Preprocessor.h revision 148772a841cae6f32db16d890e788b92a763bb3f
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 IdentifierInfo *Ident__has_feature; // __has_feature 73 IdentifierInfo *Ident__has_builtin; // __has_builtin 74 75 SourceLocation DATELoc, TIMELoc; 76 unsigned CounterValue; // Next __COUNTER__ value. 77 78 enum { 79 /// MaxIncludeStackDepth - Maximum depth of #includes. 80 MaxAllowedIncludeStackDepth = 200 81 }; 82 83 // State that is set before the preprocessor begins. 84 bool KeepComments : 1; 85 bool KeepMacroComments : 1; 86 87 // State that changes while the preprocessor runs: 88 bool DisableMacroExpansion : 1; // True if macro expansion is disabled. 89 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 90 91 /// Identifiers - This is mapping/lookup information for all identifiers in 92 /// the program, including program keywords. 93 IdentifierTable Identifiers; 94 95 /// Selectors - This table contains all the selectors in the program. Unlike 96 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 97 /// It is declared/instantiated here because it's role/lifetime is 98 /// conceptually similar the IdentifierTable. In addition, the current control 99 /// flow (in clang::ParseAST()), make it convenient to put here. 100 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 101 /// the lifetime fo the preprocessor. 102 SelectorTable Selectors; 103 104 /// PragmaHandlers - This tracks all of the pragmas that the client registered 105 /// with this preprocessor. 106 PragmaNamespace *PragmaHandlers; 107 108 /// CurLexer - This is the current top of the stack that we're lexing from if 109 /// not expanding a macro and we are lexing directly from source code. 110 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 111 llvm::OwningPtr<Lexer> CurLexer; 112 113 /// CurPTHLexer - This is the current top of stack that we're lexing from if 114 /// not expanding from a macro and we are lexing from a PTH cache. 115 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 116 llvm::OwningPtr<PTHLexer> CurPTHLexer; 117 118 /// CurPPLexer - This is the current top of the stack what we're lexing from 119 /// if not expanding a macro. This is an alias for either CurLexer or 120 /// CurPTHLexer. 121 PreprocessorLexer* CurPPLexer; 122 123 /// CurLookup - The DirectoryLookup structure used to find the current 124 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 125 /// implement #include_next and find directory-specific properties. 126 const DirectoryLookup *CurDirLookup; 127 128 /// CurTokenLexer - This is the current macro we are expanding, if we are 129 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 130 llvm::OwningPtr<TokenLexer> CurTokenLexer; 131 132 /// IncludeMacroStack - This keeps track of the stack of files currently 133 /// #included, and macros currently being expanded from, not counting 134 /// CurLexer/CurTokenLexer. 135 struct IncludeStackInfo { 136 Lexer *TheLexer; 137 PTHLexer *ThePTHLexer; 138 PreprocessorLexer *ThePPLexer; 139 TokenLexer *TheTokenLexer; 140 const DirectoryLookup *TheDirLookup; 141 142 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 143 TokenLexer* TL, const DirectoryLookup *D) 144 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 145 TheDirLookup(D) {} 146 }; 147 std::vector<IncludeStackInfo> IncludeMacroStack; 148 149 /// Callbacks - These are actions invoked when some preprocessor activity is 150 /// encountered (e.g. a file is #included, etc). 151 PPCallbacks *Callbacks; 152 153 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 154 /// to the actual definition of the macro. 155 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 156 157 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 158 /// allocation. 159 std::vector<MacroInfo*> MICache; 160 161 // Various statistics we track for performance analysis. 162 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 163 unsigned NumIf, NumElse, NumEndif; 164 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 165 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 166 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 167 unsigned NumSkipped; 168 169 /// Predefines - This string is the predefined macros that preprocessor 170 /// should use from the command line etc. 171 std::string Predefines; 172 173 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 174 enum { TokenLexerCacheSize = 8 }; 175 unsigned NumCachedTokenLexers; 176 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 177 178private: // Cached tokens state. 179 typedef std::vector<Token> CachedTokensTy; 180 181 /// CachedTokens - Cached tokens are stored here when we do backtracking or 182 /// lookahead. They are "lexed" by the CachingLex() method. 183 CachedTokensTy CachedTokens; 184 185 /// CachedLexPos - The position of the cached token that CachingLex() should 186 /// "lex" next. If it points beyond the CachedTokens vector, it means that 187 /// a normal Lex() should be invoked. 188 CachedTokensTy::size_type CachedLexPos; 189 190 /// BacktrackPositions - Stack of backtrack positions, allowing nested 191 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 192 /// indicate where CachedLexPos should be set when the BackTrack() method is 193 /// invoked (at which point the last position is popped). 