Preprocessor.h revision 7c175fb196a2bc3dbc86ea3865c713e1875f3f6d
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_Pragma, *Ident__VA_ARGS__; // _Pragma, __VA_ARGS__ 71 72 SourceLocation DATELoc, TIMELoc; 73 74 enum { 75 /// MaxIncludeStackDepth - Maximum depth of #includes. 76 MaxAllowedIncludeStackDepth = 200 77 }; 78 79 // State that is set before the preprocessor begins. 80 bool KeepComments : 1; 81 bool KeepMacroComments : 1; 82 83 // State that changes while the preprocessor runs: 84 bool DisableMacroExpansion : 1; // True if macro expansion is disabled. 85 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 86 87 /// Identifiers - This is mapping/lookup information for all identifiers in 88 /// the program, including program keywords. 89 IdentifierTable Identifiers; 90 91 /// Selectors - This table contains all the selectors in the program. Unlike 92 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 93 /// It is declared/instantiated here because it's role/lifetime is 94 /// conceptually similar the IdentifierTable. In addition, the current control 95 /// flow (in clang::ParseAST()), make it convenient to put here. 96 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 97 /// the lifetime fo the preprocessor. 98 SelectorTable Selectors; 99 100 /// PragmaHandlers - This tracks all of the pragmas that the client registered 101 /// with this preprocessor. 102 PragmaNamespace *PragmaHandlers; 103 104 /// CurLexer - This is the current top of the stack that we're lexing from if 105 /// not expanding a macro and we are lexing directly from source code. 106 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 107 llvm::OwningPtr<Lexer> CurLexer; 108 109 /// CurPTHLexer - This is the current top of stack that we're lexing from if 110 /// not expanding from a macro and we are lexing from a PTH cache. 111 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 112 llvm::OwningPtr<PTHLexer> CurPTHLexer; 113 114 /// CurPPLexer - This is the current top of the stack what we're lexing from 115 /// if not expanding a macro. This is an alias for either CurLexer or 116 /// CurPTHLexer. 117 PreprocessorLexer* CurPPLexer; 118 119 /// CurLookup - The DirectoryLookup structure used to find the current 120 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 121 /// implement #include_next and find directory-specific properties. 122 const DirectoryLookup *CurDirLookup; 123 124 /// CurTokenLexer - This is the current macro we are expanding, if we are 125 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 126 llvm::OwningPtr<TokenLexer> CurTokenLexer; 127 128 /// IncludeMacroStack - This keeps track of the stack of files currently 129 /// #included, and macros currently being expanded from, not counting 130 /// CurLexer/CurTokenLexer. 131 struct IncludeStackInfo { 132 Lexer *TheLexer; 133 PTHLexer *ThePTHLexer; 134 PreprocessorLexer *ThePPLexer; 135 TokenLexer *TheTokenLexer; 136 const DirectoryLookup *TheDirLookup; 137 138 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 139 TokenLexer* TL, const DirectoryLookup *D) 140 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 141 TheDirLookup(D) {} 142 }; 143 std::vector<IncludeStackInfo> IncludeMacroStack; 144 145 /// Callbacks - These are actions invoked when some preprocessor activity is 146 /// encountered (e.g. a file is #included, etc). 147 PPCallbacks *Callbacks; 148 149 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 150 /// to the actual definition of the macro. 151 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 152 153 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 154 /// allocation. 155 std::vector<MacroInfo*> MICache; 156 157 // Various statistics we track for performance analysis. 158 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 159 unsigned NumIf, NumElse, NumEndif; 160 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 161 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 162 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 163 unsigned NumSkipped; 164 165 /// Predefines - This string is the predefined macros that preprocessor 166 /// should use from the command line etc. 167 std::string Predefines; 168 169 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 170 enum { TokenLexerCacheSize = 8 }; 171 unsigned NumCachedTokenLexers; 172 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 173 174private: // Cached tokens state. 175 typedef std::vector<Token> CachedTokensTy; 176 177 /// CachedTokens - Cached tokens are stored here when we do backtracking or 178 /// lookahead. They are "lexed" by the CachingLex() method. 179 CachedTokensTy CachedTokens; 180 181 /// CachedLexPos - The position of the cached token that CachingLex() should 182 /// "lex" next. If it points beyond the CachedTokens vector, it means that 183 /// a normal Lex() should be invoked. 184 CachedTokensTy::size_type CachedLexPos; 185 186 /// BacktrackPositions - Stack of backtrack positions, allowing nested 187 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 188 /// indicate where CachedLexPos should be set when the BackTrack() method is 189 /// invoked (at which point the last position is popped). 