Preprocessor.h revision ab82f41b217ce588a9456c0b4411f219d3ed0df8
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/Lexer.h" 18#include "clang/Lex/PTHLexer.h" 19#include "clang/Lex/PPCallbacks.h" 20#include "clang/Lex/TokenLexer.h" 21#include "clang/Lex/PTHManager.h" 22#include "clang/Basic/Diagnostic.h" 23#include "clang/Basic/IdentifierTable.h" 24#include "clang/Basic/SourceLocation.h" 25#include "llvm/ADT/DenseMap.h" 26#include "llvm/ADT/OwningPtr.h" 27#include "llvm/Support/Allocator.h" 28 29namespace clang { 30 31class SourceManager; 32class FileManager; 33class FileEntry; 34class HeaderSearch; 35class PragmaNamespace; 36class PragmaHandler; 37class ScratchBuffer; 38class TargetInfo; 39class PPCallbacks; 40class DirectoryLookup; 41 42/// Preprocessor - This object engages in a tight little dance with the lexer to 43/// efficiently preprocess tokens. Lexers know only about tokens within a 44/// single source file, and don't know anything about preprocessor-level issues 45/// like the #include stack, token expansion, etc. 46/// 47class Preprocessor { 48 Diagnostic *Diags; 49 const LangOptions &Features; 50 TargetInfo &Target; 51 FileManager &FileMgr; 52 SourceManager &SourceMgr; 53 ScratchBuffer *ScratchBuf; 54 HeaderSearch &HeaderInfo; 55 56 /// PTH - An optional PTHManager object used for getting tokens from 57 /// a token cache rather than lexing the original source file. 58 llvm::OwningPtr<PTHManager> PTH; 59 60 /// BP - A BumpPtrAllocator object used to quickly allocate and release 61 /// objects internal to the Preprocessor. 62 llvm::BumpPtrAllocator BP; 63 64 /// Identifiers for builtin macros and other builtins. 65 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 66 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 67 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 68 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 69 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 70 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 71 IdentifierInfo *Ident_Pragma, *Ident__VA_ARGS__; // _Pragma, __VA_ARGS__ 72 73 SourceLocation DATELoc, TIMELoc; 74 unsigned CounterValue; // Next __COUNTER__ value. 75 76 enum { 77 /// MaxIncludeStackDepth - Maximum depth of #includes. 78 MaxAllowedIncludeStackDepth = 200 79 }; 80 81 // State that is set before the preprocessor begins. 82 bool KeepComments : 1; 83 bool KeepMacroComments : 1; 84 85 // State that changes while the preprocessor runs: 86 bool DisableMacroExpansion : 1; // True if macro expansion is disabled. 87 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 88 89 /// Identifiers - This is mapping/lookup information for all identifiers in 90 /// the program, including program keywords. 91 IdentifierTable Identifiers; 92 93 /// Selectors - This table contains all the selectors in the program. Unlike 94 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 95 /// It is declared/instantiated here because it's role/lifetime is 96 /// conceptually similar the IdentifierTable. In addition, the current control 97 /// flow (in clang::ParseAST()), make it convenient to put here. 98 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 99 /// the lifetime fo the preprocessor. 100 SelectorTable Selectors; 101 102 /// PragmaHandlers - This tracks all of the pragmas that the client registered 103 /// with this preprocessor. 104 PragmaNamespace *PragmaHandlers; 105 106 /// CurLexer - This is the current top of the stack that we're lexing from if 107 /// not expanding a macro and we are lexing directly from source code. 108 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 109 llvm::OwningPtr<Lexer> CurLexer; 110 111 /// CurPTHLexer - This is the current top of stack that we're lexing from if 112 /// not expanding from a macro and we are lexing from a PTH cache. 113 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 114 llvm::OwningPtr<PTHLexer> CurPTHLexer; 115 116 /// CurPPLexer - This is the current top of the stack what we're lexing from 117 /// if not expanding a macro. This is an alias for either CurLexer or 118 /// CurPTHLexer. 119 PreprocessorLexer* CurPPLexer; 120 121 /// CurLookup - The DirectoryLookup structure used to find the current 122 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 123 /// implement #include_next and find directory-specific properties. 124 const DirectoryLookup *CurDirLookup; 125 126 /// CurTokenLexer - This is the current macro we are expanding, if we are 127 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 128 llvm::OwningPtr<TokenLexer> CurTokenLexer; 129 130 /// IncludeMacroStack - This keeps track of the stack of files currently 131 /// #included, and macros currently being expanded from, not counting 132 /// CurLexer/CurTokenLexer. 