194 std::vector<CachedTokensTy::size_type> BacktrackPositions; 195 196public: 197 Preprocessor(Diagnostic &diags, const LangOptions &opts, TargetInfo &target, 198 SourceManager &SM, HeaderSearch &Headers, 199 IdentifierInfoLookup *IILookup = 0); 200 201 ~Preprocessor(); 202 203 Diagnostic &getDiagnostics() const { return *Diags; } 204 void setDiagnostics(Diagnostic &D) { Diags = &D; } 205 206 const LangOptions &getLangOptions() const { return Features; } 207 TargetInfo &getTargetInfo() const { return Target; } 208 FileManager &getFileManager() const { return FileMgr; } 209 SourceManager &getSourceManager() const { return SourceMgr; } 210 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 211 212 IdentifierTable &getIdentifierTable() { return Identifiers; } 213 SelectorTable &getSelectorTable() { return Selectors; } 214 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 215 216 void setPTHManager(PTHManager* pm); 217 218 PTHManager *getPTHManager() { return PTH.get(); } 219 220 /// SetCommentRetentionState - Control whether or not the preprocessor retains 221 /// comments in output. 222 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 223 this->KeepComments = KeepComments | KeepMacroComments; 224 this->KeepMacroComments = KeepMacroComments; 225 } 226 227 bool getCommentRetentionState() const { return KeepComments; } 228 229 /// isCurrentLexer - Return true if we are lexing directly from the specified 230 /// lexer. 231 bool isCurrentLexer(const PreprocessorLexer *L) const { 232 return CurPPLexer == L; 233 } 234 235 /// getCurrentLexer - Return the current file lexer being lexed from. Note 236 /// that this ignores any potentially active macro expansions and _Pragma 237 /// expansions going on at the time. 238 PreprocessorLexer *getCurrentFileLexer() const; 239 240 /// getPPCallbacks/setPPCallbacks - Accessors for preprocessor callbacks. 241 /// Note that this class takes ownership of any PPCallbacks object given to 242 /// it. 243 PPCallbacks *getPPCallbacks() const { return Callbacks; } 244 void setPPCallbacks(PPCallbacks *C) { 245 if (Callbacks) 246 C = new PPChainedCallbacks(C, Callbacks); 247 Callbacks = C; 248 } 249 250 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 251 /// or null if it isn't #define'd. 252 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 253 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 254 } 255 256 /// setMacroInfo - Specify a macro for this identifier. 257 /// 258 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 259 260 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 261 /// state of the macro table. This visits every currently-defined macro. 262 typedef llvm::DenseMap<IdentifierInfo*, 263 MacroInfo*>::const_iterator macro_iterator; 264 macro_iterator macro_begin() const { return Macros.begin(); } 265 macro_iterator macro_end() const { return Macros.end(); } 266 267 268 269 const std::string &getPredefines() const { return Predefines; } 270 /// setPredefines - Set the predefines for this Preprocessor. These 271 /// predefines are automatically injected when parsing the main file. 272 void setPredefines(const char *P) { Predefines = P; } 273 void setPredefines(const std::string &P) { Predefines = P; } 274 275 /// getIdentifierInfo - Return information about the specified preprocessor 276 /// identifier token. The version of this method that takes two character 277 /// pointers is preferred unless the identifier is already available as a 278 /// string (this avoids allocation and copying of memory to construct an 279 /// std::string). 280 IdentifierInfo *getIdentifierInfo(const char *NameStart, 281 const char *NameEnd) { 282 return &Identifiers.get(NameStart, NameEnd); 283 } 284 IdentifierInfo *getIdentifierInfo(const char *NameStr) { 285 return getIdentifierInfo(NameStr, NameStr+strlen(NameStr)); 286 } 287 288 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 289 /// If 'Namespace' is non-null, then it is a token required to exist on the 290 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 291 void AddPragmaHandler(const char *Namespace, PragmaHandler *Handler); 292 293 /// RemovePragmaHandler - Remove the specific pragma handler from 294 /// the preprocessor. If \arg Namespace is non-null, then it should 295 /// be the namespace that \arg Handler was added to. It is an error 296 /// to remove a handler that has not been registered. 297 void RemovePragmaHandler(const char *Namespace, PragmaHandler *Handler); 298 299 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 300 /// which implicitly adds the builtin defines etc. 301 void EnterMainSourceFile(); 302 303 /// EnterSourceFile - Add a source file to the top of the include stack and 304 /// start lexing tokens from it instead of the current buffer. If isMainFile 305 /// is true, this is the main file for the translation unit. 306 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir); 307 308 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 309 /// tokens from it instead of the current buffer. Args specifies the 310 /// tokens input to a function-like macro. 