190 std::vector<CachedTokensTy::size_type> BacktrackPositions; 191 192public: 193 Preprocessor(Diagnostic &diags, const LangOptions &opts, TargetInfo &target, 194 SourceManager &SM, HeaderSearch &Headers, 195 IdentifierInfoLookup* IILookup = 0); 196 197 ~Preprocessor(); 198 199 Diagnostic &getDiagnostics() const { return *Diags; } 200 void setDiagnostics(Diagnostic &D) { Diags = &D; } 201 202 203 const LangOptions &getLangOptions() const { return Features; } 204 TargetInfo &getTargetInfo() const { return Target; } 205 FileManager &getFileManager() const { return FileMgr; } 206 SourceManager &getSourceManager() const { return SourceMgr; } 207 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 208 209 IdentifierTable &getIdentifierTable() { return Identifiers; } 210 SelectorTable &getSelectorTable() { return Selectors; } 211 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 212 213 214 void setPTHManager(PTHManager* pm); 215 216 /// SetCommentRetentionState - Control whether or not the preprocessor retains 217 /// comments in output. 218 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 219 this->KeepComments = KeepComments | KeepMacroComments; 220 this->KeepMacroComments = KeepMacroComments; 221 } 222 223 bool getCommentRetentionState() const { return KeepComments; } 224 225 /// isCurrentLexer - Return true if we are lexing directly from the specified 226 /// lexer. 227 bool isCurrentLexer(const PreprocessorLexer *L) const { 228 return CurPPLexer == L; 229 } 230 231 /// getCurrentLexer - Return the current file lexer being lexed from. Note 232 /// that this ignores any potentially active macro expansions and _Pragma 233 /// expansions going on at the time. 234 PreprocessorLexer *getCurrentFileLexer() const; 235 236 /// getPPCallbacks/setPPCallbacks - Accessors for preprocessor callbacks. 237 /// Note that this class takes ownership of any PPCallbacks object given to 238 /// it. 239 PPCallbacks *getPPCallbacks() const { return Callbacks; } 240 void setPPCallbacks(PPCallbacks *C) { 241 delete Callbacks; 242 Callbacks = C; 243 } 244 245 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 246 /// or null if it isn't #define'd. 247 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 248 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 249 } 250 251 /// setMacroInfo - Specify a macro for this identifier. 252 /// 253 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 254 255 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 256 /// state of the macro table. This visits every currently-defined macro. 257 typedef llvm::DenseMap<IdentifierInfo*, 258 MacroInfo*>::const_iterator macro_iterator; 259 macro_iterator macro_begin() const { return Macros.begin(); } 260 macro_iterator macro_end() const { return Macros.end(); } 261 262 263 264 const std::string &getPredefines() const { return Predefines; } 265 /// setPredefines - Set the predefines for this Preprocessor. These 266 /// predefines are automatically injected when parsing the main file. 267 void setPredefines(const char *P) { Predefines = P; } 268 void setPredefines(const std::string &P) { Predefines = P; } 269 270 /// getIdentifierInfo - Return information about the specified preprocessor 271 /// identifier token. The version of this method that takes two character 272 /// pointers is preferred unless the identifier is already available as a 273 /// string (this avoids allocation and copying of memory to construct an 274 /// std::string). 275 IdentifierInfo *getIdentifierInfo(const char *NameStart, 276 const char *NameEnd) { 277 return &Identifiers.get(NameStart, NameEnd); 278 } 279 IdentifierInfo *getIdentifierInfo(const char *NameStr) { 280 return getIdentifierInfo(NameStr, NameStr+strlen(NameStr)); 281 } 282 283 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 284 /// If 'Namespace' is non-null, then it is a token required to exist on the 285 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 286 void AddPragmaHandler(const char *Namespace, PragmaHandler *Handler); 287 288 /// RemovePragmaHandler - Remove the specific pragma handler from 289 /// the preprocessor. If \arg Namespace is non-null, then it should 290 /// be the namespace that \arg Handler was added to. It is an error 291 /// to remove a handler that has not been registered. 292 void RemovePragmaHandler(const char *Namespace, PragmaHandler *Handler); 293 294 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 295 /// which implicitly adds the builtin defines etc. 296 void EnterMainSourceFile(); 297 298 /// EnterSourceFile - Add a source file to the top of the include stack and 299 /// start lexing tokens from it instead of the current buffer. If isMainFile 300 /// is true, this is the main file for the translation unit. 301 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir); 302 303 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 304 /// tokens from it instead of the current buffer. Args specifies the 305 /// tokens input to a function-like macro. 