133 struct IncludeStackInfo { 134 Lexer *TheLexer; 135 PTHLexer *ThePTHLexer; 136 PreprocessorLexer *ThePPLexer; 137 TokenLexer *TheTokenLexer; 138 const DirectoryLookup *TheDirLookup; 139 140 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 141 TokenLexer* TL, const DirectoryLookup *D) 142 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 143 TheDirLookup(D) {} 144 }; 145 std::vector<IncludeStackInfo> IncludeMacroStack; 146 147 /// Callbacks - These are actions invoked when some preprocessor activity is 148 /// encountered (e.g. a file is #included, etc). 149 PPCallbacks *Callbacks; 150 151 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 152 /// to the actual definition of the macro. 153 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 154 155 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 156 /// allocation. 157 std::vector<MacroInfo*> MICache; 158 159 // Various statistics we track for performance analysis. 160 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 161 unsigned NumIf, NumElse, NumEndif; 162 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 163 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 164 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 165 unsigned NumSkipped; 166 167 /// Predefines - This string is the predefined macros that preprocessor 168 /// should use from the command line etc. 169 std::string Predefines; 170 171 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 172 enum { TokenLexerCacheSize = 8 }; 173 unsigned NumCachedTokenLexers; 174 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 175 176private: // Cached tokens state. 177 typedef std::vector<Token> CachedTokensTy; 178 179 /// CachedTokens - Cached tokens are stored here when we do backtracking or 180 /// lookahead. They are "lexed" by the CachingLex() method. 181 CachedTokensTy CachedTokens; 182 183 /// CachedLexPos - The position of the cached token that CachingLex() should 184 /// "lex" next. If it points beyond the CachedTokens vector, it means that 185 /// a normal Lex() should be invoked. 186 CachedTokensTy::size_type CachedLexPos; 187 188 /// BacktrackPositions - Stack of backtrack positions, allowing nested 189 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 190 /// indicate where CachedLexPos should be set when the BackTrack() method is 191 /// invoked (at which point the last position is popped). 192 std::vector<CachedTokensTy::size_type> BacktrackPositions; 193 194public: 195 Preprocessor(Diagnostic &diags, const LangOptions &opts, TargetInfo &target, 196 SourceManager &SM, HeaderSearch &Headers, 197 IdentifierInfoLookup* IILookup = 0); 198 199 ~Preprocessor(); 200 201 Diagnostic &getDiagnostics() const { return *Diags; } 202 void setDiagnostics(Diagnostic &D) { Diags = &D; } 203 204 205 const LangOptions &getLangOptions() const { return Features; } 206 TargetInfo &getTargetInfo() const { return Target; } 207 FileManager &getFileManager() const { return FileMgr; } 208 SourceManager &getSourceManager() const { return SourceMgr; } 209 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 210 211 IdentifierTable &getIdentifierTable() { return Identifiers; } 212 SelectorTable &getSelectorTable() { return Selectors; } 213 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 214 215 void setPTHManager(PTHManager* pm); 216 217 PTHManager *getPTHManager() { return PTH.get(); } 218 219 /// SetCommentRetentionState - Control whether or not the preprocessor retains 220 /// comments in output. 221 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 222 this->KeepComments = KeepComments | KeepMacroComments; 223 this->KeepMacroComments = KeepMacroComments; 224 } 225 226 bool getCommentRetentionState() const { return KeepComments; } 227 228 /// isCurrentLexer - Return true if we are lexing directly from the specified 229 /// lexer. 230 bool isCurrentLexer(const PreprocessorLexer *L) const { 231 return CurPPLexer == L; 232 } 233 234 /// getCurrentLexer - Return the current file lexer being lexed from. Note 235 /// that this ignores any potentially active macro expansions and _Pragma 236 /// expansions going on at the time. 237 PreprocessorLexer *getCurrentFileLexer() const; 238 239 /// getPPCallbacks/setPPCallbacks - Accessors for preprocessor callbacks. 240 /// Note that this class takes ownership of any PPCallbacks object given to 241 /// it. 242 PPCallbacks *getPPCallbacks() const { return Callbacks; } 243 void setPPCallbacks(PPCallbacks *C) { 244 delete Callbacks; 245 Callbacks = C; 246 } 247 248 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 249 /// or null if it isn't #define'd. 250 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 251 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 252 } 253 254 /// setMacroInfo - Specify a macro for this identifier. 