311 /// 312 /// ILEnd specifies the location of the ')' for a function-like macro or the 313 /// identifier for an object-like macro. 314 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 315 316 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 317 /// which will cause the lexer to start returning the specified tokens. 318 /// 319 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 320 /// not be subject to further macro expansion. Otherwise, these tokens will 321 /// be re-macro-expanded when/if expansion is enabled. 322 /// 323 /// If OwnsTokens is false, this method assumes that the specified stream of 324 /// tokens has a permanent owner somewhere, so they do not need to be copied. 325 /// If it is true, it assumes the array of tokens is allocated with new[] and 326 /// must be freed. 327 /// 328 void EnterTokenStream(const Token *Toks, unsigned NumToks, 329 bool DisableMacroExpansion, bool OwnsTokens); 330 331 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 332 /// lexer stack. This should only be used in situations where the current 333 /// state of the top-of-stack lexer is known. 334 void RemoveTopOfLexerStack(); 335 336 /// EnableBacktrackAtThisPos - From the point that this method is called, and 337 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 338 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 339 /// make the Preprocessor re-lex the same tokens. 340 /// 341 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 342 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 343 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 344 /// 345 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 346 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 347 /// tokens will continue indefinitely. 348 /// 349 void EnableBacktrackAtThisPos(); 350 351 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 352 void CommitBacktrackedTokens(); 353 354 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 355 /// EnableBacktrackAtThisPos() was previously called. 356 void Backtrack(); 357 358 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 359 /// caching of tokens is on. 360 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 361 362 /// Lex - To lex a token from the preprocessor, just pull a token from the 363 /// current lexer or macro object. 364 void Lex(Token &Result) { 365 if (CurLexer) 366 CurLexer->Lex(Result); 367 else if (CurPTHLexer) 368 CurPTHLexer->Lex(Result); 369 else if (CurTokenLexer) 370 CurTokenLexer->Lex(Result); 371 else 372 CachingLex(Result); 373 } 374 375 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 376 /// something not a comment. This is useful in -E -C mode where comments 377 /// would foul up preprocessor directive handling. 378 void LexNonComment(Token &Result) { 379 do 380 Lex(Result); 381 while (Result.getKind() == tok::comment); 382 } 383 384 /// LexUnexpandedToken - This is just like Lex, but this disables macro 385 /// expansion of identifier tokens. 386 void LexUnexpandedToken(Token &Result) { 387 // Disable macro expansion. 388 bool OldVal = DisableMacroExpansion; 389 DisableMacroExpansion = true; 390 // Lex the token. 391 Lex(Result); 392 393 // Reenable it. 394 DisableMacroExpansion = OldVal; 395 } 396 397 /// LookAhead - This peeks ahead N tokens and returns that token without 398 /// consuming any tokens. LookAhead(0) returns the next token that would be 399 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 400 /// returns normal tokens after phase 5. As such, it is equivalent to using 401 /// 'Lex', not 'LexUnexpandedToken'. 402 const Token &LookAhead(unsigned N) { 403 if (CachedLexPos + N < CachedTokens.size()) 404 return CachedTokens[CachedLexPos+N]; 405 else 406 return PeekAhead(N+1); 407 } 408 409 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 410 /// this allows to revert a specific number of tokens. 411 /// Note that the number of tokens being reverted should be up to the last 412 /// backtrack position, not more. 413 void RevertCachedTokens(unsigned N) { 414 assert(isBacktrackEnabled() && 415 "Should only be called when tokens are cached for backtracking"); 416 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 417 && "Should revert tokens up to the last backtrack position, not more"); 418 assert(signed(CachedLexPos) - signed(N) >= 0 && 419 "Corrupted backtrack positions ?"); 420 CachedLexPos -= N; 421 } 422 423 /// EnterToken - Enters a token in the token stream to be lexed next. If 424 /// BackTrack() is called afterwards, the token will remain at the insertion 425 /// point. 426 void EnterToken(const Token &Tok) { 427 EnterCachingLexMode(); 428 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 429 } 430 431 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 432 /// tokens (because backtrack is enabled) it should replace the most recent 433 /// cached tokens with the given annotation token. This function has no effect 434 /// if backtracking is not enabled. 435 /// 436 /// Note that the use of this function is just for optimization; so that the 437 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 438 /// invoked. 