306 /// 307 /// ILEnd specifies the location of the ')' for a function-like macro or the 308 /// identifier for an object-like macro. 309 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 310 311 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 312 /// which will cause the lexer to start returning the specified tokens. 313 /// 314 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 315 /// not be subject to further macro expansion. Otherwise, these tokens will 316 /// be re-macro-expanded when/if expansion is enabled. 317 /// 318 /// If OwnsTokens is false, this method assumes that the specified stream of 319 /// tokens has a permanent owner somewhere, so they do not need to be copied. 320 /// If it is true, it assumes the array of tokens is allocated with new[] and 321 /// must be freed. 322 /// 323 void EnterTokenStream(const Token *Toks, unsigned NumToks, 324 bool DisableMacroExpansion, bool OwnsTokens); 325 326 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 327 /// lexer stack. This should only be used in situations where the current 328 /// state of the top-of-stack lexer is known. 329 void RemoveTopOfLexerStack(); 330 331 /// EnableBacktrackAtThisPos - From the point that this method is called, and 332 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 333 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 334 /// make the Preprocessor re-lex the same tokens. 335 /// 336 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 337 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 338 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 339 /// 340 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 341 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 342 /// tokens will continue indefinitely. 343 /// 344 void EnableBacktrackAtThisPos(); 345 346 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 347 void CommitBacktrackedTokens(); 348 349 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 350 /// EnableBacktrackAtThisPos() was previously called. 351 void Backtrack(); 352 353 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 354 /// caching of tokens is on. 355 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 356 357 /// Lex - To lex a token from the preprocessor, just pull a token from the 358 /// current lexer or macro object. 359 void Lex(Token &Result) { 360 if (CurLexer) 361 CurLexer->Lex(Result); 362 else if (CurPTHLexer) 363 CurPTHLexer->Lex(Result); 364 else if (CurTokenLexer) 365 CurTokenLexer->Lex(Result); 366 else 367 CachingLex(Result); 368 } 369 370 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 371 /// something not a comment. This is useful in -E -C mode where comments 372 /// would foul up preprocessor directive handling. 373 void LexNonComment(Token &Result) { 374 do 375 Lex(Result); 376 while (Result.getKind() == tok::comment); 377 } 378 379 /// LexUnexpandedToken - This is just like Lex, but this disables macro 380 /// expansion of identifier tokens. 381 void LexUnexpandedToken(Token &Result) { 382 // Disable macro expansion. 383 bool OldVal = DisableMacroExpansion; 384 DisableMacroExpansion = true; 385 // Lex the token. 386 Lex(Result); 387 388 // Reenable it. 389 DisableMacroExpansion = OldVal; 390 } 391 392 /// LookAhead - This peeks ahead N tokens and returns that token without 393 /// consuming any tokens. LookAhead(0) returns the next token that would be 394 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 395 /// returns normal tokens after phase 5. As such, it is equivalent to using 396 /// 'Lex', not 'LexUnexpandedToken'. 397 const Token &LookAhead(unsigned N) { 398 if (CachedLexPos + N < CachedTokens.size()) 399 return CachedTokens[CachedLexPos+N]; 400 else 401 return PeekAhead(N+1); 402 } 403 404 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 405 /// this allows to revert a specific number of tokens. 406 /// Note that the number of tokens being reverted should be up to the last 407 /// backtrack position, not more. 408 void RevertCachedTokens(unsigned N) { 409 assert(isBacktrackEnabled() && 410 "Should only be called when tokens are cached for backtracking"); 411 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 412 && "Should revert tokens up to the last backtrack position, not more"); 413 assert(signed(CachedLexPos) - signed(N) >= 0 && 414 "Corrupted backtrack positions ?"); 415 CachedLexPos -= N; 416 } 417 418 /// EnterToken - Enters a token in the token stream to be lexed next. If 419 /// BackTrack() is called afterwards, the token will remain at the insertion 420 /// point. 421 void EnterToken(const Token &Tok) { 422 EnterCachingLexMode(); 423 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 424 } 425 426 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 427 /// tokens (because backtrack is enabled) it should replace the most recent 428 /// cached tokens with the given annotation token. This function has no effect 429 /// if backtracking is not enabled. 430 /// 431 /// Note that the use of this function is just for optimization; so that the 432 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 433 /// invoked. 