Preprocessor.h revision f4d5eb4866a27d497f0bb75b12c2ffd48ad4d9c0
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/Builtins.h" 23#include "clang/Basic/Diagnostic.h" 24#include "clang/Basic/IdentifierTable.h" 25#include "clang/Basic/SourceLocation.h" 26#include "llvm/ADT/DenseMap.h" 27#include "llvm/ADT/OwningPtr.h" 28#include "llvm/ADT/SmallVector.h" 29#include "llvm/Support/Allocator.h" 30#include <vector> 31 32namespace clang { 33 34class SourceManager; 35class FileManager; 36class FileEntry; 37class HeaderSearch; 38class PragmaNamespace; 39class PragmaHandler; 40class CommentHandler; 41class ScratchBuffer; 42class TargetInfo; 43class PPCallbacks; 44class DirectoryLookup; 45 46/// Preprocessor - This object engages in a tight little dance with the lexer to 47/// efficiently preprocess tokens. Lexers know only about tokens within a 48/// single source file, and don't know anything about preprocessor-level issues 49/// like the #include stack, token expansion, etc. 50/// 51class Preprocessor { 52 Diagnostic *Diags; 53 LangOptions Features; 54 TargetInfo &Target; 55 FileManager &FileMgr; 56 SourceManager &SourceMgr; 57 ScratchBuffer *ScratchBuf; 58 HeaderSearch &HeaderInfo; 59 60 /// PTH - An optional PTHManager object used for getting tokens from 61 /// a token cache rather than lexing the original source file. 62 llvm::OwningPtr<PTHManager> PTH; 63 64 /// BP - A BumpPtrAllocator object used to quickly allocate and release 65 /// objects internal to the Preprocessor. 66 llvm::BumpPtrAllocator BP; 67 68 /// Identifiers for builtin macros and other builtins. 69 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 70 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 71 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 72 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 73 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 74 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 75 IdentifierInfo *Ident_Pragma, *Ident__VA_ARGS__; // _Pragma, __VA_ARGS__ 76 IdentifierInfo *Ident__has_feature; // __has_feature 77 IdentifierInfo *Ident__has_builtin; // __has_builtin 78 79 SourceLocation DATELoc, TIMELoc; 80 unsigned CounterValue; // Next __COUNTER__ value. 81 82 enum { 83 /// MaxIncludeStackDepth - Maximum depth of #includes. 84 MaxAllowedIncludeStackDepth = 200 85 }; 86 87 // State that is set before the preprocessor begins. 88 bool KeepComments : 1; 89 bool KeepMacroComments : 1; 90 91 // State that changes while the preprocessor runs: 92 bool DisableMacroExpansion : 1; // True if macro expansion is disabled. 93 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 94 95 /// Identifiers - This is mapping/lookup information for all identifiers in 96 /// the program, including program keywords. 97 IdentifierTable Identifiers; 98 99 /// Selectors - This table contains all the selectors in the program. Unlike 100 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 101 /// It is declared/instantiated here because it's role/lifetime is 102 /// conceptually similar the IdentifierTable. In addition, the current control 103 /// flow (in clang::ParseAST()), make it convenient to put here. 104 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 105 /// the lifetime fo the preprocessor. 106 SelectorTable Selectors; 107 108 /// BuiltinInfo - Information about builtins. 109 Builtin::Context BuiltinInfo; 110 111 /// PragmaHandlers - This tracks all of the pragmas that the client registered 112 /// with this preprocessor. 113 PragmaNamespace *PragmaHandlers; 114 115 /// \brief Tracks all of the comment handlers that the client registered 116 /// with this preprocessor. 117 std::vector<CommentHandler *> CommentHandlers; 118 119 /// CurLexer - This is the current top of the stack that we're lexing from if 120 /// not expanding a macro and we are lexing directly from source code. 121 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 122 llvm::OwningPtr<Lexer> CurLexer; 123 124 /// CurPTHLexer - This is the current top of stack that we're lexing from if 125 /// not expanding from a macro and we are lexing from a PTH cache. 126 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 127 llvm::OwningPtr<PTHLexer> CurPTHLexer; 128 129 /// CurPPLexer - This is the current top of the stack what we're lexing from 130 /// if not expanding a macro. This is an alias for either CurLexer or 131 /// CurPTHLexer. 132 PreprocessorLexer* CurPPLexer; 133 134 /// CurLookup - The DirectoryLookup structure used to find the current 135 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 136 /// implement #include_next and find directory-specific properties. 137 const DirectoryLookup *CurDirLookup; 138 139 /// CurTokenLexer - This is the current macro we are expanding, if we are 140 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 141 llvm::OwningPtr<TokenLexer> CurTokenLexer; 142 143 /// IncludeMacroStack - This keeps track of the stack of files currently 144 /// #included, and macros currently being expanded from, not counting 145 /// CurLexer/CurTokenLexer. 