255 /// 256 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 257 258 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 259 /// state of the macro table. This visits every currently-defined macro. 260 typedef llvm::DenseMap<IdentifierInfo*, 261 MacroInfo*>::const_iterator macro_iterator; 262 macro_iterator macro_begin() const { return Macros.begin(); } 263 macro_iterator macro_end() const { return Macros.end(); } 264 265 266 267 const std::string &getPredefines() const { return Predefines; } 268 /// setPredefines - Set the predefines for this Preprocessor. These 269 /// predefines are automatically injected when parsing the main file. 270 void setPredefines(const char *P) { Predefines = P; } 271 void setPredefines(const std::string &P) { Predefines = P; } 272 273 /// getIdentifierInfo - Return information about the specified preprocessor 274 /// identifier token. The version of this method that takes two character 275 /// pointers is preferred unless the identifier is already available as a 276 /// string (this avoids allocation and copying of memory to construct an 277 /// std::string). 278 IdentifierInfo *getIdentifierInfo(const char *NameStart, 279 const char *NameEnd) { 280 return &Identifiers.get(NameStart, NameEnd); 281 } 282 IdentifierInfo *getIdentifierInfo(const char *NameStr) { 283 return getIdentifierInfo(NameStr, NameStr+strlen(NameStr)); 284 } 285 286 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 287 /// If 'Namespace' is non-null, then it is a token required to exist on the 288 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 289 void AddPragmaHandler(const char *Namespace, PragmaHandler *Handler); 290 291 /// RemovePragmaHandler - Remove the specific pragma handler from 292 /// the preprocessor. If \arg Namespace is non-null, then it should 293 /// be the namespace that \arg Handler was added to. It is an error 294 /// to remove a handler that has not been registered. 295 void RemovePragmaHandler(const char *Namespace, PragmaHandler *Handler); 296 297 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 298 /// which implicitly adds the builtin defines etc. 299 void EnterMainSourceFile(); 300 301 /// EnterSourceFile - Add a source file to the top of the include stack and 302 /// start lexing tokens from it instead of the current buffer. If isMainFile 303 /// is true, this is the main file for the translation unit. 304 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir); 305 306 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 307 /// tokens from it instead of the current buffer. Args specifies the 308 /// tokens input to a function-like macro. 309 /// 310 /// ILEnd specifies the location of the ')' for a function-like macro or the 311 /// identifier for an object-like macro. 312 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 313 314 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 315 /// which will cause the lexer to start returning the specified tokens. 316 /// 317 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 318 /// not be subject to further macro expansion. Otherwise, these tokens will 319 /// be re-macro-expanded when/if expansion is enabled. 320 /// 321 /// If OwnsTokens is false, this method assumes that the specified stream of 322 /// tokens has a permanent owner somewhere, so they do not need to be copied. 323 /// If it is true, it assumes the array of tokens is allocated with new[] and 324 /// must be freed. 325 /// 326 void EnterTokenStream(const Token *Toks, unsigned NumToks, 327 bool DisableMacroExpansion, bool OwnsTokens); 328 329 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 330 /// lexer stack. This should only be used in situations where the current 331 /// state of the top-of-stack lexer is known. 332 void RemoveTopOfLexerStack(); 333 334 /// EnableBacktrackAtThisPos - From the point that this method is called, and 335 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 336 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 337 /// make the Preprocessor re-lex the same tokens. 338 /// 339 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 340 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 341 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 342 /// 343 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 344 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 345 /// tokens will continue indefinitely. 346 /// 347 void EnableBacktrackAtThisPos(); 348 349 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 350 void CommitBacktrackedTokens(); 351 352 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 353 /// EnableBacktrackAtThisPos() was previously called. 