The location of the currently-active \#pragma clang 834 /// arc_cf_code_audited begin. Returns an invalid location if there 835 /// is no such pragma active. 836 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 837 return PragmaARCCFCodeAuditedLoc; 838 } 839 840 /// \brief Set the location of the currently-active \#pragma clang 841 /// arc_cf_code_audited begin. An invalid location ends the pragma. 842 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 843 PragmaARCCFCodeAuditedLoc = Loc; 844 } 845 846 /// \brief Instruct the preprocessor to skip part of the main source file. 847 /// 848 /// \param Bytes The number of bytes in the preamble to skip. 849 /// 850 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 851 /// start of a line. 852 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 853 SkipMainFilePreamble.first = Bytes; 854 SkipMainFilePreamble.second = StartOfLine; 855 } 856 857 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 858 /// the specified Token's location, translating the token's start 859 /// position in the current buffer into a SourcePosition object for rendering. 860 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 861 return Diags->Report(Loc, DiagID); 862 } 863 864 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 865 return Diags->Report(Tok.getLocation(), DiagID); 866 } 867 868 /// getSpelling() - Return the 'spelling' of the token at the given 869 /// location; does not go up to the spelling location or down to the 870 /// expansion location. 871 /// 872 /// \param buffer A buffer which will be used only if the token requires 873 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 874 /// \param invalid If non-null, will be set \c true if an error occurs. 875 StringRef getSpelling(SourceLocation loc, 876 SmallVectorImpl<char> &buffer, 877 bool *invalid = 0) const { 878 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 879 } 880 881 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 882 /// token is the characters used to represent the token in the source file 883 /// after trigraph expansion and escaped-newline folding. In particular, this 884 /// wants to get the true, uncanonicalized, spelling of things like digraphs 885 /// UCNs, etc. 886 /// 887 /// \param Invalid If non-null, will be set \c true if an error occurs. 888 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 889 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 890 } 891 892 /// getSpelling - This method is used to get the spelling of a token into a 893 /// preallocated buffer, instead of as an std::string. The caller is required 894 /// to allocate enough space for the token, which is guaranteed to be at least 895 /// Tok.getLength() bytes long. The length of the actual result is returned. 896 /// 897 /// Note that this method may do two possible things: it may either fill in 898 /// the buffer specified with characters, or it may *change the input pointer* 899 /// to point to a constant buffer with the data already in it (avoiding a 900 /// copy). The caller is not allowed to modify the returned buffer pointer 901 /// if an internal buffer is returned. 902 unsigned getSpelling(const Token &Tok, const char *&Buffer, 903 bool *Invalid = 0) const { 904 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 905 } 906 907 /// getSpelling - This method is used to get the spelling of a token into a 908 /// SmallVector. Note that the returned StringRef may not point to the 909 /// supplied buffer if a copy can be avoided. 910 StringRef getSpelling(const Token &Tok, 911 SmallVectorImpl<char> &Buffer, 912 bool *Invalid = 0) const; 913 914 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 915 /// with length 1, return the character. 916 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 917 bool *Invalid = 0) const { 918 assert(Tok.is(tok::numeric_constant) && 919 Tok.getLength() == 1 && "Called on unsupported token"); 920 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 921 922 // If the token is carrying a literal data pointer, just use it. 923 if (const char *D = Tok.getLiteralData()) 924 return *D; 925 926 // Otherwise, fall back on getCharacterData, which is slower, but always 927 // works. 928 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 929 } 930 931 /// \brief Retrieve the name of the immediate macro expansion. 932 /// 933 /// This routine starts from a source location, and finds the name of the macro 934 /// responsible for its immediate expansion. It looks through any intervening 935 /// macro argument expansions to compute this. It returns a StringRef which 936 /// refers to the SourceManager-owned buffer of the source where that macro 937 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 938 StringRef getImmediateMacroName(SourceLocation Loc) { 939 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 940 } 941 942 /// CreateString - Plop the specified string into a scratch buffer and set the 943 /// specified token's location and length to it. If specified, the source 944 /// location provides a location of the expansion point of the token. 945 void CreateString(StringRef Str, Token &Tok, 946 SourceLocation ExpansionLocStart = SourceLocation(), 947 SourceLocation ExpansionLocEnd = SourceLocation()); 948 949 /// \brief Computes the source location just past the end of the 950 /// token at this source location. 951 /// 952 /// This routine can be used to produce a source location that 953 /// points just past the end of the token referenced by \p Loc, and 954 /// is generally used when a diagnostic needs to point just after a 955 /// token where it expected something different that it received. If 956 /// the returned source location would not be meaningful (e.g., if 957 /// it points into a macro), this routine returns an invalid 958 /// source location. 