434 void AnnotateCachedTokens(const Token &Tok) { 435 assert(Tok.isAnnotation() && "Expected annotation token"); 436 if (CachedLexPos != 0 && isBacktrackEnabled()) 437 AnnotatePreviousCachedTokens(Tok); 438 } 439 440 /// \brief Replace the last token with an annotation token. 441 /// 442 /// Like AnnotateCachedTokens(), this routine replaces an 443 /// already-parsed (and resolved) token with an annotation 444 /// token. However, this routine only replaces the last token with 445 /// the annotation token; it does not affect any other cached 446 /// tokens. This function has no effect if backtracking is not 447 /// enabled. 448 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 449 assert(Tok.isAnnotation() && "Expected annotation token"); 450 if (CachedLexPos != 0 && isBacktrackEnabled()) 451 CachedTokens[CachedLexPos-1] = Tok; 452 } 453 454 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 455 /// the specified Token's location, translating the token's start 456 /// position in the current buffer into a SourcePosition object for rendering. 457 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 458 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 459 } 460 461 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 462 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 463 DiagID); 464 } 465 466 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 467 /// token is the characters used to represent the token in the source file 468 /// after trigraph expansion and escaped-newline folding. In particular, this 469 /// wants to get the true, uncanonicalized, spelling of things like digraphs 470 /// UCNs, etc. 471 std::string getSpelling(const Token &Tok) const; 472 473 /// getSpelling - This method is used to get the spelling of a token into a 474 /// preallocated buffer, instead of as an std::string. The caller is required 475 /// to allocate enough space for the token, which is guaranteed to be at least 476 /// Tok.getLength() bytes long. The length of the actual result is returned. 477 /// 478 /// Note that this method may do two possible things: it may either fill in 479 /// the buffer specified with characters, or it may *change the input pointer* 480 /// to point to a constant buffer with the data already in it (avoiding a 481 /// copy). The caller is not allowed to modify the returned buffer pointer 482 /// if an internal buffer is returned. 483 unsigned getSpelling(const Token &Tok, const char *&Buffer) const; 484 485 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 486 /// with length 1, return the character. 487 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok) const { 488 assert(Tok.is(tok::numeric_constant) && 489 Tok.getLength() == 1 && "Called on unsupported token"); 490 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 491 492 // If the token is carrying a literal data pointer, just use it. 493 if (const char *D = Tok.getLiteralData()) 494 return *D; 495 496 // Otherwise, fall back on getCharacterData, which is slower, but always 497 // works. 498 return *SourceMgr.getCharacterData(Tok.getLocation()); 499 } 500 501 /// CreateString - Plop the specified string into a scratch buffer and set the 502 /// specified token's location and length to it. If specified, the source 503 /// location provides a location of the instantiation point of the token. 504 void CreateString(const char *Buf, unsigned Len, 505 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 506 507 /// \brief Computes the source location just past the end of the 508 /// token at this source location. 509 /// 510 /// This routine can be used to produce a source location that 511 /// points just past the end of the token referenced by \p Loc, and 512 /// is generally used when a diagnostic needs to point just after a 513 /// token where it expected something different that it received. If 514 /// the returned source location would not be meaningful (e.g., if 515 /// it points into a macro), this routine returns an invalid 516 /// source location. 517 SourceLocation getLocForEndOfToken(SourceLocation Loc); 518 519 /// DumpToken - Print the token to stderr, used for debugging. 520 /// 521 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 522 void DumpLocation(SourceLocation Loc) const; 523 void DumpMacro(const MacroInfo &MI) const; 524 525 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 526 /// token, return a new location that specifies a character within the token. 527 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 528 529 /// IncrementPasteCounter - Increment the counters for the number of token 530 /// paste operations performed. If fast was specified, this is a 'fast paste' 531 /// case we handled. 532 /// 533 void IncrementPasteCounter(bool isFast) { 534 if (isFast) 535 ++NumFastTokenPaste; 536 else 537 ++NumTokenPaste; 538 } 539 540 void PrintStats(); 541 542 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 543 /// comment (/##/) in microsoft mode, this method handles updating the current 544 /// state, returning the token on the next source line. 545 void HandleMicrosoftCommentPaste(Token &Tok); 546 547 //===--------------------------------------------------------------------===// 548 // Preprocessor callback methods. These are invoked by a lexer as various 549 // directives and events are found. 