439 void AnnotateCachedTokens(const Token &Tok) { 440 assert(Tok.isAnnotation() && "Expected annotation token"); 441 if (CachedLexPos != 0 && isBacktrackEnabled()) 442 AnnotatePreviousCachedTokens(Tok); 443 } 444 445 /// \brief Replace the last token with an annotation token. 446 /// 447 /// Like AnnotateCachedTokens(), this routine replaces an 448 /// already-parsed (and resolved) token with an annotation 449 /// token. However, this routine only replaces the last token with 450 /// the annotation token; it does not affect any other cached 451 /// tokens. This function has no effect if backtracking is not 452 /// enabled. 453 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 454 assert(Tok.isAnnotation() && "Expected annotation token"); 455 if (CachedLexPos != 0 && isBacktrackEnabled()) 456 CachedTokens[CachedLexPos-1] = Tok; 457 } 458 459 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 460 /// the specified Token's location, translating the token's start 461 /// position in the current buffer into a SourcePosition object for rendering. 462 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 463 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 464 } 465 466 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 467 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 468 DiagID); 469 } 470 471 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 472 /// token is the characters used to represent the token in the source file 473 /// after trigraph expansion and escaped-newline folding. In particular, this 474 /// wants to get the true, uncanonicalized, spelling of things like digraphs 475 /// UCNs, etc. 476 std::string getSpelling(const Token &Tok) const; 477 478 /// getSpelling - This method is used to get the spelling of a token into a 479 /// preallocated buffer, instead of as an std::string. The caller is required 480 /// to allocate enough space for the token, which is guaranteed to be at least 481 /// Tok.getLength() bytes long. The length of the actual result is returned. 482 /// 483 /// Note that this method may do two possible things: it may either fill in 484 /// the buffer specified with characters, or it may *change the input pointer* 485 /// to point to a constant buffer with the data already in it (avoiding a 486 /// copy). The caller is not allowed to modify the returned buffer pointer 487 /// if an internal buffer is returned. 488 unsigned getSpelling(const Token &Tok, const char *&Buffer) const; 489 490 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 491 /// with length 1, return the character. 492 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok) const { 493 assert(Tok.is(tok::numeric_constant) && 494 Tok.getLength() == 1 && "Called on unsupported token"); 495 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 496 497 // If the token is carrying a literal data pointer, just use it. 498 if (const char *D = Tok.getLiteralData()) 499 return *D; 500 501 // Otherwise, fall back on getCharacterData, which is slower, but always 502 // works. 503 return *SourceMgr.getCharacterData(Tok.getLocation()); 504 } 505 506 /// CreateString - Plop the specified string into a scratch buffer and set the 507 /// specified token's location and length to it. If specified, the source 508 /// location provides a location of the instantiation point of the token. 509 void CreateString(const char *Buf, unsigned Len, 510 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 511 512 /// \brief Computes the source location just past the end of the 513 /// token at this source location. 514 /// 515 /// This routine can be used to produce a source location that 516 /// points just past the end of the token referenced by \p Loc, and 517 /// is generally used when a diagnostic needs to point just after a 518 /// token where it expected something different that it received. If 519 /// the returned source location would not be meaningful (e.g., if 520 /// it points into a macro), this routine returns an invalid 521 /// source location. 522 SourceLocation getLocForEndOfToken(SourceLocation Loc); 523 524 /// DumpToken - Print the token to stderr, used for debugging. 525 /// 526 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 527 void DumpLocation(SourceLocation Loc) const; 528 void DumpMacro(const MacroInfo &MI) const; 529 530 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 531 /// token, return a new location that specifies a character within the token. 532 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 533 534 /// IncrementPasteCounter - Increment the counters for the number of token 535 /// paste operations performed. If fast was specified, this is a 'fast paste' 536 /// case we handled. 537 /// 538 void IncrementPasteCounter(bool isFast) { 539 if (isFast) 540 ++NumFastTokenPaste; 541 else 542 ++NumTokenPaste; 543 } 544 545 void PrintStats(); 546 547 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 548 /// comment (/##/) in microsoft mode, this method handles updating the current 549 /// state, returning the token on the next source line. 550 void HandleMicrosoftCommentPaste(Token &Tok); 551 552 //===--------------------------------------------------------------------===// 553 // Preprocessor callback methods. These are invoked by a lexer as various 554 // directives and events are found. 