Preprocessor.h revision fd61d6fe0aa9853f0577ca88a63901c3773e8101
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 the current build action is a preprocessing action. 164 bool PreprocessedOutput : 1; 165 166 /// \brief True if we are currently preprocessing a #if or #elif directive 167 bool ParsingIfOrElifDirective; 168 169 /// \brief True if we are pre-expanding macro arguments. 170 bool InMacroArgPreExpansion; 171 172 /// Identifiers - This is mapping/lookup information for all identifiers in 173 /// the program, including program keywords. 174 mutable IdentifierTable Identifiers; 175 176 /// Selectors - This table contains all the selectors in the program. Unlike 177 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 178 /// It is declared/expanded here because it's role/lifetime is 179 /// conceptually similar the IdentifierTable. In addition, the current control 180 /// flow (in clang::ParseAST()), make it convenient to put here. 181 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 182 /// the lifetime of the preprocessor. 183 SelectorTable Selectors; 184 185 /// BuiltinInfo - Information about builtins. 186 Builtin::Context BuiltinInfo; 187 188 /// PragmaHandlers - This tracks all of the pragmas that the client registered 189 /// with this preprocessor. 190 PragmaNamespace *PragmaHandlers; 191 192 /// \brief Tracks all of the comment handlers that the client registered 193 /// with this preprocessor. 194 std::vector<CommentHandler *> CommentHandlers; 195 196 /// \brief True if we want to ignore EOF token and continue later on (thus 197 /// avoid tearing the Lexer and etc. down). 198 bool IncrementalProcessing; 199 200 /// \brief The code-completion handler. 201 CodeCompletionHandler *CodeComplete; 202 203 /// \brief The file that we're performing code-completion for, if any. 204 const FileEntry *CodeCompletionFile; 205 206 /// \brief The offset in file for the code-completion point. 207 unsigned CodeCompletionOffset; 208 209 /// \brief The location for the code-completion point. This gets instantiated 210 /// when the CodeCompletionFile gets \#include'ed for preprocessing. 211 SourceLocation CodeCompletionLoc; 212 213 /// \brief The start location for the file of the code-completion point. 214 /// 215 /// This gets instantiated when the CodeCompletionFile gets \#include'ed 216 /// for preprocessing. 217 SourceLocation CodeCompletionFileLoc; 218 219 /// \brief The source location of the 'import' contextual keyword we just 220 /// lexed, if any. 221 SourceLocation ModuleImportLoc; 222 223 /// \brief The module import path that we're currently processing. 224 SmallVector<std::pair<IdentifierInfo *, SourceLocation>, 2> ModuleImportPath; 225 226 /// \brief Whether the module import expectes an identifier next. Otherwise, 227 /// it expects a '.' or ';'. 228 bool ModuleImportExpectsIdentifier; 229 230 /// \brief The source location of the currently-active 231 /// #pragma clang arc_cf_code_audited begin. 232 SourceLocation PragmaARCCFCodeAuditedLoc; 233 234 /// \brief True if we hit the code-completion point. 235 bool CodeCompletionReached; 236 237 /// \brief The number of bytes that we will initially skip when entering the 238 /// main file, which is used when loading a precompiled preamble, along 239 /// with a flag that indicates whether skipping this number of bytes will 240 /// place the lexer at the start of a line. 241 std::pair<unsigned, bool> SkipMainFilePreamble; 242 243 /// CurLexer - This is the current top of the stack that we're lexing from if 244 /// not expanding a macro and we are lexing directly from source code. 245 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 246 OwningPtr<Lexer> CurLexer; 247 248 /// CurPTHLexer - This is the current top of stack that we're lexing from if 249 /// not expanding from a macro and we are lexing from a PTH cache. 250 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 251 OwningPtr<PTHLexer> CurPTHLexer; 252 253 /// CurPPLexer - This is the current top of the stack what we're lexing from 254 /// if not expanding a macro. This is an alias for either CurLexer or 255 /// CurPTHLexer. 256 PreprocessorLexer *CurPPLexer; 257 258 /// CurLookup - The DirectoryLookup structure used to find the current 259 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 260 /// implement \#include_next and find directory-specific properties. 261 const DirectoryLookup *CurDirLookup; 262 263 /// CurTokenLexer - This is the current macro we are expanding, if we are 264 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 265 OwningPtr<TokenLexer> CurTokenLexer; 266 267 /// \brief The kind of lexer we're currently working with. 268 enum CurLexerKind { 269 CLK_Lexer, 270 CLK_PTHLexer, 271 CLK_TokenLexer, 272 CLK_CachingLexer, 273 CLK_LexAfterModuleImport 274 } CurLexerKind; 275 276 /// IncludeMacroStack - This keeps track of the stack of files currently 277 /// \#included, and macros currently being expanded from, not counting 278 /// CurLexer/CurTokenLexer. 279 struct IncludeStackInfo { 280 enum CurLexerKind CurLexerKind; 281 Lexer *TheLexer; 282 PTHLexer *ThePTHLexer; 283 PreprocessorLexer *ThePPLexer; 284 TokenLexer *TheTokenLexer; 285 const DirectoryLookup *TheDirLookup; 286 287 IncludeStackInfo(enum CurLexerKind K, Lexer *L, PTHLexer* P, 288 PreprocessorLexer* PPL, 289 TokenLexer* TL, const DirectoryLookup *D) 290 : CurLexerKind(K), TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), 291 TheTokenLexer(TL), TheDirLookup(D) {} 292 }; 293 std::vector<IncludeStackInfo> IncludeMacroStack; 294 295 /// Callbacks - These are actions invoked when some preprocessor activity is 296 /// encountered (e.g. a file is \#included, etc). 297 PPCallbacks *Callbacks; 298 299 /// \brief Listener whose actions are invoked when an entity in the 300 /// preprocessor (e.g., a macro) that was loaded from an AST file is 301 /// later mutated. 302 PPMutationListener *Listener; 303 304 struct MacroExpandsInfo { 305 Token Tok; 306 MacroDirective *MD; 307 SourceRange Range; 308 MacroExpandsInfo(Token Tok, MacroDirective *MD, SourceRange Range) 309 : Tok(Tok), MD(MD), Range(Range) { } 310 }; 311 SmallVector<MacroExpandsInfo, 2> DelayedMacroExpandsCallbacks; 312 313 /// Macros - For each IdentifierInfo that was associated with a macro, we 314 /// keep a mapping to the history of all macro definitions and #undefs in 315 /// the reverse order (the latest one is in the head of the list). 316 llvm::DenseMap<const IdentifierInfo*, MacroDirective*> Macros; 317 friend class ASTReader; 318 319 /// \brief Macros that we want to warn because they are not used at the end 320 /// of the translation unit; we store just their SourceLocations instead 321 /// something like MacroInfo*. The benefit of this is that when we are 322 /// deserializing from PCH, we don't need to deserialize identifier & macros 323 /// just so that we can report that they are unused, we just warn using 324 /// the SourceLocations of this set (that will be filled by the ASTReader). 325 /// We are using SmallPtrSet instead of a vector for faster removal. 