550 551 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 552 /// identifier information for the token and install it into the token. 553 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 554 const char *BufPtr = 0); 555 556 /// HandleIdentifier - This callback is invoked when the lexer reads an 557 /// identifier and has filled in the tokens IdentifierInfo member. This 558 /// callback potentially macro expands it or turns it into a named token (like 559 /// 'for'). 560 void HandleIdentifier(Token &Identifier); 561 562 563 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 564 /// the current file. This either returns the EOF token and returns true, or 565 /// pops a level off the include stack and returns false, at which point the 566 /// client should call lex again. 567 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 568 569 /// HandleEndOfTokenLexer - This callback is invoked when the current 570 /// TokenLexer hits the end of its token stream. 571 bool HandleEndOfTokenLexer(Token &Result); 572 573 /// HandleDirective - This callback is invoked when the lexer sees a # token 574 /// at the start of a line. This consumes the directive, modifies the 575 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 576 /// read is the correct one. 577 void HandleDirective(Token &Result); 578 579 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 580 /// not, emit a diagnostic and consume up until the eom. 581 void CheckEndOfDirective(const char *Directive); 582 583 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 584 /// current line until the tok::eom token is found. 585 void DiscardUntilEndOfDirective(); 586 587private: 588 589 void PushIncludeMacroStack() { 590 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 591 CurPTHLexer.take(), 592 CurPPLexer, 593 CurTokenLexer.take(), 594 CurDirLookup)); 595 CurPPLexer = 0; 596 } 597 598 void PopIncludeMacroStack() { 599 CurLexer.reset(IncludeMacroStack.back().TheLexer); 600 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 601 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 602 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 603 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 604 IncludeMacroStack.pop_back(); 605 } 606 607 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 608 /// SourceLocation. 609 MacroInfo* AllocateMacroInfo(SourceLocation L); 610 611 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 612 /// be reused for allocating new MacroInfo objects. 613 void ReleaseMacroInfo(MacroInfo* MI); 614 615 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 616 /// #include. 617 bool isInPrimaryFile() const; 618 619 /// ReadMacroName - Lex and validate a macro name, which occurs after a 620 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 621 /// and discards the rest of the macro line if the macro name is invalid. 622 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 623 624 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 625 /// definition has just been read. Lex the rest of the arguments and the 626 /// closing ), updating MI with what we learn. Return true if an error occurs 627 /// parsing the arg list. 628 bool ReadMacroDefinitionArgList(MacroInfo *MI); 629 630 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 631 /// decided that the subsequent tokens are in the #if'd out portion of the 632 /// file. Lex the rest of the file, until we see an #endif. If 633 /// FoundNonSkipPortion is true, then we have already emitted code for part of 634 /// this #if directive, so #else/#elif blocks should never be entered. If 635 /// FoundElse is false, then #else directives are ok, if not, then we have 636 /// already seen one so a #else directive is a duplicate. When this returns, 637 /// the caller can lex the first valid token. 638 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 639 bool FoundNonSkipPortion, bool FoundElse); 640 641 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 642 /// SkipExcludedConditionalBlock. 643 void PTHSkipExcludedConditionalBlock(); 644 645 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 646 /// may occur after a #if or #elif directive and return it as a bool. If the 647 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 648 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 649 650 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 651 /// #pragma GCC poison/system_header/dependency and #pragma once. 652 void RegisterBuiltinPragmas(); 653 654 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 655 /// identifier table. 656 void RegisterBuiltinMacros(); 657 IdentifierInfo *RegisterBuiltinMacro(const char *Name); 658 659 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 660 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 661 /// the macro should not be expanded return true, otherwise return false. 