555 556 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 557 /// identifier information for the token and install it into the token. 558 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 559 const char *BufPtr = 0); 560 561 /// HandleIdentifier - This callback is invoked when the lexer reads an 562 /// identifier and has filled in the tokens IdentifierInfo member. This 563 /// callback potentially macro expands it or turns it into a named token (like 564 /// 'for'). 565 void HandleIdentifier(Token &Identifier); 566 567 568 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 569 /// the current file. This either returns the EOF token and returns true, or 570 /// pops a level off the include stack and returns false, at which point the 571 /// client should call lex again. 572 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 573 574 /// HandleEndOfTokenLexer - This callback is invoked when the current 575 /// TokenLexer hits the end of its token stream. 576 bool HandleEndOfTokenLexer(Token &Result); 577 578 /// HandleDirective - This callback is invoked when the lexer sees a # token 579 /// at the start of a line. This consumes the directive, modifies the 580 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 581 /// read is the correct one. 582 void HandleDirective(Token &Result); 583 584 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 585 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 586 /// true, then we consider macros that expand to zero tokens as being ok. 587 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 588 589 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 590 /// current line until the tok::eom token is found. 591 void DiscardUntilEndOfDirective(); 592 593 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 594 /// __DATE__ or __TIME__ in the file so far. 595 bool SawDateOrTime() const { 596 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 597 } 598 unsigned getCounterValue() const { return CounterValue; } 599 void setCounterValue(unsigned V) { CounterValue = V; } 600 601 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 602 /// SourceLocation. 603 MacroInfo* AllocateMacroInfo(SourceLocation L); 604 605private: 606 607 void PushIncludeMacroStack() { 608 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 609 CurPTHLexer.take(), 610 CurPPLexer, 611 CurTokenLexer.take(), 612 CurDirLookup)); 613 CurPPLexer = 0; 614 } 615 616 void PopIncludeMacroStack() { 617 CurLexer.reset(IncludeMacroStack.back().TheLexer); 618 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 619 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 620 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 621 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 622 IncludeMacroStack.pop_back(); 623 } 624 625 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 626 /// be reused for allocating new MacroInfo objects. 627 void ReleaseMacroInfo(MacroInfo* MI); 628 629 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 630 /// #include. 631 bool isInPrimaryFile() const; 632 633 /// ReadMacroName - Lex and validate a macro name, which occurs after a 634 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 635 /// and discards the rest of the macro line if the macro name is invalid. 636 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 637 638 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 639 /// definition has just been read. Lex the rest of the arguments and the 640 /// closing ), updating MI with what we learn. Return true if an error occurs 641 /// parsing the arg list. 642 bool ReadMacroDefinitionArgList(MacroInfo *MI); 643 644 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 645 /// decided that the subsequent tokens are in the #if'd out portion of the 646 /// file. Lex the rest of the file, until we see an #endif. If 647 /// FoundNonSkipPortion is true, then we have already emitted code for part of 648 /// this #if directive, so #else/#elif blocks should never be entered. If 649 /// FoundElse is false, then #else directives are ok, if not, then we have 650 /// already seen one so a #else directive is a duplicate. When this returns, 651 /// the caller can lex the first valid token. 652 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 653 bool FoundNonSkipPortion, bool FoundElse); 654 655 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 656 /// SkipExcludedConditionalBlock. 657 void PTHSkipExcludedConditionalBlock(); 658 659 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 660 /// may occur after a #if or #elif directive and return it as a bool. If the 661 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 662 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 663 664 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 665 /// #pragma GCC poison/system_header/dependency and #pragma once. 666 void RegisterBuiltinPragmas(); 667 668 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 669 /// identifier table. 670 void RegisterBuiltinMacros(); 671 672 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 673 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 674 /// the macro should not be expanded return true, otherwise return false. 