146 struct IncludeStackInfo { 147 Lexer *TheLexer; 148 PTHLexer *ThePTHLexer; 149 PreprocessorLexer *ThePPLexer; 150 TokenLexer *TheTokenLexer; 151 const DirectoryLookup *TheDirLookup; 152 153 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 154 TokenLexer* TL, const DirectoryLookup *D) 155 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 156 TheDirLookup(D) {} 157 }; 158 std::vector<IncludeStackInfo> IncludeMacroStack; 159 160 /// Callbacks - These are actions invoked when some preprocessor activity is 161 /// encountered (e.g. a file is #included, etc). 162 PPCallbacks *Callbacks; 163 164 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 165 /// to the actual definition of the macro. 166 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 167 168 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 169 /// allocation. 170 std::vector<MacroInfo*> MICache; 171 172 // Various statistics we track for performance analysis. 173 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 174 unsigned NumIf, NumElse, NumEndif; 175 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 176 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 177 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 178 unsigned NumSkipped; 179 180 /// Predefines - This string is the predefined macros that preprocessor 181 /// should use from the command line etc. 182 std::string Predefines; 183 184 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 185 enum { TokenLexerCacheSize = 8 }; 186 unsigned NumCachedTokenLexers; 187 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 188 189private: // Cached tokens state. 190 typedef llvm::SmallVector<Token, 1> CachedTokensTy; 191 192 /// CachedTokens - Cached tokens are stored here when we do backtracking or 193 /// lookahead. They are "lexed" by the CachingLex() method. 194 CachedTokensTy CachedTokens; 195 196 /// CachedLexPos - The position of the cached token that CachingLex() should 197 /// "lex" next. If it points beyond the CachedTokens vector, it means that 198 /// a normal Lex() should be invoked. 199 CachedTokensTy::size_type CachedLexPos; 200 201 /// BacktrackPositions - Stack of backtrack positions, allowing nested 202 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 203 /// indicate where CachedLexPos should be set when the BackTrack() method is 204 /// invoked (at which point the last position is popped). 205 std::vector<CachedTokensTy::size_type> BacktrackPositions; 206 207public: 208 Preprocessor(Diagnostic &diags, const LangOptions &opts, TargetInfo &target, 209 SourceManager &SM, HeaderSearch &Headers, 210 IdentifierInfoLookup *IILookup = 0); 211 212 ~Preprocessor(); 213 214 Diagnostic &getDiagnostics() const { return *Diags; } 215 void setDiagnostics(Diagnostic &D) { Diags = &D; } 216 217 const LangOptions &getLangOptions() const { return Features; } 218 TargetInfo &getTargetInfo() const { return Target; } 219 FileManager &getFileManager() const { return FileMgr; } 220 SourceManager &getSourceManager() const { return SourceMgr; } 221 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 222 223 IdentifierTable &getIdentifierTable() { return Identifiers; } 224 SelectorTable &getSelectorTable() { return Selectors; } 225 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 226 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 227 228 void setPTHManager(PTHManager* pm); 229 230 PTHManager *getPTHManager() { return PTH.get(); } 231 232 /// SetCommentRetentionState - Control whether or not the preprocessor retains 233 /// comments in output. 234 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 235 this->KeepComments = KeepComments | KeepMacroComments; 236 this->KeepMacroComments = KeepMacroComments; 237 } 238 239 bool getCommentRetentionState() const { return KeepComments; } 240 241 /// isCurrentLexer - Return true if we are lexing directly from the specified 242 /// lexer. 243 bool isCurrentLexer(const PreprocessorLexer *L) const { 244 return CurPPLexer == L; 245 } 246 247 /// getCurrentLexer - Return the current file lexer being lexed from. Note 248 /// that this ignores any potentially active macro expansions and _Pragma 249 /// expansions going on at the time. 250 PreprocessorLexer *getCurrentFileLexer() const; 251 252 /// getPPCallbacks/setPPCallbacks - Accessors for preprocessor callbacks. 253 /// Note that this class takes ownership of any PPCallbacks object given to 254 /// it. 255 PPCallbacks *getPPCallbacks() const { return Callbacks; } 256 void setPPCallbacks(PPCallbacks *C) { 257 if (Callbacks) 258 C = new PPChainedCallbacks(C, Callbacks); 259 Callbacks = C; 260 } 261 262 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 263 /// or null if it isn't #define'd. 264 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 265 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 266 } 267 268 /// setMacroInfo - Specify a macro for this identifier. 