326 typedef llvm::SmallPtrSet<SourceLocation, 32> WarnUnusedMacroLocsTy; 327 WarnUnusedMacroLocsTy WarnUnusedMacroLocs; 328 329 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 330 /// reused for quick allocation. 331 MacroArgs *MacroArgCache; 332 friend class MacroArgs; 333 334 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 335 /// push_macro directive, we keep a MacroInfo stack used to restore 336 /// previous macro value. 337 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 338 339 // Various statistics we track for performance analysis. 340 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 341 unsigned NumIf, NumElse, NumEndif; 342 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 343 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 344 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 345 unsigned NumSkipped; 346 347 /// Predefines - This string is the predefined macros that preprocessor 348 /// should use from the command line etc. 349 std::string Predefines; 350 351 /// \brief The file ID for the preprocessor predefines. 352 FileID PredefinesFileID; 353 354 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 355 enum { TokenLexerCacheSize = 8 }; 356 unsigned NumCachedTokenLexers; 357 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 358 359 /// \brief Keeps macro expanded tokens for TokenLexers. 360 // 361 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 362 /// going to lex in the cache and when it finishes the tokens are removed 363 /// from the end of the cache. 364 SmallVector<Token, 16> MacroExpandedTokens; 365 std::vector<std::pair<TokenLexer *, size_t> > MacroExpandingLexersStack; 366 367 /// \brief A record of the macro definitions and expansions that 368 /// occurred during preprocessing. 369 /// 370 /// This is an optional side structure that can be enabled with 371 /// \c createPreprocessingRecord() prior to preprocessing. 372 PreprocessingRecord *Record; 373 374private: // Cached tokens state. 375 typedef SmallVector<Token, 1> CachedTokensTy; 376 377 /// CachedTokens - Cached tokens are stored here when we do backtracking or 378 /// lookahead. They are "lexed" by the CachingLex() method. 379 CachedTokensTy CachedTokens; 380 381 /// CachedLexPos - The position of the cached token that CachingLex() should 382 /// "lex" next. If it points beyond the CachedTokens vector, it means that 383 /// a normal Lex() should be invoked. 384 CachedTokensTy::size_type CachedLexPos; 385 386 /// BacktrackPositions - Stack of backtrack positions, allowing nested 387 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 388 /// indicate where CachedLexPos should be set when the BackTrack() method is 389 /// invoked (at which point the last position is popped). 390 std::vector<CachedTokensTy::size_type> BacktrackPositions; 391 392 struct MacroInfoChain { 393 MacroInfo MI; 394 MacroInfoChain *Next; 395 MacroInfoChain *Prev; 396 }; 397 398 /// MacroInfos are managed as a chain for easy disposal. This is the head 399 /// of that list. 400 MacroInfoChain *MIChainHead; 401 402 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 403 /// allocation. 404 MacroInfoChain *MICache; 405 406public: 407 Preprocessor(IntrusiveRefCntPtr<PreprocessorOptions> PPOpts, 408 DiagnosticsEngine &diags, LangOptions &opts, 409 const TargetInfo *target, 410 SourceManager &SM, HeaderSearch &Headers, 411 ModuleLoader &TheModuleLoader, 412 IdentifierInfoLookup *IILookup = 0, 413 bool OwnsHeaderSearch = false, 414 bool DelayInitialization = false, 415 bool IncrProcessing = false); 416 417 ~Preprocessor(); 418 419 /// \brief Initialize the preprocessor, if the constructor did not already 420 /// perform the initialization. 421 /// 422 /// \param Target Information about the target. 423 void Initialize(const TargetInfo &Target); 424 425 /// \brief Retrieve the preprocessor options used to initialize this 426 /// preprocessor. 427 PreprocessorOptions &getPreprocessorOpts() const { return *PPOpts; } 428 429 DiagnosticsEngine &getDiagnostics() const { return *Diags; } 430 void setDiagnostics(DiagnosticsEngine &D) { Diags = &D; } 431 432 const LangOptions &getLangOpts() const { return LangOpts; } 433 const TargetInfo &getTargetInfo() const { return *Target; } 434 FileManager &getFileManager() const { return FileMgr; } 435 SourceManager &getSourceManager() const { return SourceMgr; } 436 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 437 438 IdentifierTable &getIdentifierTable() { return Identifiers; } 439 SelectorTable &getSelectorTable() { return Selectors; } 440 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 441 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 442 443 void setPTHManager(PTHManager* pm); 444 445 PTHManager *getPTHManager() { return PTH.get(); } 446 447 void setExternalSource(ExternalPreprocessorSource *Source) { 448 ExternalSource = Source; 449 } 450 451 ExternalPreprocessorSource *getExternalSource() const { 452 return ExternalSource; 453 } 454 455 /// \brief Retrieve the module loader associated with this preprocessor. 456 ModuleLoader &getModuleLoader() const { return TheModuleLoader; } 457 458 /// \brief True if we are currently preprocessing a #if or #elif directive 459 bool isParsingIfOrElifDirective() const { 460 return ParsingIfOrElifDirective; 461 } 462 463 /// SetCommentRetentionState - Control whether or not the preprocessor retains 464 /// comments in output. 465 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 466 this->KeepComments = KeepComments | KeepMacroComments; 467 this->KeepMacroComments = KeepMacroComments; 468 } 469 470 bool getCommentRetentionState() const { return KeepComments; } 471 472 void setPragmasEnabled(bool Enabled) { PragmasEnabled = Enabled; } 473 bool getPragmasEnabled() const { return PragmasEnabled; } 474 475 void SetSuppressIncludeNotFoundError(bool Suppress) { 476 SuppressIncludeNotFoundError = Suppress; 477 } 478 479 bool GetSuppressIncludeNotFoundError() { 480 return SuppressIncludeNotFoundError; 481 } 482 483 /// Sets whether the preprocessor is responsible for producing output or if 484 /// it is producing tokens to be consumed by Parse and Sema. 485 void setPreprocessedOutput(bool IsPreprocessedOutput) { 486 PreprocessedOutput = IsPreprocessedOutput; 487 } 488 489 /// Returns true if the preprocessor is responsible for generating output, 490 /// false if it is producing tokens to be consumed by Parse and Sema. 491 bool isPreprocessedOutput() const { return PreprocessedOutput; } 492 493 /// isCurrentLexer - Return true if we are lexing directly from the specified 494 /// lexer. 495 bool isCurrentLexer(const PreprocessorLexer *L) const { 496 return CurPPLexer == L; 497 } 498 499 /// getCurrentLexer - Return the current lexer being lexed from. Note 500 /// that this ignores any potentially active macro expansions and _Pragma 501 /// expansions going on at the time. 502 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 503 504 /// getCurrentFileLexer - Return the current file lexer being lexed from. 505 /// Note that this ignores any potentially active macro expansions and _Pragma 506 /// expansions going on at the time. 507 PreprocessorLexer *getCurrentFileLexer() const; 508 509 /// \brief Returns the file ID for the preprocessor predefines. 