354 void Backtrack(); 355 356 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 357 /// caching of tokens is on. 358 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 359 360 /// Lex - To lex a token from the preprocessor, just pull a token from the 361 /// current lexer or macro object. 362 void Lex(Token &Result) { 363 if (CurLexer) 364 CurLexer->Lex(Result); 365 else if (CurPTHLexer) 366 CurPTHLexer->Lex(Result); 367 else if (CurTokenLexer) 368 CurTokenLexer->Lex(Result); 369 else 370 CachingLex(Result); 371 } 372 373 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 374 /// something not a comment. This is useful in -E -C mode where comments 375 /// would foul up preprocessor directive handling. 376 void LexNonComment(Token &Result) { 377 do 378 Lex(Result); 379 while (Result.getKind() == tok::comment); 380 } 381 382 /// LexUnexpandedToken - This is just like Lex, but this disables macro 383 /// expansion of identifier tokens. 384 void LexUnexpandedToken(Token &Result) { 385 // Disable macro expansion. 386 bool OldVal = DisableMacroExpansion; 387 DisableMacroExpansion = true; 388 // Lex the token. 389 Lex(Result); 390 391 // Reenable it. 392 DisableMacroExpansion = OldVal; 393 } 394 395 /// LookAhead - This peeks ahead N tokens and returns that token without 396 /// consuming any tokens. LookAhead(0) returns the next token that would be 397 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 398 /// returns normal tokens after phase 5. As such, it is equivalent to using 399 /// 'Lex', not 'LexUnexpandedToken'. 400 const Token &LookAhead(unsigned N) { 401 if (CachedLexPos + N < CachedTokens.size()) 402 return CachedTokens[CachedLexPos+N]; 403 else 404 return PeekAhead(N+1); 405 } 406 407 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 408 /// this allows to revert a specific number of tokens. 409 /// Note that the number of tokens being reverted should be up to the last 410 /// backtrack position, not more. 411 void RevertCachedTokens(unsigned N) { 412 assert(isBacktrackEnabled() && 413 "Should only be called when tokens are cached for backtracking"); 414 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 415 && "Should revert tokens up to the last backtrack position, not more"); 416 assert(signed(CachedLexPos) - signed(N) >= 0 && 417 "Corrupted backtrack positions ?"); 418 CachedLexPos -= N; 419 } 420 421 /// EnterToken - Enters a token in the token stream to be lexed next. If 422 /// BackTrack() is called afterwards, the token will remain at the insertion 423 /// point. 424 void EnterToken(const Token &Tok) { 425 EnterCachingLexMode(); 426 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 427 } 428 429 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 430 /// tokens (because backtrack is enabled) it should replace the most recent 431 /// cached tokens with the given annotation token. This function has no effect 432 /// if backtracking is not enabled. 433 /// 434 /// Note that the use of this function is just for optimization; so that the 435 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 436 /// invoked. 437 void AnnotateCachedTokens(const Token &Tok) { 438 assert(Tok.isAnnotation() && "Expected annotation token"); 439 if (CachedLexPos != 0 && isBacktrackEnabled()) 440 AnnotatePreviousCachedTokens(Tok); 441 } 442 443 /// \brief Replace the last token with an annotation token. 444 /// 445 /// Like AnnotateCachedTokens(), this routine replaces an 446 /// already-parsed (and resolved) token with an annotation 447 /// token. However, this routine only replaces the last token with 448 /// the annotation token; it does not affect any other cached 449 /// tokens. This function has no effect if backtracking is not 450 /// enabled. 451 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 452 assert(Tok.isAnnotation() && "Expected annotation token"); 453 if (CachedLexPos != 0 && isBacktrackEnabled()) 454 CachedTokens[CachedLexPos-1] = Tok; 455 } 456 457 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 458 /// the specified Token's location, translating the token's start 459 /// position in the current buffer into a SourcePosition object for rendering. 460 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 461 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 462 } 463 464 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 465 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 466 DiagID); 467 } 468 469 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 470 /// token is the characters used to represent the token in the source file 471 /// after trigraph expansion and escaped-newline folding. In particular, this 472 /// wants to get the true, uncanonicalized, spelling of things like digraphs 473 /// UCNs, etc. 474 std::string getSpelling(const Token &Tok) const; 475 476 /// getSpelling - This method is used to get the spelling of a token into a 477 /// preallocated buffer, instead of as an std::string. The caller is required 478 /// to allocate enough space for the token, which is guaranteed to be at least 479 /// Tok.getLength() bytes long. The length of the actual result is returned. 