269 /// 270 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 271 272 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 273 /// state of the macro table. This visits every currently-defined macro. 274 typedef llvm::DenseMap<IdentifierInfo*, 275 MacroInfo*>::const_iterator macro_iterator; 276 macro_iterator macro_begin() const { return Macros.begin(); } 277 macro_iterator macro_end() const { return Macros.end(); } 278 279 280 281 const std::string &getPredefines() const { return Predefines; } 282 /// setPredefines - Set the predefines for this Preprocessor. These 283 /// predefines are automatically injected when parsing the main file. 284 void setPredefines(const char *P) { Predefines = P; } 285 void setPredefines(const std::string &P) { Predefines = P; } 286 287 /// getIdentifierInfo - Return information about the specified preprocessor 288 /// identifier token. The version of this method that takes two character 289 /// pointers is preferred unless the identifier is already available as a 290 /// string (this avoids allocation and copying of memory to construct an 291 /// std::string). 292 IdentifierInfo *getIdentifierInfo(const char *NameStart, 293 const char *NameEnd) { 294 return &Identifiers.get(NameStart, NameEnd); 295 } 296 IdentifierInfo *getIdentifierInfo(const char *NameStr) { 297 return getIdentifierInfo(NameStr, NameStr+strlen(NameStr)); 298 } 299 300 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 301 /// If 'Namespace' is non-null, then it is a token required to exist on the 302 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 303 void AddPragmaHandler(const char *Namespace, PragmaHandler *Handler); 304 305 /// RemovePragmaHandler - Remove the specific pragma handler from 306 /// the preprocessor. If \arg Namespace is non-null, then it should 307 /// be the namespace that \arg Handler was added to. It is an error 308 /// to remove a handler that has not been registered. 309 void RemovePragmaHandler(const char *Namespace, PragmaHandler *Handler); 310 311 /// \brief Add the specified comment handler to the preprocessor. 312 void AddCommentHandler(CommentHandler *Handler); 313 314 /// \brief Remove the specified comment handler. 315 /// 316 /// It is an error to remove a handler that has not been registered. 317 void RemoveCommentHandler(CommentHandler *Handler); 318 319 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 320 /// which implicitly adds the builtin defines etc. 321 void EnterMainSourceFile(); 322 323 /// EnterSourceFile - Add a source file to the top of the include stack and 324 /// start lexing tokens from it instead of the current buffer. If isMainFile 325 /// is true, this is the main file for the translation unit. 326 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir); 327 328 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 329 /// tokens from it instead of the current buffer. Args specifies the 330 /// tokens input to a function-like macro. 331 /// 332 /// ILEnd specifies the location of the ')' for a function-like macro or the 333 /// identifier for an object-like macro. 334 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 335 336 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 337 /// which will cause the lexer to start returning the specified tokens. 338 /// 339 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 340 /// not be subject to further macro expansion. Otherwise, these tokens will 341 /// be re-macro-expanded when/if expansion is enabled. 342 /// 343 /// If OwnsTokens is false, this method assumes that the specified stream of 344 /// tokens has a permanent owner somewhere, so they do not need to be copied. 345 /// If it is true, it assumes the array of tokens is allocated with new[] and 346 /// must be freed. 347 /// 348 void EnterTokenStream(const Token *Toks, unsigned NumToks, 349 bool DisableMacroExpansion, bool OwnsTokens); 350 351 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 352 /// lexer stack. This should only be used in situations where the current 353 /// state of the top-of-stack lexer is known. 354 void RemoveTopOfLexerStack(); 355 356 /// EnableBacktrackAtThisPos - From the point that this method is called, and 357 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 358 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 359 /// make the Preprocessor re-lex the same tokens. 360 /// 361 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 362 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 363 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 364 /// 365 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 366 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 367 /// tokens will continue indefinitely. 