675 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 676 677 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 678 /// lexed is a '('. If so, consume the token and return true, if not, this 679 /// method should have no observable side-effect on the lexed tokens. 680 bool isNextPPTokenLParen(); 681 682 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 683 /// invoked to read all of the formal arguments specified for the macro 684 /// invocation. This returns null on error. 685 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 686 SourceLocation &InstantiationEnd); 687 688 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 689 /// as a builtin macro, handle it and return the next token as 'Tok'. 690 void ExpandBuiltinMacro(Token &Tok); 691 692 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 693 /// return the first token after the directive. The _Pragma token has just 694 /// been read into 'Tok'. 695 void Handle_Pragma(Token &Tok); 696 697 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 698 /// start lexing tokens from it instead of the current buffer. 699 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 700 701 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 702 /// start getting tokens from it using the PTH cache. 703 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 704 705 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 706 /// checked and spelled filename, e.g. as an operand of #include. This returns 707 /// true if the input filename was in <>'s or false if it were in ""'s. The 708 /// caller is expected to provide a buffer that is large enough to hold the 709 /// spelling of the filename, but is also expected to handle the case when 710 /// this method decides to use a different buffer. 711 bool GetIncludeFilenameSpelling(SourceLocation Loc, 712 const char *&BufStart, const char *&BufEnd); 713 714 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 715 /// return null on failure. isAngled indicates whether the file reference is 716 /// for system #include's or not (i.e. using <> instead of ""). 717 const FileEntry *LookupFile(const char *FilenameStart,const char *FilenameEnd, 718 bool isAngled, const DirectoryLookup *FromDir, 719 const DirectoryLookup *&CurDir); 720 721 722 723 /// IsFileLexer - Returns true if we are lexing from a file and not a 724 /// pragma or a macro. 725 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 726 return L ? !L->isPragmaLexer() : P != 0; 727 } 728 729 static bool IsFileLexer(const IncludeStackInfo& I) { 730 return IsFileLexer(I.TheLexer, I.ThePPLexer); 731 } 732 733 bool IsFileLexer() const { 734 return IsFileLexer(CurLexer.get(), CurPPLexer); 735 } 736 737 //===--------------------------------------------------------------------===// 738 // Caching stuff. 739 void CachingLex(Token &Result); 740 bool InCachingLexMode() const { return CurPPLexer == 0 && CurTokenLexer == 0;} 741 void EnterCachingLexMode(); 742 void ExitCachingLexMode() { 743 if (InCachingLexMode()) 744 RemoveTopOfLexerStack(); 745 } 746 const Token &PeekAhead(unsigned N); 747 void AnnotatePreviousCachedTokens(const Token &Tok); 748 749 //===--------------------------------------------------------------------===// 750 /// Handle*Directive - implement the various preprocessor directives. These 751 /// should side-effect the current preprocessor object so that the next call 752 /// to Lex() will return the appropriate token next. 753 void HandleLineDirective(Token &Tok); 754 void HandleDigitDirective(Token &Tok); 755 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 756 void HandleIdentSCCSDirective(Token &Tok); 757 758 // File inclusion. 759 void HandleIncludeDirective(Token &Tok, 760 const DirectoryLookup *LookupFrom = 0, 761 bool isImport = false); 762 void HandleIncludeNextDirective(Token &Tok); 763 void HandleIncludeMacrosDirective(Token &Tok); 764 void HandleImportDirective(Token &Tok); 765 766 // Macro handling. 767 void HandleDefineDirective(Token &Tok); 768 void HandleUndefDirective(Token &Tok); 769 // HandleAssertDirective(Token &Tok); 770 // HandleUnassertDirective(Token &Tok); 771 772 // Conditional Inclusion. 773 void HandleIfdefDirective(Token &Tok, bool isIfndef, 774 bool ReadAnyTokensBeforeDirective); 775 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 776 void HandleEndifDirective(Token &Tok); 777 void HandleElseDirective(Token &Tok); 778 void HandleElifDirective(Token &Tok); 779 780 // Pragmas. 781 void HandlePragmaDirective(); 782public: 783 void HandlePragmaOnce(Token &OnceTok); 784 void HandlePragmaMark(); 785 void HandlePragmaPoison(Token &PoisonTok); 786 void HandlePragmaSystemHeader(Token &SysHeaderTok); 787 void HandlePragmaDependency(Token &DependencyTok); 788 void HandlePragmaComment(Token &CommentTok); 789}; 790 791/// PreprocessorFactory - A generic factory interface for lazily creating 792/// Preprocessor objects on-demand when they are needed. 793class PreprocessorFactory { 794public: 795 virtual ~PreprocessorFactory(); 796 virtual Preprocessor* CreatePreprocessor() = 0; 797}; 798 799} // end namespace clang 800 801#endif 802