510 FileID getPredefinesFileID() const { return PredefinesFileID; } 511 512 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 513 /// Note that this class takes ownership of any PPCallbacks object given to 514 /// it. 515 PPCallbacks *getPPCallbacks() const { return Callbacks; } 516 void addPPCallbacks(PPCallbacks *C) { 517 if (Callbacks) 518 C = new PPChainedCallbacks(C, Callbacks); 519 Callbacks = C; 520 } 521 522 /// \brief Attach an preprocessor mutation listener to the preprocessor. 523 /// 524 /// The preprocessor mutation listener provides the ability to track 525 /// modifications to the preprocessor entities committed after they were 526 /// initially created. 527 void setPPMutationListener(PPMutationListener *Listener) { 528 this->Listener = Listener; 529 } 530 531 /// \brief Retrieve a pointer to the preprocessor mutation listener 532 /// associated with this preprocessor, if any. 533 PPMutationListener *getPPMutationListener() const { return Listener; } 534 535 /// \brief Given an identifier, return the MacroInfo it is \#defined to 536 /// or null if it isn't \#define'd. 537 MacroDirective *getMacroDirective(IdentifierInfo *II) const { 538 if (!II->hasMacroDefinition()) 539 return 0; 540 541 MacroDirective *MD = getMacroDirectiveHistory(II); 542 assert(MD->getUndefLoc().isInvalid() && "Macro is undefined!"); 543 return MD; 544 } 545 546 const MacroInfo *getMacroInfo(IdentifierInfo *II) const { 547 return const_cast<Preprocessor*>(this)->getMacroInfo(II); 548 } 549 550 MacroInfo *getMacroInfo(IdentifierInfo *II) { 551 if (MacroDirective *MD = getMacroDirective(II)) 552 return MD->getInfo(); 553 return 0; 554 } 555 556 /// \brief Given an identifier, return the (probably #undef'd) MacroInfo 557 /// representing the most recent macro definition. One can iterate over all 558 /// previous macro definitions from it. This method should only be called for 559 /// identifiers that hadMacroDefinition(). 560 MacroDirective *getMacroDirectiveHistory(const IdentifierInfo *II) const; 561 562 /// \brief Specify a macro for this identifier. 563 void setMacroDirective(IdentifierInfo *II, MacroDirective *MD); 564 MacroDirective *setMacroDirective(IdentifierInfo *II, MacroInfo *MI, 565 SourceLocation Loc, bool isImported) { 566 MacroDirective *MD = AllocateMacroDirective(MI, Loc, isImported); 567 setMacroDirective(II, MD); 568 return MD; 569 } 570 MacroDirective *setMacroDirective(IdentifierInfo *II, MacroInfo *MI) { 571 return setMacroDirective(II, MI, MI->getDefinitionLoc(), false); 572 } 573 /// \brief Set a MacroDirective that was loaded from a PCH file. 574 void setLoadedMacroDirective(IdentifierInfo *II, MacroDirective *MD); 575 /// \brief Add a MacroInfo that was loaded from an AST file. 576 void addLoadedMacroInfo(IdentifierInfo *II, MacroDirective *MD, 577 MacroDirective *Hint = 0); 578 /// \brief Make the given MacroInfo, that was loaded from an AST file and 579 /// previously hidden, visible. 580 void makeLoadedMacroInfoVisible(IdentifierInfo *II, MacroDirective *MD); 581 /// \brief Undefine a macro for this identifier. 582 void clearMacroInfo(IdentifierInfo *II); 583 584 /// macro_iterator/macro_begin/macro_end - This allows you to walk the macro 585 /// history table. Currently defined macros have 586 /// IdentifierInfo::hasMacroDefinition() set and an empty 587 /// MacroInfo::getUndefLoc() at the head of the list. 588 typedef llvm::DenseMap<const IdentifierInfo *, 589 MacroDirective*>::const_iterator macro_iterator; 590 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 591 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 592 593 /// \brief Return the name of the macro defined before \p Loc that has 594 /// spelling \p Tokens. If there are multiple macros with same spelling, 595 /// return the last one defined. 596 StringRef getLastMacroWithSpelling(SourceLocation Loc, 597 ArrayRef<TokenValue> Tokens) const; 598 599 const std::string &getPredefines() const { return Predefines; } 600 /// setPredefines - Set the predefines for this Preprocessor. These 601 /// predefines are automatically injected when parsing the main file. 602 void setPredefines(const char *P) { Predefines = P; } 603 void setPredefines(const std::string &P) { Predefines = P; } 604 605 /// Return information about the specified preprocessor 606 /// identifier token. 607 IdentifierInfo *getIdentifierInfo(StringRef Name) const { 608 return &Identifiers.get(Name); 609 } 610 611 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 612 /// If 'Namespace' is non-null, then it is a token required to exist on the 613 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 614 void AddPragmaHandler(StringRef Namespace, PragmaHandler *Handler); 615 void AddPragmaHandler(PragmaHandler *Handler) { 616 AddPragmaHandler(StringRef(), Handler); 617 } 618 619 /// RemovePragmaHandler - Remove the specific pragma handler from 620 /// the preprocessor. If \p Namespace is non-null, then it should 621 /// be the namespace that \p Handler was added to. It is an error 622 /// to remove a handler that has not been registered. 623 void RemovePragmaHandler(StringRef Namespace, PragmaHandler *Handler); 624 void RemovePragmaHandler(PragmaHandler *Handler) { 625 RemovePragmaHandler(StringRef(), Handler); 626 } 627 628 /// \brief Add the specified comment handler to the preprocessor. 629 void addCommentHandler(CommentHandler *Handler); 630 631 /// \brief Remove the specified comment handler. 632 /// 633 /// It is an error to remove a handler that has not been registered. 634 void removeCommentHandler(CommentHandler *Handler); 635 636 /// \brief Set the code completion handler to the given object. 637 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 638 CodeComplete = &Handler; 639 } 640 641 /// \brief Retrieve the current code-completion handler. 642 CodeCompletionHandler *getCodeCompletionHandler() const { 643 return CodeComplete; 644 } 645 646 /// \brief Clear out the code completion handler. 647 void clearCodeCompletionHandler() { 648 CodeComplete = 0; 649 } 650 651 /// \brief Hook used by the lexer to invoke the "natural language" code 652 /// completion point. 653 void CodeCompleteNaturalLanguage(); 654 655 /// \brief Retrieve the preprocessing record, or NULL if there is no 656 /// preprocessing record. 657 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 658 659 /// \brief Create a new preprocessing record, which will keep track of 660 /// all macro expansions, macro definitions, etc. 661 void createPreprocessingRecord(); 662 663 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 664 /// which implicitly adds the builtin defines etc. 665 void EnterMainSourceFile(); 666 667 /// EndSourceFile - Inform the preprocessor callbacks that processing is 668 /// complete. 669 void EndSourceFile(); 670 671 /// EnterSourceFile - Add a source file to the top of the include stack and 672 /// start lexing tokens from it instead of the current buffer. Emit an error 673 /// and don't enter the file on error. 