368 /// 369 void EnableBacktrackAtThisPos(); 370 371 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 372 void CommitBacktrackedTokens(); 373 374 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 375 /// EnableBacktrackAtThisPos() was previously called. 376 void Backtrack(); 377 378 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 379 /// caching of tokens is on. 380 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 381 382 /// Lex - To lex a token from the preprocessor, just pull a token from the 383 /// current lexer or macro object. 384 void Lex(Token &Result) { 385 if (CurLexer) 386 CurLexer->Lex(Result); 387 else if (CurPTHLexer) 388 CurPTHLexer->Lex(Result); 389 else if (CurTokenLexer) 390 CurTokenLexer->Lex(Result); 391 else 392 CachingLex(Result); 393 } 394 395 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 396 /// something not a comment. This is useful in -E -C mode where comments 397 /// would foul up preprocessor directive handling. 398 void LexNonComment(Token &Result) { 399 do 400 Lex(Result); 401 while (Result.getKind() == tok::comment); 402 } 403 404 /// LexUnexpandedToken - This is just like Lex, but this disables macro 405 /// expansion of identifier tokens. 406 void LexUnexpandedToken(Token &Result) { 407 // Disable macro expansion. 408 bool OldVal = DisableMacroExpansion; 409 DisableMacroExpansion = true; 410 // Lex the token. 411 Lex(Result); 412 413 // Reenable it. 414 DisableMacroExpansion = OldVal; 415 } 416 417 /// LookAhead - This peeks ahead N tokens and returns that token without 418 /// consuming any tokens. LookAhead(0) returns the next token that would be 419 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 420 /// returns normal tokens after phase 5. As such, it is equivalent to using 421 /// 'Lex', not 'LexUnexpandedToken'. 422 const Token &LookAhead(unsigned N) { 423 if (CachedLexPos + N < CachedTokens.size()) 424 return CachedTokens[CachedLexPos+N]; 425 else 426 return PeekAhead(N+1); 427 } 428 429 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 430 /// this allows to revert a specific number of tokens. 431 /// Note that the number of tokens being reverted should be up to the last 432 /// backtrack position, not more. 433 void RevertCachedTokens(unsigned N) { 434 assert(isBacktrackEnabled() && 435 "Should only be called when tokens are cached for backtracking"); 436 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 437 && "Should revert tokens up to the last backtrack position, not more"); 438 assert(signed(CachedLexPos) - signed(N) >= 0 && 439 "Corrupted backtrack positions ?"); 440 CachedLexPos -= N; 441 } 442 443 /// EnterToken - Enters a token in the token stream to be lexed next. If 444 /// BackTrack() is called afterwards, the token will remain at the insertion 445 /// point. 446 void EnterToken(const Token &Tok) { 447 EnterCachingLexMode(); 448 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 449 } 450 451 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 452 /// tokens (because backtrack is enabled) it should replace the most recent 453 /// cached tokens with the given annotation token. This function has no effect 454 /// if backtracking is not enabled. 455 /// 456 /// Note that the use of this function is just for optimization; so that the 457 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 458 /// invoked. 459 void AnnotateCachedTokens(const Token &Tok) { 460 assert(Tok.isAnnotation() && "Expected annotation token"); 461 if (CachedLexPos != 0 && isBacktrackEnabled()) 462 AnnotatePreviousCachedTokens(Tok); 463 } 464 465 /// \brief Replace the last token with an annotation token. 466 /// 467 /// Like AnnotateCachedTokens(), this routine replaces an 468 /// already-parsed (and resolved) token with an annotation 469 /// token. However, this routine only replaces the last token with 470 /// the annotation token; it does not affect any other cached 471 /// tokens. This function has no effect if backtracking is not 472 /// enabled. 473 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 474 assert(Tok.isAnnotation() && "Expected annotation token"); 475 if (CachedLexPos != 0 && isBacktrackEnabled()) 476 CachedTokens[CachedLexPos-1] = Tok; 477 } 478 479 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 480 /// the specified Token's location, translating the token's start 481 /// position in the current buffer into a SourcePosition object for rendering. 