959 /// 960 /// \param Offset an offset from the end of the token, where the source 961 /// location should refer to. The default offset (0) produces a source 962 /// location pointing just past the end of the token; an offset of 1 produces 963 /// a source location pointing to the last character in the token, etc. 964 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 965 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 966 } 967 968 /// \brief Returns true if the given MacroID location points at the first 969 /// token of the macro expansion. 970 /// 971 /// \param MacroBegin If non-null and function returns true, it is set to 972 /// begin location of the macro. 973 bool isAtStartOfMacroExpansion(SourceLocation loc, 974 SourceLocation *MacroBegin = 0) const { 975 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 976 MacroBegin); 977 } 978 979 /// \brief Returns true if the given MacroID location points at the last 980 /// token of the macro expansion. 981 /// 982 /// \param MacroEnd If non-null and function returns true, it is set to 983 /// end location of the macro. 984 bool isAtEndOfMacroExpansion(SourceLocation loc, 985 SourceLocation *MacroEnd = 0) const { 986 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 987 } 988 989 /// DumpToken - Print the token to stderr, used for debugging. 990 /// 991 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 992 void DumpLocation(SourceLocation Loc) const; 993 void DumpMacro(const MacroInfo &MI) const; 994 995 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 996 /// token, return a new location that specifies a character within the token. 997 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 998 unsigned Char) const { 999 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 1000 } 1001 1002 /// IncrementPasteCounter - Increment the counters for the number of token 1003 /// paste operations performed. If fast was specified, this is a 'fast paste' 1004 /// case we handled. 1005 /// 1006 void IncrementPasteCounter(bool isFast) { 1007 if (isFast) 1008 ++NumFastTokenPaste; 1009 else 1010 ++NumTokenPaste; 1011 } 1012 1013 void PrintStats(); 1014 1015 size_t getTotalMemory() const; 1016 1017 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 1018 /// comment (/##/) in microsoft mode, this method handles updating the current 1019 /// state, returning the token on the next source line. 1020 void HandleMicrosoftCommentPaste(Token &Tok); 1021 1022 //===--------------------------------------------------------------------===// 1023 // Preprocessor callback methods. These are invoked by a lexer as various 1024 // directives and events are found. 1025 1026 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1027 /// identifier information for the token and install it into the token, 1028 /// updating the token kind accordingly. 1029 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1030 1031private: 1032 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1033 1034public: 1035 1036 // SetPoisonReason - Call this function to indicate the reason for 1037 // poisoning an identifier. If that identifier is accessed while 1038 // poisoned, then this reason will be used instead of the default 1039 // "poisoned" diagnostic. 1040 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1041 1042 // HandlePoisonedIdentifier - Display reason for poisoned 1043 // identifier. 1044 void HandlePoisonedIdentifier(Token & Tok); 1045 1046 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1047 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1048 if(II->isPoisoned()) { 1049 HandlePoisonedIdentifier(Identifier); 1050 } 1051 } 1052 } 1053 1054private: 1055 /// Identifiers used for SEH handling in Borland. These are only 1056 /// allowed in particular circumstances 1057 // __except block 1058 IdentifierInfo *Ident__exception_code, 1059 *Ident___exception_code, 1060 *Ident_GetExceptionCode; 1061 // __except filter expression 1062 IdentifierInfo *Ident__exception_info, 1063 *Ident___exception_info, 1064 *Ident_GetExceptionInfo; 1065 // __finally 1066 IdentifierInfo *Ident__abnormal_termination, 1067 *Ident___abnormal_termination, 1068 *Ident_AbnormalTermination; 1069public: 1070 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1071 1072 /// HandleIdentifier - This callback is invoked when the lexer reads an 1073 /// identifier and has filled in the tokens IdentifierInfo member. This 1074 /// callback potentially macro expands it or turns it into a named token (like 1075 /// 'for'). 1076 void HandleIdentifier(Token &Identifier); 1077 1078 1079 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1080 /// the current file. This either returns the EOF token and returns true, or 1081 /// pops a level off the include stack and returns false, at which point the 1082 /// client should call lex again. 1083 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1084 1085 /// HandleEndOfTokenLexer - This callback is invoked when the current 1086 /// TokenLexer hits the end of its token stream. 1087 bool HandleEndOfTokenLexer(Token &Result); 1088 1089 /// HandleDirective - This callback is invoked when the lexer sees a # token 1090 /// at the start of a line. This consumes the directive, modifies the 1091 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1092 /// read is the correct one. 1093 void HandleDirective(Token &Result); 1094 1095 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1096 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1097 /// true, then we consider macros that expand to zero tokens as being ok. 1098 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1099 1100 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1101 /// current line until the tok::eod token is found. 