674 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 675 SourceLocation Loc); 676 677 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 678 /// tokens from it instead of the current buffer. Args specifies the 679 /// tokens input to a function-like macro. 680 /// 681 /// ILEnd specifies the location of the ')' for a function-like macro or the 682 /// identifier for an object-like macro. 683 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroInfo *Macro, 684 MacroArgs *Args); 685 686 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 687 /// which will cause the lexer to start returning the specified tokens. 688 /// 689 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 690 /// not be subject to further macro expansion. Otherwise, these tokens will 691 /// be re-macro-expanded when/if expansion is enabled. 692 /// 693 /// If OwnsTokens is false, this method assumes that the specified stream of 694 /// tokens has a permanent owner somewhere, so they do not need to be copied. 695 /// If it is true, it assumes the array of tokens is allocated with new[] and 696 /// must be freed. 697 /// 698 void EnterTokenStream(const Token *Toks, unsigned NumToks, 699 bool DisableMacroExpansion, bool OwnsTokens); 700 701 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 702 /// lexer stack. This should only be used in situations where the current 703 /// state of the top-of-stack lexer is known. 704 void RemoveTopOfLexerStack(); 705 706 /// EnableBacktrackAtThisPos - From the point that this method is called, and 707 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 708 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 709 /// make the Preprocessor re-lex the same tokens. 710 /// 711 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 712 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 713 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 714 /// 715 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 716 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 717 /// tokens will continue indefinitely. 718 /// 719 void EnableBacktrackAtThisPos(); 720 721 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 722 void CommitBacktrackedTokens(); 723 724 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 725 /// EnableBacktrackAtThisPos() was previously called. 726 void Backtrack(); 727 728 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 729 /// caching of tokens is on. 730 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 731 732 /// Lex - To lex a token from the preprocessor, just pull a token from the 733 /// current lexer or macro object. 734 void Lex(Token &Result) { 735 switch (CurLexerKind) { 736 case CLK_Lexer: CurLexer->Lex(Result); break; 737 case CLK_PTHLexer: CurPTHLexer->Lex(Result); break; 738 case CLK_TokenLexer: CurTokenLexer->Lex(Result); break; 739 case CLK_CachingLexer: CachingLex(Result); break; 740 case CLK_LexAfterModuleImport: LexAfterModuleImport(Result); break; 741 } 742 } 743 744 void LexAfterModuleImport(Token &Result); 745 746 /// \brief Lex a string literal, which may be the concatenation of multiple 747 /// string literals and may even come from macro expansion. 748 /// \returns true on success, false if a error diagnostic has been generated. 749 bool LexStringLiteral(Token &Result, std::string &String, 750 const char *DiagnosticTag, bool AllowMacroExpansion) { 751 if (AllowMacroExpansion) 752 Lex(Result); 753 else 754 LexUnexpandedToken(Result); 755 return FinishLexStringLiteral(Result, String, DiagnosticTag, 756 AllowMacroExpansion); 757 } 758 759 /// \brief Complete the lexing of a string literal where the first token has 760 /// already been lexed (see LexStringLiteral). 761 bool FinishLexStringLiteral(Token &Result, std::string &String, 762 const char *DiagnosticTag, 763 bool AllowMacroExpansion); 764 765 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 766 /// something not a comment. This is useful in -E -C mode where comments 767 /// would foul up preprocessor directive handling. 768 void LexNonComment(Token &Result) { 769 do 770 Lex(Result); 771 while (Result.getKind() == tok::comment); 772 } 773 774 /// LexUnexpandedToken - This is just like Lex, but this disables macro 775 /// expansion of identifier tokens. 776 void LexUnexpandedToken(Token &Result) { 777 // Disable macro expansion. 778 bool OldVal = DisableMacroExpansion; 779 DisableMacroExpansion = true; 780 // Lex the token. 781 Lex(Result); 782 783 // Reenable it. 784 DisableMacroExpansion = OldVal; 785 } 786 787 /// LexUnexpandedNonComment - Like LexNonComment, but this disables macro 788 /// expansion of identifier tokens. 789 void LexUnexpandedNonComment(Token &Result) { 790 do 791 LexUnexpandedToken(Result); 792 while (Result.getKind() == tok::comment); 793 } 794 795 /// Disables macro expansion everywhere except for preprocessor directives. 796 void SetMacroExpansionOnlyInDirectives() { 797 DisableMacroExpansion = true; 798 MacroExpansionInDirectivesOverride = true; 799 } 800 801 /// LookAhead - This peeks ahead N tokens and returns that token without 802 /// consuming any tokens. LookAhead(0) returns the next token that would be 803 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 804 /// returns normal tokens after phase 5. As such, it is equivalent to using 805 /// 'Lex', not 'LexUnexpandedToken'. 806 const Token &LookAhead(unsigned N) { 807 if (CachedLexPos + N < CachedTokens.size()) 808 return CachedTokens[CachedLexPos+N]; 809 else 810 return PeekAhead(N+1); 811 } 812 813 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 814 /// this allows to revert a specific number of tokens. 815 /// Note that the number of tokens being reverted should be up to the last 816 /// backtrack position, not more. 817 void RevertCachedTokens(unsigned N) { 818 assert(isBacktrackEnabled() && 819 "Should only be called when tokens are cached for backtracking"); 820 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 821 && "Should revert tokens up to the last backtrack position, not more"); 822 assert(signed(CachedLexPos) - signed(N) >= 0 && 823 "Corrupted backtrack positions ?"); 824 CachedLexPos -= N; 825 } 826 827 /// EnterToken - Enters a token in the token stream to be lexed next. If 828 /// BackTrack() is called afterwards, the token will remain at the insertion 829 /// point. 830 void EnterToken(const Token &Tok) { 831 EnterCachingLexMode(); 832 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 833 } 834 835 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 836 /// tokens (because backtrack is enabled) it should replace the most recent 837 /// cached tokens with the given annotation token. This function has no effect 838 /// if backtracking is not enabled. 839 /// 840 /// Note that the use of this function is just for optimization; so that the 841 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 842 /// invoked. 843 void AnnotateCachedTokens(const Token &Tok) { 844 assert(Tok.isAnnotation() && "Expected annotation token"); 845 if (CachedLexPos != 0 && isBacktrackEnabled()) 846 AnnotatePreviousCachedTokens(Tok); 847 } 848 849 /// \brief Replace the last token with an annotation token. 