482 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 483 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 484 } 485 486 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 487 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 488 DiagID); 489 } 490 491 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 492 /// token is the characters used to represent the token in the source file 493 /// after trigraph expansion and escaped-newline folding. In particular, this 494 /// wants to get the true, uncanonicalized, spelling of things like digraphs 495 /// UCNs, etc. 496 std::string getSpelling(const Token &Tok) const; 497 498 /// getSpelling - This method is used to get the spelling of a token into a 499 /// preallocated buffer, instead of as an std::string. The caller is required 500 /// to allocate enough space for the token, which is guaranteed to be at least 501 /// Tok.getLength() bytes long. The length of the actual result is returned. 502 /// 503 /// Note that this method may do two possible things: it may either fill in 504 /// the buffer specified with characters, or it may *change the input pointer* 505 /// to point to a constant buffer with the data already in it (avoiding a 506 /// copy). The caller is not allowed to modify the returned buffer pointer 507 /// if an internal buffer is returned. 508 unsigned getSpelling(const Token &Tok, const char *&Buffer) const; 509 510 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 511 /// with length 1, return the character. 512 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok) const { 513 assert(Tok.is(tok::numeric_constant) && 514 Tok.getLength() == 1 && "Called on unsupported token"); 515 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 516 517 // If the token is carrying a literal data pointer, just use it. 518 if (const char *D = Tok.getLiteralData()) 519 return *D; 520 521 // Otherwise, fall back on getCharacterData, which is slower, but always 522 // works. 523 return *SourceMgr.getCharacterData(Tok.getLocation()); 524 } 525 526 /// CreateString - Plop the specified string into a scratch buffer and set the 527 /// specified token's location and length to it. If specified, the source 528 /// location provides a location of the instantiation point of the token. 529 void CreateString(const char *Buf, unsigned Len, 530 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 531 532 /// \brief Computes the source location just past the end of the 533 /// token at this source location. 534 /// 535 /// This routine can be used to produce a source location that 536 /// points just past the end of the token referenced by \p Loc, and 537 /// is generally used when a diagnostic needs to point just after a 538 /// token where it expected something different that it received. If 539 /// the returned source location would not be meaningful (e.g., if 540 /// it points into a macro), this routine returns an invalid 541 /// source location. 542 SourceLocation getLocForEndOfToken(SourceLocation Loc); 543 544 /// DumpToken - Print the token to stderr, used for debugging. 545 /// 546 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 547 void DumpLocation(SourceLocation Loc) const; 548 void DumpMacro(const MacroInfo &MI) const; 549 550 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 551 /// token, return a new location that specifies a character within the token. 552 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 553 554 /// IncrementPasteCounter - Increment the counters for the number of token 555 /// paste operations performed. If fast was specified, this is a 'fast paste' 556 /// case we handled. 557 /// 558 void IncrementPasteCounter(bool isFast) { 559 if (isFast) 560 ++NumFastTokenPaste; 561 else 562 ++NumTokenPaste; 563 } 564 565 void PrintStats(); 566 567 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 568 /// comment (/##/) in microsoft mode, this method handles updating the current 569 /// state, returning the token on the next source line. 570 void HandleMicrosoftCommentPaste(Token &Tok); 571 572 //===--------------------------------------------------------------------===// 573 // Preprocessor callback methods. These are invoked by a lexer as various 574 // directives and events are found. 575 576 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 577 /// identifier information for the token and install it into the token. 578 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 579 const char *BufPtr = 0); 580 581 /// HandleIdentifier - This callback is invoked when the lexer reads an 582 /// identifier and has filled in the tokens IdentifierInfo member. This 583 /// callback potentially macro expands it or turns it into a named token (like 584 /// 'for'). 585 void HandleIdentifier(Token &Identifier); 586 587 588 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 589 /// the current file. This either returns the EOF token and returns true, or 590 /// pops a level off the include stack and returns false, at which point the 591 /// client should call lex again. 592 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 593 594 /// HandleEndOfTokenLexer - This callback is invoked when the current 595 /// TokenLexer hits the end of its token stream. 