1102 void DiscardUntilEndOfDirective(); 1103 1104 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1105 /// __DATE__ or __TIME__ in the file so far. 1106 bool SawDateOrTime() const { 1107 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1108 } 1109 unsigned getCounterValue() const { return CounterValue; } 1110 void setCounterValue(unsigned V) { CounterValue = V; } 1111 1112 /// \brief Retrieves the module that we're currently building, if any. 1113 Module *getCurrentModule(); 1114 1115 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1116 MacroInfo *AllocateMacroInfo(SourceLocation L); 1117 1118 /// \brief Allocate a new MacroInfo object which is clone of \p MI. 1119 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 1120 1121 /// \brief Turn the specified lexer token into a fully checked and spelled 1122 /// filename, e.g. as an operand of \#include. 1123 /// 1124 /// The caller is expected to provide a buffer that is large enough to hold 1125 /// the spelling of the filename, but is also expected to handle the case 1126 /// when this method decides to use a different buffer. 1127 /// 1128 /// \returns true if the input filename was in <>'s or false if it was 1129 /// in ""'s. 1130 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1131 1132 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1133 /// 1134 /// Returns null on failure. \p isAngled indicates whether the file 1135 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1136 const FileEntry *LookupFile(StringRef Filename, 1137 bool isAngled, const DirectoryLookup *FromDir, 1138 const DirectoryLookup *&CurDir, 1139 SmallVectorImpl<char> *SearchPath, 1140 SmallVectorImpl<char> *RelativePath, 1141 Module **SuggestedModule, 1142 bool SkipCache = false); 1143 1144 /// GetCurLookup - The DirectoryLookup structure used to find the current 1145 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1146 /// implement \#include_next and find directory-specific properties. 1147 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1148 1149 /// \brief Return true if we're in the top-level file, not in a \#include. 1150 bool isInPrimaryFile() const; 1151 1152 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1153 /// from a macro as multiple tokens, which need to be glued together. This 1154 /// occurs for code like: 1155 /// \code 1156 /// \#define FOO <x/y.h> 1157 /// \#include FOO 1158 /// \endcode 1159 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1160 /// 1161 /// This code concatenates and consumes tokens up to the '>' token. It 1162 /// returns false if the > was found, otherwise it returns true if it finds 1163 /// and consumes the EOD marker. 1164 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1165 SourceLocation &End); 1166 1167 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1168 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1169 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1170 1171private: 1172 1173 void PushIncludeMacroStack() { 1174 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1175 CurLexer.take(), 1176 CurPTHLexer.take(), 1177 CurPPLexer, 1178 CurTokenLexer.take(), 1179 CurDirLookup)); 1180 CurPPLexer = 0; 1181 } 1182 1183 void PopIncludeMacroStack() { 1184 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1185 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1186 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1187 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1188 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1189 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1190 IncludeMacroStack.pop_back(); 1191 } 1192 1193 /// \brief Allocate a new MacroInfo object. 1194 MacroInfo *AllocateMacroInfo(); 1195 1196 /// \brief Release the specified MacroInfo for re-use. 1197 /// 1198 /// This memory will be reused for allocating new MacroInfo objects. 1199 void ReleaseMacroInfo(MacroInfo* MI); 1200 1201 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1202 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1203 /// and discards the rest of the macro line if the macro name is invalid. 1204 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1205 1206 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1207 /// definition has just been read. Lex the rest of the arguments and the 1208 /// closing ), updating MI with what we learn and saving in LastTok the 1209 /// last token read. 1210 /// Return true if an error occurs parsing the arg list. 1211 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1212 1213 /// We just read a \#if or related directive and decided that the 1214 /// subsequent tokens are in the \#if'd out portion of the 1215 /// file. Lex the rest of the file, until we see an \#endif. If \p 1216 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1217 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1218 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1219 /// already seen one so a \#else directive is a duplicate. When this returns, 1220 /// the caller can lex the first valid token. 1221 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1222 bool FoundNonSkipPortion, bool FoundElse, 1223 SourceLocation ElseLoc = SourceLocation()); 1224 1225 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1226 void PTHSkipExcludedConditionalBlock(); 1227 1228 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1229 /// may occur after a #if or #elif directive and return it as a bool. If the 1230 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1231 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1232 1233 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1234 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1235 void RegisterBuiltinPragmas(); 1236 1237 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1238 void RegisterBuiltinMacros(); 1239 1240 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1241 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1242 /// the macro should not be expanded return true, otherwise return false. 1243 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 1244 1245 /// \brief Cache macro expanded tokens for TokenLexers. 