850 /// 851 /// Like AnnotateCachedTokens(), this routine replaces an 852 /// already-parsed (and resolved) token with an annotation 853 /// token. However, this routine only replaces the last token with 854 /// the annotation token; it does not affect any other cached 855 /// tokens. This function has no effect if backtracking is not 856 /// enabled. 857 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 858 assert(Tok.isAnnotation() && "Expected annotation token"); 859 if (CachedLexPos != 0 && isBacktrackEnabled()) 860 CachedTokens[CachedLexPos-1] = Tok; 861 } 862 863 /// TypoCorrectToken - Update the current token to represent the provided 864 /// identifier, in order to cache an action performed by typo correction. 865 void TypoCorrectToken(const Token &Tok) { 866 assert(Tok.getIdentifierInfo() && "Expected identifier token"); 867 if (CachedLexPos != 0 && isBacktrackEnabled()) 868 CachedTokens[CachedLexPos-1] = Tok; 869 } 870 871 /// \brief Recompute the current lexer kind based on the CurLexer/CurPTHLexer/ 872 /// CurTokenLexer pointers. 873 void recomputeCurLexerKind(); 874 875 /// \brief Returns true if incremental processing is enabled 876 bool isIncrementalProcessingEnabled() const { return IncrementalProcessing; } 877 878 /// \brief Enables the incremental processing 879 void enableIncrementalProcessing(bool value = true) { 880 IncrementalProcessing = value; 881 } 882 883 /// \brief Specify the point at which code-completion will be performed. 884 /// 885 /// \param File the file in which code completion should occur. If 886 /// this file is included multiple times, code-completion will 887 /// perform completion the first time it is included. If NULL, this 888 /// function clears out the code-completion point. 889 /// 890 /// \param Line the line at which code completion should occur 891 /// (1-based). 892 /// 893 /// \param Column the column at which code completion should occur 894 /// (1-based). 895 /// 896 /// \returns true if an error occurred, false otherwise. 897 bool SetCodeCompletionPoint(const FileEntry *File, 898 unsigned Line, unsigned Column); 899 900 /// \brief Determine if we are performing code completion. 901 bool isCodeCompletionEnabled() const { return CodeCompletionFile != 0; } 902 903 /// \brief Returns the location of the code-completion point. 904 /// Returns an invalid location if code-completion is not enabled or the file 905 /// containing the code-completion point has not been lexed yet. 906 SourceLocation getCodeCompletionLoc() const { return CodeCompletionLoc; } 907 908 /// \brief Returns the start location of the file of code-completion point. 909 /// Returns an invalid location if code-completion is not enabled or the file 910 /// containing the code-completion point has not been lexed yet. 911 SourceLocation getCodeCompletionFileLoc() const { 912 return CodeCompletionFileLoc; 913 } 914 915 /// \brief Returns true if code-completion is enabled and we have hit the 916 /// code-completion point. 917 bool isCodeCompletionReached() const { return CodeCompletionReached; } 918 919 /// \brief Note that we hit the code-completion point. 920 void setCodeCompletionReached() { 921 assert(isCodeCompletionEnabled() && "Code-completion not enabled!"); 922 CodeCompletionReached = true; 923 // Silence any diagnostics that occur after we hit the code-completion. 924 getDiagnostics().setSuppressAllDiagnostics(true); 925 } 926 927 /// \brief The location of the currently-active \#pragma clang 928 /// arc_cf_code_audited begin. Returns an invalid location if there 929 /// is no such pragma active. 930 SourceLocation getPragmaARCCFCodeAuditedLoc() const { 931 return PragmaARCCFCodeAuditedLoc; 932 } 933 934 /// \brief Set the location of the currently-active \#pragma clang 935 /// arc_cf_code_audited begin. An invalid location ends the pragma. 936 void setPragmaARCCFCodeAuditedLoc(SourceLocation Loc) { 937 PragmaARCCFCodeAuditedLoc = Loc; 938 } 939 940 /// \brief Instruct the preprocessor to skip part of the main source file. 941 /// 942 /// \param Bytes The number of bytes in the preamble to skip. 943 /// 944 /// \param StartOfLine Whether skipping these bytes puts the lexer at the 945 /// start of a line. 946 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 947 SkipMainFilePreamble.first = Bytes; 948 SkipMainFilePreamble.second = StartOfLine; 949 } 950 951 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 952 /// the specified Token's location, translating the token's start 953 /// position in the current buffer into a SourcePosition object for rendering. 954 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) const { 955 return Diags->Report(Loc, DiagID); 956 } 957 958 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) const { 959 return Diags->Report(Tok.getLocation(), DiagID); 960 } 961 962 /// getSpelling() - Return the 'spelling' of the token at the given 963 /// location; does not go up to the spelling location or down to the 964 /// expansion location. 965 /// 966 /// \param buffer A buffer which will be used only if the token requires 967 /// "cleaning", e.g. if it contains trigraphs or escaped newlines 968 /// \param invalid If non-null, will be set \c true if an error occurs. 969 StringRef getSpelling(SourceLocation loc, 970 SmallVectorImpl<char> &buffer, 971 bool *invalid = 0) const { 972 return Lexer::getSpelling(loc, buffer, SourceMgr, LangOpts, invalid); 973 } 974 975 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 976 /// token is the characters used to represent the token in the source file 977 /// after trigraph expansion and escaped-newline folding. In particular, this 978 /// wants to get the true, uncanonicalized, spelling of things like digraphs 979 /// UCNs, etc. 980 /// 981 /// \param Invalid If non-null, will be set \c true if an error occurs. 982 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 983 return Lexer::getSpelling(Tok, SourceMgr, LangOpts, Invalid); 984 } 985 986 /// getSpelling - This method is used to get the spelling of a token into a 987 /// preallocated buffer, instead of as an std::string. The caller is required 988 /// to allocate enough space for the token, which is guaranteed to be at least 989 /// Tok.getLength() bytes long. The length of the actual result is returned. 990 /// 991 /// Note that this method may do two possible things: it may either fill in 992 /// the buffer specified with characters, or it may *change the input pointer* 993 /// to point to a constant buffer with the data already in it (avoiding a 994 /// copy). The caller is not allowed to modify the returned buffer pointer 995 /// if an internal buffer is returned. 996 unsigned getSpelling(const Token &Tok, const char *&Buffer, 997 bool *Invalid = 0) const { 998 return Lexer::getSpelling(Tok, Buffer, SourceMgr, LangOpts, Invalid); 999 } 1000 1001 /// getSpelling - This method is used to get the spelling of a token into a 1002 /// SmallVector. Note that the returned StringRef may not point to the 1003 /// supplied buffer if a copy can be avoided. 1004 StringRef getSpelling(const Token &Tok, 1005 SmallVectorImpl<char> &Buffer, 1006 bool *Invalid = 0) const; 1007 1008 /// \brief Relex the token at the specified location. 