596 bool HandleEndOfTokenLexer(Token &Result); 597 598 /// HandleDirective - This callback is invoked when the lexer sees a # token 599 /// at the start of a line. This consumes the directive, modifies the 600 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 601 /// read is the correct one. 602 void HandleDirective(Token &Result); 603 604 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 605 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 606 /// true, then we consider macros that expand to zero tokens as being ok. 607 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 608 609 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 610 /// current line until the tok::eom token is found. 611 void DiscardUntilEndOfDirective(); 612 613 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 614 /// __DATE__ or __TIME__ in the file so far. 615 bool SawDateOrTime() const { 616 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 617 } 618 unsigned getCounterValue() const { return CounterValue; } 619 void setCounterValue(unsigned V) { CounterValue = V; } 620 621 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 622 /// SourceLocation. 623 MacroInfo* AllocateMacroInfo(SourceLocation L); 624 625private: 626 627 void PushIncludeMacroStack() { 628 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 629 CurPTHLexer.take(), 630 CurPPLexer, 631 CurTokenLexer.take(), 632 CurDirLookup)); 633 CurPPLexer = 0; 634 } 635 636 void PopIncludeMacroStack() { 637 CurLexer.reset(IncludeMacroStack.back().TheLexer); 638 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 639 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 640 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 641 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 642 IncludeMacroStack.pop_back(); 643 } 644 645 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 646 /// be reused for allocating new MacroInfo objects. 647 void ReleaseMacroInfo(MacroInfo* MI); 648 649 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 650 /// #include. 651 bool isInPrimaryFile() const; 652 653 /// ReadMacroName - Lex and validate a macro name, which occurs after a 654 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 655 /// and discards the rest of the macro line if the macro name is invalid. 656 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 657 658 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 659 /// definition has just been read. Lex the rest of the arguments and the 660 /// closing ), updating MI with what we learn. Return true if an error occurs 661 /// parsing the arg list. 662 bool ReadMacroDefinitionArgList(MacroInfo *MI); 663 664 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 665 /// decided that the subsequent tokens are in the #if'd out portion of the 666 /// file. Lex the rest of the file, until we see an #endif. If 667 /// FoundNonSkipPortion is true, then we have already emitted code for part of 668 /// this #if directive, so #else/#elif blocks should never be entered. If 669 /// FoundElse is false, then #else directives are ok, if not, then we have 670 /// already seen one so a #else directive is a duplicate. When this returns, 671 /// the caller can lex the first valid token. 672 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 673 bool FoundNonSkipPortion, bool FoundElse); 674 675 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 676 /// SkipExcludedConditionalBlock. 677 void PTHSkipExcludedConditionalBlock(); 678 679 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 680 /// may occur after a #if or #elif directive and return it as a bool. If the 681 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 682 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 683 684 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 685 /// #pragma GCC poison/system_header/dependency and #pragma once. 686 void RegisterBuiltinPragmas(); 687 688 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 689 /// identifier table. 690 void RegisterBuiltinMacros(); 691 692 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 693 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 694 /// the macro should not be expanded return true, otherwise return false. 695 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 696 697 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 698 /// lexed is a '('. If so, consume the token and return true, if not, this 699 /// method should have no observable side-effect on the lexed tokens. 700 bool isNextPPTokenLParen(); 701 702 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 703 /// invoked to read all of the formal arguments specified for the macro 704 /// invocation. This returns null on error. 705 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 706 SourceLocation &InstantiationEnd); 707 708 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 709 /// as a builtin macro, handle it and return the next token as 'Tok'. 