1246 // 1247 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1248 /// going to lex in the cache and when it finishes the tokens are removed 1249 /// from the end of the cache. 1250 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1251 ArrayRef<Token> tokens); 1252 void removeCachedMacroExpandedTokensOfLastLexer(); 1253 friend void TokenLexer::ExpandFunctionArguments(); 1254 1255 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1256 /// lexed is a '('. If so, consume the token and return true, if not, this 1257 /// method should have no observable side-effect on the lexed tokens. 1258 bool isNextPPTokenLParen(); 1259 1260 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1261 /// invoked to read all of the formal arguments specified for the macro 1262 /// invocation. This returns null on error. 1263 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1264 SourceLocation &ExpansionEnd); 1265 1266 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1267 /// as a builtin macro, handle it and return the next token as 'Tok'. 1268 void ExpandBuiltinMacro(Token &Tok); 1269 1270 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1271 /// return the first token after the directive. The _Pragma token has just 1272 /// been read into 'Tok'. 1273 void Handle_Pragma(Token &Tok); 1274 1275 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1276 /// is not enclosed within a string literal. 1277 void HandleMicrosoft__pragma(Token &Tok); 1278 1279 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1280 /// start lexing tokens from it instead of the current buffer. 1281 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1282 1283 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1284 /// start getting tokens from it using the PTH cache. 1285 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1286 1287 /// IsFileLexer - Returns true if we are lexing from a file and not a 1288 /// pragma or a macro. 1289 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1290 return L ? !L->isPragmaLexer() : P != 0; 1291 } 1292 1293 static bool IsFileLexer(const IncludeStackInfo& I) { 1294 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1295 } 1296 1297 bool IsFileLexer() const { 1298 return IsFileLexer(CurLexer.get(), CurPPLexer); 1299 } 1300 1301 //===--------------------------------------------------------------------===// 1302 // Caching stuff. 1303 void CachingLex(Token &Result); 1304 bool InCachingLexMode() const { 1305 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1306 // that we are past EOF, not that we are in CachingLex mode. 1307 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1308 !IncludeMacroStack.empty(); 1309 } 1310 void EnterCachingLexMode(); 1311 void ExitCachingLexMode() { 1312 if (InCachingLexMode()) 1313 RemoveTopOfLexerStack(); 1314 } 1315 const Token &PeekAhead(unsigned N); 1316 void AnnotatePreviousCachedTokens(const Token &Tok); 1317 1318 //===--------------------------------------------------------------------===// 1319 /// Handle*Directive - implement the various preprocessor directives. These 1320 /// should side-effect the current preprocessor object so that the next call 1321 /// to Lex() will return the appropriate token next. 1322 void HandleLineDirective(Token &Tok); 1323 void HandleDigitDirective(Token &Tok); 1324 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1325 void HandleIdentSCCSDirective(Token &Tok); 1326 void HandleMacroPublicDirective(Token &Tok); 1327 void HandleMacroPrivateDirective(Token &Tok); 1328 1329 // File inclusion. 1330 void HandleIncludeDirective(SourceLocation HashLoc, 1331 Token &Tok, 1332 const DirectoryLookup *LookupFrom = 0, 1333 bool isImport = false); 1334 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1335 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1336 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1337 void HandleMicrosoftImportDirective(Token &Tok); 1338 1339 // Macro handling. 1340 void HandleDefineDirective(Token &Tok); 1341 void HandleUndefDirective(Token &Tok); 1342 1343 // Conditional Inclusion. 1344 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1345 bool ReadAnyTokensBeforeDirective); 1346 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1347 void HandleEndifDirective(Token &Tok); 1348 void HandleElseDirective(Token &Tok); 1349 void HandleElifDirective(Token &Tok); 1350 1351 // Pragmas. 1352 void HandlePragmaDirective(unsigned Introducer); 1353public: 1354 void HandlePragmaOnce(Token &OnceTok); 1355 void HandlePragmaMark(); 1356 void HandlePragmaPoison(Token &PoisonTok); 1357 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1358 void HandlePragmaDependency(Token &DependencyTok); 1359 void HandlePragmaComment(Token &CommentTok); 1360 void HandlePragmaMessage(Token &MessageTok); 1361 void HandlePragmaPushMacro(Token &Tok); 1362 void HandlePragmaPopMacro(Token &Tok); 1363 void HandlePragmaIncludeAlias(Token &Tok); 1364 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1365 1366 // Return true and store the first token only if any CommentHandler 1367 // has inserted some tokens and getCommentRetentionState() is false. 1368 bool HandleComment(Token &Token, SourceRange Comment); 1369 1370 /// \brief A macro is used, update information about macros that need unused 1371 /// warnings. 1372 void markMacroAsUsed(MacroInfo *MI); 1373}; 1374 1375/// \brief Abstract base class that describes a handler that will receive 1376/// source ranges for each of the comments encountered in the source file. 1377class CommentHandler { 1378public: 1379 virtual ~CommentHandler(); 1380 1381 // The handler shall return true if it has pushed any tokens 1382 // to be read using e.g. EnterToken or EnterTokenStream. 1383 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1384}; 1385 1386} // end namespace clang 1387 1388#endif 1389