662 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 663 664 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 665 /// lexed is a '('. If so, consume the token and return true, if not, this 666 /// method should have no observable side-effect on the lexed tokens. 667 bool isNextPPTokenLParen(); 668 669 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 670 /// invoked to read all of the formal arguments specified for the macro 671 /// invocation. This returns null on error. 672 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 673 SourceLocation &InstantiationEnd); 674 675 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 676 /// as a builtin macro, handle it and return the next token as 'Tok'. 677 void ExpandBuiltinMacro(Token &Tok); 678 679 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 680 /// return the first token after the directive. The _Pragma token has just 681 /// been read into 'Tok'. 682 void Handle_Pragma(Token &Tok); 683 684 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 685 /// start lexing tokens from it instead of the current buffer. 686 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 687 688 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 689 /// start getting tokens from it using the PTH cache. 690 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 691 692 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 693 /// checked and spelled filename, e.g. as an operand of #include. This returns 694 /// true if the input filename was in <>'s or false if it were in ""'s. The 695 /// caller is expected to provide a buffer that is large enough to hold the 696 /// spelling of the filename, but is also expected to handle the case when 697 /// this method decides to use a different buffer. 698 bool GetIncludeFilenameSpelling(SourceLocation Loc, 699 const char *&BufStart, const char *&BufEnd); 700 701 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 702 /// return null on failure. isAngled indicates whether the file reference is 703 /// for system #include's or not (i.e. using <> instead of ""). 704 const FileEntry *LookupFile(const char *FilenameStart,const char *FilenameEnd, 705 bool isAngled, const DirectoryLookup *FromDir, 706 const DirectoryLookup *&CurDir); 707 708 709 710 /// IsFileLexer - Returns true if we are lexing from a file and not a 711 /// pragma or a macro. 712 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 713 return L ? !L->isPragmaLexer() : P != 0; 714 } 715 716 static bool IsFileLexer(const IncludeStackInfo& I) { 717 return IsFileLexer(I.TheLexer, I.ThePPLexer); 718 } 719 720 bool IsFileLexer() const { 721 return IsFileLexer(CurLexer.get(), CurPPLexer); 722 } 723 724 //===--------------------------------------------------------------------===// 725 // Caching stuff. 726 void CachingLex(Token &Result); 727 bool InCachingLexMode() const { return CurPPLexer == 0 && CurTokenLexer == 0;} 728 void EnterCachingLexMode(); 729 void ExitCachingLexMode() { 730 if (InCachingLexMode()) 731 RemoveTopOfLexerStack(); 732 } 733 const Token &PeekAhead(unsigned N); 734 void AnnotatePreviousCachedTokens(const Token &Tok); 735 736 //===--------------------------------------------------------------------===// 737 /// Handle*Directive - implement the various preprocessor directives. These 738 /// should side-effect the current preprocessor object so that the next call 739 /// to Lex() will return the appropriate token next. 740 void HandleLineDirective(Token &Tok); 741 void HandleDigitDirective(Token &Tok); 742 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 743 void HandleIdentSCCSDirective(Token &Tok); 744 745 // File inclusion. 746 void HandleIncludeDirective(Token &Tok, 747 const DirectoryLookup *LookupFrom = 0, 748 bool isImport = false); 749 void HandleIncludeNextDirective(Token &Tok); 750 void HandleImportDirective(Token &Tok); 751 752 // Macro handling. 753 void HandleDefineDirective(Token &Tok); 754 void HandleUndefDirective(Token &Tok); 755 // HandleAssertDirective(Token &Tok); 756 // HandleUnassertDirective(Token &Tok); 757 758 // Conditional Inclusion. 759 void HandleIfdefDirective(Token &Tok, bool isIfndef, 760 bool ReadAnyTokensBeforeDirective); 761 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 762 void HandleEndifDirective(Token &Tok); 763 void HandleElseDirective(Token &Tok); 764 void HandleElifDirective(Token &Tok); 765 766 // Pragmas. 767 void HandlePragmaDirective(); 768public: 769 void HandlePragmaOnce(Token &OnceTok); 770 void HandlePragmaMark(); 771 void HandlePragmaPoison(Token &PoisonTok); 772 void HandlePragmaSystemHeader(Token &SysHeaderTok); 773 void HandlePragmaDependency(Token &DependencyTok); 774 void HandlePragmaComment(Token &CommentTok); 775}; 776 777/// PreprocessorFactory - A generic factory interface for lazily creating 778/// Preprocessor objects on-demand when they are needed. 779class PreprocessorFactory { 780public: 781 virtual ~PreprocessorFactory(); 782 virtual Preprocessor* CreatePreprocessor() = 0; 783}; 784 785} // end namespace clang 786 787#endif 788