1009 /// \returns true if there was a failure, false on success. 1010 bool getRawToken(SourceLocation Loc, Token &Result) { 1011 return Lexer::getRawToken(Loc, Result, SourceMgr, LangOpts); 1012 } 1013 1014 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 1015 /// with length 1, return the character. 1016 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 1017 bool *Invalid = 0) const { 1018 assert(Tok.is(tok::numeric_constant) && 1019 Tok.getLength() == 1 && "Called on unsupported token"); 1020 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 1021 1022 // If the token is carrying a literal data pointer, just use it. 1023 if (const char *D = Tok.getLiteralData()) 1024 return *D; 1025 1026 // Otherwise, fall back on getCharacterData, which is slower, but always 1027 // works. 1028 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 1029 } 1030 1031 /// \brief Retrieve the name of the immediate macro expansion. 1032 /// 1033 /// This routine starts from a source location, and finds the name of the macro 1034 /// responsible for its immediate expansion. It looks through any intervening 1035 /// macro argument expansions to compute this. It returns a StringRef which 1036 /// refers to the SourceManager-owned buffer of the source where that macro 1037 /// name is spelled. Thus, the result shouldn't out-live the SourceManager. 1038 StringRef getImmediateMacroName(SourceLocation Loc) { 1039 return Lexer::getImmediateMacroName(Loc, SourceMgr, getLangOpts()); 1040 } 1041 1042 /// CreateString - Plop the specified string into a scratch buffer and set the 1043 /// specified token's location and length to it. If specified, the source 1044 /// location provides a location of the expansion point of the token. 1045 void CreateString(StringRef Str, Token &Tok, 1046 SourceLocation ExpansionLocStart = SourceLocation(), 1047 SourceLocation ExpansionLocEnd = SourceLocation()); 1048 1049 /// \brief Computes the source location just past the end of the 1050 /// token at this source location. 1051 /// 1052 /// This routine can be used to produce a source location that 1053 /// points just past the end of the token referenced by \p Loc, and 1054 /// is generally used when a diagnostic needs to point just after a 1055 /// token where it expected something different that it received. If 1056 /// the returned source location would not be meaningful (e.g., if 1057 /// it points into a macro), this routine returns an invalid 1058 /// source location. 1059 /// 1060 /// \param Offset an offset from the end of the token, where the source 1061 /// location should refer to. The default offset (0) produces a source 1062 /// location pointing just past the end of the token; an offset of 1 produces 1063 /// a source location pointing to the last character in the token, etc. 1064 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 1065 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, LangOpts); 1066 } 1067 1068 /// \brief Returns true if the given MacroID location points at the first 1069 /// token of the macro expansion. 1070 /// 1071 /// \param MacroBegin If non-null and function returns true, it is set to 1072 /// begin location of the macro. 1073 bool isAtStartOfMacroExpansion(SourceLocation loc, 1074 SourceLocation *MacroBegin = 0) const { 1075 return Lexer::isAtStartOfMacroExpansion(loc, SourceMgr, LangOpts, 1076 MacroBegin); 1077 } 1078 1079 /// \brief Returns true if the given MacroID location points at the last 1080 /// token of the macro expansion. 1081 /// 1082 /// \param MacroEnd If non-null and function returns true, it is set to 1083 /// end location of the macro. 1084 bool isAtEndOfMacroExpansion(SourceLocation loc, 1085 SourceLocation *MacroEnd = 0) const { 1086 return Lexer::isAtEndOfMacroExpansion(loc, SourceMgr, LangOpts, MacroEnd); 1087 } 1088 1089 /// DumpToken - Print the token to stderr, used for debugging. 1090 /// 1091 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 1092 void DumpLocation(SourceLocation Loc) const; 1093 void DumpMacro(const MacroInfo &MI) const; 1094 1095 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 1096 /// token, return a new location that specifies a character within the token. 1097 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 1098 unsigned Char) const { 1099 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, LangOpts); 1100 } 1101 1102 /// IncrementPasteCounter - Increment the counters for the number of token 1103 /// paste operations performed. If fast was specified, this is a 'fast paste' 1104 /// case we handled. 1105 /// 1106 void IncrementPasteCounter(bool isFast) { 1107 if (isFast) 1108 ++NumFastTokenPaste; 1109 else 1110 ++NumTokenPaste; 1111 } 1112 1113 void PrintStats(); 1114 1115 size_t getTotalMemory() const; 1116 1117 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 1118 /// comment (/##/) in microsoft mode, this method handles updating the current 1119 /// state, returning the token on the next source line. 1120 void HandleMicrosoftCommentPaste(Token &Tok); 1121 1122 //===--------------------------------------------------------------------===// 1123 // Preprocessor callback methods. These are invoked by a lexer as various 1124 // directives and events are found. 1125 1126 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the 1127 /// identifier information for the token and install it into the token, 1128 /// updating the token kind accordingly. 1129 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier) const; 1130 1131private: 1132 llvm::DenseMap<IdentifierInfo*,unsigned> PoisonReasons; 1133 1134public: 1135 1136 // SetPoisonReason - Call this function to indicate the reason for 1137 // poisoning an identifier. If that identifier is accessed while 1138 // poisoned, then this reason will be used instead of the default 1139 // "poisoned" diagnostic. 1140 void SetPoisonReason(IdentifierInfo *II, unsigned DiagID); 1141 1142 // HandlePoisonedIdentifier - Display reason for poisoned 1143 // identifier. 1144 void HandlePoisonedIdentifier(Token & Tok); 1145 1146 void MaybeHandlePoisonedIdentifier(Token & Identifier) { 1147 if(IdentifierInfo * II = Identifier.getIdentifierInfo()) { 1148 if(II->isPoisoned()) { 1149 HandlePoisonedIdentifier(Identifier); 1150 } 1151 } 1152 } 1153 1154private: 1155 /// Identifiers used for SEH handling in Borland. These are only 1156 /// allowed in particular circumstances 1157 // __except block 1158 IdentifierInfo *Ident__exception_code, 1159 *Ident___exception_code, 1160 *Ident_GetExceptionCode; 1161 // __except filter expression 1162 IdentifierInfo *Ident__exception_info, 1163 *Ident___exception_info, 1164 *Ident_GetExceptionInfo; 1165 // __finally 1166 IdentifierInfo *Ident__abnormal_termination, 1167 *Ident___abnormal_termination, 1168 *Ident_AbnormalTermination; 1169public: 1170 void PoisonSEHIdentifiers(bool Poison = true); // Borland 1171 1172 /// HandleIdentifier - This callback is invoked when the lexer reads an 1173 /// identifier and has filled in the tokens IdentifierInfo member. This 1174 /// callback potentially macro expands it or turns it into a named token (like 1175 /// 'for'). 