710 void ExpandBuiltinMacro(Token &Tok); 711 712 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 713 /// return the first token after the directive. The _Pragma token has just 714 /// been read into 'Tok'. 715 void Handle_Pragma(Token &Tok); 716 717 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 718 /// start lexing tokens from it instead of the current buffer. 719 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 720 721 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 722 /// start getting tokens from it using the PTH cache. 723 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 724 725 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 726 /// checked and spelled filename, e.g. as an operand of #include. This returns 727 /// true if the input filename was in <>'s or false if it were in ""'s. The 728 /// caller is expected to provide a buffer that is large enough to hold the 729 /// spelling of the filename, but is also expected to handle the case when 730 /// this method decides to use a different buffer. 731 bool GetIncludeFilenameSpelling(SourceLocation Loc, 732 const char *&BufStart, const char *&BufEnd); 733 734 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 735 /// return null on failure. isAngled indicates whether the file reference is 736 /// for system #include's or not (i.e. using <> instead of ""). 737 const FileEntry *LookupFile(const char *FilenameStart,const char *FilenameEnd, 738 bool isAngled, const DirectoryLookup *FromDir, 739 const DirectoryLookup *&CurDir); 740 741 742 743 /// IsFileLexer - Returns true if we are lexing from a file and not a 744 /// pragma or a macro. 745 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 746 return L ? !L->isPragmaLexer() : P != 0; 747 } 748 749 static bool IsFileLexer(const IncludeStackInfo& I) { 750 return IsFileLexer(I.TheLexer, I.ThePPLexer); 751 } 752 753 bool IsFileLexer() const { 754 return IsFileLexer(CurLexer.get(), CurPPLexer); 755 } 756 757 //===--------------------------------------------------------------------===// 758 // Caching stuff. 759 void CachingLex(Token &Result); 760 bool InCachingLexMode() const { return CurPPLexer == 0 && CurTokenLexer == 0;} 761 void EnterCachingLexMode(); 762 void ExitCachingLexMode() { 763 if (InCachingLexMode()) 764 RemoveTopOfLexerStack(); 765 } 766 const Token &PeekAhead(unsigned N); 767 void AnnotatePreviousCachedTokens(const Token &Tok); 768 769 //===--------------------------------------------------------------------===// 770 /// Handle*Directive - implement the various preprocessor directives. These 771 /// should side-effect the current preprocessor object so that the next call 772 /// to Lex() will return the appropriate token next. 773 void HandleLineDirective(Token &Tok); 774 void HandleDigitDirective(Token &Tok); 775 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 776 void HandleIdentSCCSDirective(Token &Tok); 777 778 // File inclusion. 779 void HandleIncludeDirective(Token &Tok, 780 const DirectoryLookup *LookupFrom = 0, 781 bool isImport = false); 782 void HandleIncludeNextDirective(Token &Tok); 783 void HandleIncludeMacrosDirective(Token &Tok); 784 void HandleImportDirective(Token &Tok); 785 786 // Macro handling. 787 void HandleDefineDirective(Token &Tok); 788 void HandleUndefDirective(Token &Tok); 789 // HandleAssertDirective(Token &Tok); 790 // HandleUnassertDirective(Token &Tok); 791 792 // Conditional Inclusion. 793 void HandleIfdefDirective(Token &Tok, bool isIfndef, 794 bool ReadAnyTokensBeforeDirective); 795 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 796 void HandleEndifDirective(Token &Tok); 797 void HandleElseDirective(Token &Tok); 798 void HandleElifDirective(Token &Tok); 799 800 // Pragmas. 801 void HandlePragmaDirective(); 802public: 803 void HandlePragmaOnce(Token &OnceTok); 804 void HandlePragmaMark(); 805 void HandlePragmaPoison(Token &PoisonTok); 806 void HandlePragmaSystemHeader(Token &SysHeaderTok); 807 void HandlePragmaDependency(Token &DependencyTok); 808 void HandlePragmaComment(Token &CommentTok); 809 void HandleComment(SourceRange Comment); 810}; 811 812/// PreprocessorFactory - A generic factory interface for lazily creating 813/// Preprocessor objects on-demand when they are needed. 814class PreprocessorFactory { 815public: 816 virtual ~PreprocessorFactory(); 817 virtual Preprocessor* CreatePreprocessor() = 0; 818}; 819 820/// \brief Abstract base class that describes a handler that will receive 821/// source ranges for each of the comments encountered in the source file. 822class CommentHandler { 823public: 824 virtual ~CommentHandler(); 825 826 virtual void HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 827}; 828 829} // end namespace clang 830 831#endif 832