480 /// 481 /// Note that this method may do two possible things: it may either fill in 482 /// the buffer specified with characters, or it may *change the input pointer* 483 /// to point to a constant buffer with the data already in it (avoiding a 484 /// copy). The caller is not allowed to modify the returned buffer pointer 485 /// if an internal buffer is returned. 486 unsigned getSpelling(const Token &Tok, const char *&Buffer) const; 487 488 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 489 /// with length 1, return the character. 490 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok) const { 491 assert(Tok.is(tok::numeric_constant) && 492 Tok.getLength() == 1 && "Called on unsupported token"); 493 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 494 495 // If the token is carrying a literal data pointer, just use it. 496 if (const char *D = Tok.getLiteralData()) 497 return *D; 498 499 // Otherwise, fall back on getCharacterData, which is slower, but always 500 // works. 501 return *SourceMgr.getCharacterData(Tok.getLocation()); 502 } 503 504 /// CreateString - Plop the specified string into a scratch buffer and set the 505 /// specified token's location and length to it. If specified, the source 506 /// location provides a location of the instantiation point of the token. 507 void CreateString(const char *Buf, unsigned Len, 508 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 509 510 /// \brief Computes the source location just past the end of the 511 /// token at this source location. 512 /// 513 /// This routine can be used to produce a source location that 514 /// points just past the end of the token referenced by \p Loc, and 515 /// is generally used when a diagnostic needs to point just after a 516 /// token where it expected something different that it received. If 517 /// the returned source location would not be meaningful (e.g., if 518 /// it points into a macro), this routine returns an invalid 519 /// source location. 520 SourceLocation getLocForEndOfToken(SourceLocation Loc); 521 522 /// DumpToken - Print the token to stderr, used for debugging. 523 /// 524 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 525 void DumpLocation(SourceLocation Loc) const; 526 void DumpMacro(const MacroInfo &MI) const; 527 528 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 529 /// token, return a new location that specifies a character within the token. 530 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 531 532 /// IncrementPasteCounter - Increment the counters for the number of token 533 /// paste operations performed. If fast was specified, this is a 'fast paste' 534 /// case we handled. 535 /// 536 void IncrementPasteCounter(bool isFast) { 537 if (isFast) 538 ++NumFastTokenPaste; 539 else 540 ++NumTokenPaste; 541 } 542 543 void PrintStats(); 544 545 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 546 /// comment (/##/) in microsoft mode, this method handles updating the current 547 /// state, returning the token on the next source line. 548 void HandleMicrosoftCommentPaste(Token &Tok); 549 550 //===--------------------------------------------------------------------===// 551 // Preprocessor callback methods. These are invoked by a lexer as various 552 // directives and events are found. 553 554 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 555 /// identifier information for the token and install it into the token. 556 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 557 const char *BufPtr = 0); 558 559 /// HandleIdentifier - This callback is invoked when the lexer reads an 560 /// identifier and has filled in the tokens IdentifierInfo member. This 561 /// callback potentially macro expands it or turns it into a named token (like 562 /// 'for'). 563 void HandleIdentifier(Token &Identifier); 564 565 566 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 567 /// the current file. This either returns the EOF token and returns true, or 568 /// pops a level off the include stack and returns false, at which point the 569 /// client should call lex again. 570 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 571 572 /// HandleEndOfTokenLexer - This callback is invoked when the current 573 /// TokenLexer hits the end of its token stream. 