1176 void HandleIdentifier(Token &Identifier); 1177 1178 1179 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 1180 /// the current file. This either returns the EOF token and returns true, or 1181 /// pops a level off the include stack and returns false, at which point the 1182 /// client should call lex again. 1183 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 1184 1185 /// HandleEndOfTokenLexer - This callback is invoked when the current 1186 /// TokenLexer hits the end of its token stream. 1187 bool HandleEndOfTokenLexer(Token &Result); 1188 1189 /// HandleDirective - This callback is invoked when the lexer sees a # token 1190 /// at the start of a line. This consumes the directive, modifies the 1191 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 1192 /// read is the correct one. 1193 void HandleDirective(Token &Result); 1194 1195 /// CheckEndOfDirective - Ensure that the next token is a tok::eod token. If 1196 /// not, emit a diagnostic and consume up until the eod. If EnableMacros is 1197 /// true, then we consider macros that expand to zero tokens as being ok. 1198 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 1199 1200 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 1201 /// current line until the tok::eod token is found. 1202 void DiscardUntilEndOfDirective(); 1203 1204 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 1205 /// __DATE__ or __TIME__ in the file so far. 1206 bool SawDateOrTime() const { 1207 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 1208 } 1209 unsigned getCounterValue() const { return CounterValue; } 1210 void setCounterValue(unsigned V) { CounterValue = V; } 1211 1212 /// \brief Retrieves the module that we're currently building, if any. 1213 Module *getCurrentModule(); 1214 1215 /// \brief Allocate a new MacroInfo object with the provided SourceLocation. 1216 MacroInfo *AllocateMacroInfo(SourceLocation L); 1217 1218 /// \brief Allocate a new MacroInfo object loaded from an AST file. 1219 MacroInfo *AllocateDeserializedMacroInfo(SourceLocation L, 1220 unsigned SubModuleID); 1221 1222 /// \brief Turn the specified lexer token into a fully checked and spelled 1223 /// filename, e.g. as an operand of \#include. 1224 /// 1225 /// The caller is expected to provide a buffer that is large enough to hold 1226 /// the spelling of the filename, but is also expected to handle the case 1227 /// when this method decides to use a different buffer. 1228 /// 1229 /// \returns true if the input filename was in <>'s or false if it was 1230 /// in ""'s. 1231 bool GetIncludeFilenameSpelling(SourceLocation Loc,StringRef &Filename); 1232 1233 /// \brief Given a "foo" or \<foo> reference, look up the indicated file. 1234 /// 1235 /// Returns null on failure. \p isAngled indicates whether the file 1236 /// reference is for system \#include's or not (i.e. using <> instead of ""). 1237 const FileEntry *LookupFile(StringRef Filename, 1238 bool isAngled, const DirectoryLookup *FromDir, 1239 const DirectoryLookup *&CurDir, 1240 SmallVectorImpl<char> *SearchPath, 1241 SmallVectorImpl<char> *RelativePath, 1242 Module **SuggestedModule, 1243 bool SkipCache = false); 1244 1245 /// GetCurLookup - The DirectoryLookup structure used to find the current 1246 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 1247 /// implement \#include_next and find directory-specific properties. 1248 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 1249 1250 /// \brief Return true if we're in the top-level file, not in a \#include. 1251 bool isInPrimaryFile() const; 1252 1253 /// ConcatenateIncludeName - Handle cases where the \#include name is expanded 1254 /// from a macro as multiple tokens, which need to be glued together. This 1255 /// occurs for code like: 1256 /// \code 1257 /// \#define FOO <x/y.h> 1258 /// \#include FOO 1259 /// \endcode 1260 /// because in this case, "<x/y.h>" is returned as 7 tokens, not one. 1261 /// 1262 /// This code concatenates and consumes tokens up to the '>' token. It 1263 /// returns false if the > was found, otherwise it returns true if it finds 1264 /// and consumes the EOD marker. 1265 bool ConcatenateIncludeName(SmallString<128> &FilenameBuffer, 1266 SourceLocation &End); 1267 1268 /// LexOnOffSwitch - Lex an on-off-switch (C99 6.10.6p2) and verify that it is 1269 /// followed by EOD. Return true if the token is not a valid on-off-switch. 1270 bool LexOnOffSwitch(tok::OnOffSwitch &OOS); 1271 1272private: 1273 1274 void PushIncludeMacroStack() { 1275 IncludeMacroStack.push_back(IncludeStackInfo(CurLexerKind, 1276 CurLexer.take(), 1277 CurPTHLexer.take(), 1278 CurPPLexer, 1279 CurTokenLexer.take(), 1280 CurDirLookup)); 1281 CurPPLexer = 0; 1282 } 1283 1284 void PopIncludeMacroStack() { 1285 CurLexer.reset(IncludeMacroStack.back().TheLexer); 1286 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 1287 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 1288 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 1289 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 1290 CurLexerKind = IncludeMacroStack.back().CurLexerKind; 1291 IncludeMacroStack.pop_back(); 1292 } 1293 1294 /// \brief Allocate a new MacroInfo object. 1295 MacroInfo *AllocateMacroInfo(); 1296 1297 MacroDirective *AllocateMacroDirective(MacroInfo *MI, SourceLocation Loc, 1298 bool isImported); 1299 1300 /// \brief Release the specified MacroInfo for re-use. 1301 /// 1302 /// This memory will be reused for allocating new MacroInfo objects. 1303 void ReleaseMacroInfo(MacroInfo* MI); 1304 1305 /// ReadMacroName - Lex and validate a macro name, which occurs after a 1306 /// \#define or \#undef. This emits a diagnostic, sets the token kind to eod, 1307 /// and discards the rest of the macro line if the macro name is invalid. 1308 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 1309 1310 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 1311 /// definition has just been read. Lex the rest of the arguments and the 1312 /// closing ), updating MI with what we learn and saving in LastTok the 1313 /// last token read. 1314 /// Return true if an error occurs parsing the arg list. 1315 bool ReadMacroDefinitionArgList(MacroInfo *MI, Token& LastTok); 1316 1317 /// We just read a \#if or related directive and decided that the 1318 /// subsequent tokens are in the \#if'd out portion of the 1319 /// file. Lex the rest of the file, until we see an \#endif. If \p 1320 /// FoundNonSkipPortion is true, then we have already emitted code for part of 1321 /// this \#if directive, so \#else/\#elif blocks should never be entered. If 1322 /// \p FoundElse is false, then \#else directives are ok, if not, then we have 1323 /// already seen one so a \#else directive is a duplicate. When this returns, 1324 /// the caller can lex the first valid token. 1325 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 1326 bool FoundNonSkipPortion, bool FoundElse, 1327 SourceLocation ElseLoc = SourceLocation()); 1328 1329 /// \brief A fast PTH version of SkipExcludedConditionalBlock. 1330 void PTHSkipExcludedConditionalBlock(); 1331 1332 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 1333 /// may occur after a #if or #elif directive and return it as a bool. If the 1334 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 1335 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 1336 1337 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 1338 /// \#pragma GCC poison/system_header/dependency and \#pragma once. 1339 void RegisterBuiltinPragmas(); 1340 1341 /// \brief Register builtin macros such as __LINE__ with the identifier table. 1342 void RegisterBuiltinMacros(); 1343 1344 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 1345 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 1346 /// the macro should not be expanded return true, otherwise return false. 1347 bool HandleMacroExpandedIdentifier(Token &Tok, MacroDirective *MD); 1348 1349 /// \brief Cache macro expanded tokens for TokenLexers. 1350 // 1351 /// Works like a stack; a TokenLexer adds the macro expanded tokens that is 1352 /// going to lex in the cache and when it finishes the tokens are removed 1353 /// from the end of the cache. 1354 Token *cacheMacroExpandedTokens(TokenLexer *tokLexer, 1355 ArrayRef<Token> tokens); 1356 void removeCachedMacroExpandedTokensOfLastLexer(); 1357 friend void TokenLexer::ExpandFunctionArguments(); 1358 1359 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 1360 /// lexed is a '('. If so, consume the token and return true, if not, this 1361 /// method should have no observable side-effect on the lexed tokens. 1362 bool isNextPPTokenLParen(); 1363 1364 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 1365 /// invoked to read all of the formal arguments specified for the macro 1366 /// invocation. This returns null on error. 1367 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 1368 SourceLocation &ExpansionEnd); 1369 1370 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 1371 /// as a builtin macro, handle it and return the next token as 'Tok'. 1372 void ExpandBuiltinMacro(Token &Tok); 1373 1374 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 1375 /// return the first token after the directive. The _Pragma token has just 1376 /// been read into 'Tok'. 1377 void Handle_Pragma(Token &Tok); 1378 1379 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 1380 /// is not enclosed within a string literal. 1381 void HandleMicrosoft__pragma(Token &Tok); 1382 1383 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 1384 /// start lexing tokens from it instead of the current buffer. 1385 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 1386 1387 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 1388 /// start getting tokens from it using the PTH cache. 1389 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 1390 1391 /// \brief Set the file ID for the preprocessor predefines. 1392 void setPredefinesFileID(FileID FID) { 1393 assert(PredefinesFileID.isInvalid() && "PredefinesFileID already set!"); 1394 PredefinesFileID = FID; 1395 } 1396 1397 /// IsFileLexer - Returns true if we are lexing from a file and not a 1398 /// pragma or a macro. 1399 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 1400 return L ? !L->isPragmaLexer() : P != 0; 1401 } 1402 1403 static bool IsFileLexer(const IncludeStackInfo& I) { 1404 return IsFileLexer(I.TheLexer, I.ThePPLexer); 1405 } 1406 1407 bool IsFileLexer() const { 1408 return IsFileLexer(CurLexer.get(), CurPPLexer); 1409 } 1410 1411 //===--------------------------------------------------------------------===// 1412 // Caching stuff. 1413 void CachingLex(Token &Result); 1414 bool InCachingLexMode() const { 1415 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 1416 // that we are past EOF, not that we are in CachingLex mode. 1417 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 1418 !IncludeMacroStack.empty(); 1419 } 1420 void EnterCachingLexMode(); 1421 void ExitCachingLexMode() { 1422 if (InCachingLexMode()) 1423 RemoveTopOfLexerStack(); 1424 } 1425 const Token &PeekAhead(unsigned N); 1426 void AnnotatePreviousCachedTokens(const Token &Tok); 1427 1428 //===--------------------------------------------------------------------===// 1429 /// Handle*Directive - implement the various preprocessor directives. These 1430 /// should side-effect the current preprocessor object so that the next call 1431 /// to Lex() will return the appropriate token next. 1432 void HandleLineDirective(Token &Tok); 1433 void HandleDigitDirective(Token &Tok); 1434 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 1435 void HandleIdentSCCSDirective(Token &Tok); 1436 void HandleMacroPublicDirective(Token &Tok); 1437 void HandleMacroPrivateDirective(Token &Tok); 1438 1439 // File inclusion. 1440 void HandleIncludeDirective(SourceLocation HashLoc, 1441 Token &Tok, 1442 const DirectoryLookup *LookupFrom = 0, 1443 bool isImport = false); 1444 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 1445 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 1446 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 1447 void HandleMicrosoftImportDirective(Token &Tok); 1448 1449 // Macro handling. 1450 void HandleDefineDirective(Token &Tok); 1451 void HandleUndefDirective(Token &Tok); 1452 void UndefineMacro(IdentifierInfo *II, MacroDirective *MD, 1453 SourceLocation UndefLoc); 1454 1455 // Conditional Inclusion. 1456 void HandleIfdefDirective(Token &Tok, bool isIfndef, 1457 bool ReadAnyTokensBeforeDirective); 1458 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 1459 void HandleEndifDirective(Token &Tok); 1460 void HandleElseDirective(Token &Tok); 1461 void HandleElifDirective(Token &Tok); 1462 1463 // Pragmas. 1464 void HandlePragmaDirective(unsigned Introducer); 1465public: 1466 void HandlePragmaOnce(Token &OnceTok); 1467 void HandlePragmaMark(); 1468 void HandlePragmaPoison(Token &PoisonTok); 1469 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1470 void HandlePragmaDependency(Token &DependencyTok); 1471 void HandlePragmaComment(Token &CommentTok); 1472 void HandlePragmaMessage(Token &MessageTok); 1473 void HandlePragmaPushMacro(Token &Tok); 1474 void HandlePragmaPopMacro(Token &Tok); 1475 void HandlePragmaIncludeAlias(Token &Tok); 1476 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1477 1478 // Return true and store the first token only if any CommentHandler 1479 // has inserted some tokens and getCommentRetentionState() is false. 1480 bool HandleComment(Token &Token, SourceRange Comment); 1481 1482 /// \brief A macro is used, update information about macros that need unused 1483 /// warnings. 1484 void markMacroAsUsed(MacroInfo *MI); 1485}; 1486 1487/// \brief Abstract base class that describes a handler that will receive 1488/// source ranges for each of the comments encountered in the source file. 1489class CommentHandler { 1490public: 1491 virtual ~CommentHandler(); 1492 1493 // The handler shall return true if it has pushed any tokens 1494 // to be read using e.g. EnterToken or EnterTokenStream. 1495 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1496}; 1497 1498} // end namespace clang 1499 1500#endif 1501