574 bool HandleEndOfTokenLexer(Token &Result); 575 576 /// HandleDirective - This callback is invoked when the lexer sees a # token 577 /// at the start of a line. This consumes the directive, modifies the 578 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 579 /// read is the correct one. 580 void HandleDirective(Token &Result); 581 582 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 583 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 584 /// true, then we consider macros that expand to zero tokens as being ok. 585 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 586 587 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 588 /// current line until the tok::eom token is found. 589 void DiscardUntilEndOfDirective(); 590 591 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 592 /// __DATE__ or __TIME__ in the file so far. 593 bool SawDateOrTime() const { 594 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 595 } 596 unsigned getCounterValue() const { return CounterValue; } 597 void setCounterValue(unsigned V) { CounterValue = V; } 598 599 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 600 /// SourceLocation. 601 MacroInfo* AllocateMacroInfo(SourceLocation L); 602 603private: 604 605 void PushIncludeMacroStack() { 606 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 607 CurPTHLexer.take(), 608 CurPPLexer, 609 CurTokenLexer.take(), 610 CurDirLookup)); 611 CurPPLexer = 0; 612 } 613 614 void PopIncludeMacroStack() { 615 CurLexer.reset(IncludeMacroStack.back().TheLexer); 616 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 617 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 618 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 619 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 620 IncludeMacroStack.pop_back(); 621 } 622 623 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 624 /// be reused for allocating new MacroInfo objects. 625 void ReleaseMacroInfo(MacroInfo* MI); 626 627 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 628 /// #include. 629 bool isInPrimaryFile() const; 630 631 /// ReadMacroName - Lex and validate a macro name, which occurs after a 632 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 633 /// and discards the rest of the macro line if the macro name is invalid. 634 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 635 636 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 637 /// definition has just been read. Lex the rest of the arguments and the 638 /// closing ), updating MI with what we learn. Return true if an error occurs 639 /// parsing the arg list. 640 bool ReadMacroDefinitionArgList(MacroInfo *MI); 641 642 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 643 /// decided that the subsequent tokens are in the #if'd out portion of the 644 /// file. Lex the rest of the file, until we see an #endif. If 645 /// FoundNonSkipPortion is true, then we have already emitted code for part of 646 /// this #if directive, so #else/#elif blocks should never be entered. If 647 /// FoundElse is false, then #else directives are ok, if not, then we have 648 /// already seen one so a #else directive is a duplicate. When this returns, 649 /// the caller can lex the first valid token. 650 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 651 bool FoundNonSkipPortion, bool FoundElse); 652 653 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 654 /// SkipExcludedConditionalBlock. 655 void PTHSkipExcludedConditionalBlock(); 656 657 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 658 /// may occur after a #if or #elif directive and return it as a bool. If the 659 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 660 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 661 662 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 663 /// #pragma GCC poison/system_header/dependency and #pragma once. 664 void RegisterBuiltinPragmas(); 665 666 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 667 /// identifier table. 668 void RegisterBuiltinMacros(); 669 IdentifierInfo *RegisterBuiltinMacro(const char *Name); 670 671 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 672 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 673 /// the macro should not be expanded return true, otherwise return false. 674 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 675 676 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 677 /// lexed is a '('. If so, consume the token and return true, if not, this 678 /// method should have no observable side-effect on the lexed tokens. 679 bool isNextPPTokenLParen(); 680 681 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 682 /// invoked to read all of the formal arguments specified for the macro 683 /// invocation. This returns null on error. 684 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 685 SourceLocation &InstantiationEnd); 686 687 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 688 /// as a builtin macro, handle it and return the next token as 'Tok'. 689 void ExpandBuiltinMacro(Token &Tok); 690 691 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 692 /// return the first token after the directive. The _Pragma token has just 693 /// been read into 'Tok'. 694 void Handle_Pragma(Token &Tok); 695 696 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 697 /// start lexing tokens from it instead of the current buffer. 698 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 699 700 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 701 /// start getting tokens from it using the PTH cache. 702 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 703 704 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 705 /// checked and spelled filename, e.g. as an operand of #include. This returns 706 /// true if the input filename was in <>'s or false if it were in ""'s. The 707 /// caller is expected to provide a buffer that is large enough to hold the 708 /// spelling of the filename, but is also expected to handle the case when 709 /// this method decides to use a different buffer. 710 bool GetIncludeFilenameSpelling(SourceLocation Loc, 711 const char *&BufStart, const char *&BufEnd); 712 713 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 714 /// return null on failure. isAngled indicates whether the file reference is 715 /// for system #include's or not (i.e. using <> instead of ""). 716 const FileEntry *LookupFile(const char *FilenameStart,const char *FilenameEnd, 717 bool isAngled, const DirectoryLookup *FromDir, 718 const DirectoryLookup *&CurDir); 719 720 721 722 /// IsFileLexer - Returns true if we are lexing from a file and not a 723 /// pragma or a macro. 724 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 725 return L ? !L->isPragmaLexer() : P != 0; 726 } 727 728 static bool IsFileLexer(const IncludeStackInfo& I) { 729 return IsFileLexer(I.TheLexer, I.ThePPLexer); 730 } 731 732 bool IsFileLexer() const { 733 return IsFileLexer(CurLexer.get(), CurPPLexer); 734 } 735 736 //===--------------------------------------------------------------------===// 737 // Caching stuff. 738 void CachingLex(Token &Result); 739 bool InCachingLexMode() const { return CurPPLexer == 0 && CurTokenLexer == 0;} 740 void EnterCachingLexMode(); 741 void ExitCachingLexMode() { 742 if (InCachingLexMode()) 743 RemoveTopOfLexerStack(); 744 } 745 const Token &PeekAhead(unsigned N); 746 void AnnotatePreviousCachedTokens(const Token &Tok); 747 748 //===--------------------------------------------------------------------===// 749 /// Handle*Directive - implement the various preprocessor directives. These 750 /// should side-effect the current preprocessor object so that the next call 751 /// to Lex() will return the appropriate token next. 752 void HandleLineDirective(Token &Tok); 753 void HandleDigitDirective(Token &Tok); 754 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 755 void HandleIdentSCCSDirective(Token &Tok); 756 757 // File inclusion. 758 void HandleIncludeDirective(Token &Tok, 759 const DirectoryLookup *LookupFrom = 0, 760 bool isImport = false); 761 void HandleIncludeNextDirective(Token &Tok); 762 void HandleIncludeMacrosDirective(Token &Tok); 763 void HandleImportDirective(Token &Tok); 764 765 // Macro handling. 766 void HandleDefineDirective(Token &Tok); 767 void HandleUndefDirective(Token &Tok); 768 // HandleAssertDirective(Token &Tok); 769 // HandleUnassertDirective(Token &Tok); 770 771 // Conditional Inclusion. 772 void HandleIfdefDirective(Token &Tok, bool isIfndef, 773 bool ReadAnyTokensBeforeDirective); 774 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 775 void HandleEndifDirective(Token &Tok); 776 void HandleElseDirective(Token &Tok); 777 void HandleElifDirective(Token &Tok); 778 779 // Pragmas. 780 void HandlePragmaDirective(); 781public: 782 void HandlePragmaOnce(Token &OnceTok); 783 void HandlePragmaMark(); 784 void HandlePragmaPoison(Token &PoisonTok); 785 void HandlePragmaSystemHeader(Token &SysHeaderTok); 786 void HandlePragmaDependency(Token &DependencyTok); 787 void HandlePragmaComment(Token &CommentTok); 788}; 789 790/// PreprocessorFactory - A generic factory interface for lazily creating 791/// Preprocessor objects on-demand when they are needed. 792class PreprocessorFactory { 793public: 794 virtual ~PreprocessorFactory(); 795 virtual Preprocessor* CreatePreprocessor() = 0; 796}; 797 798} // end namespace clang 799 800#endif 801