Preprocessor.h revision 33e4e70c8c0a17e0ccb7465d96556b077a68ecb1
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/MacroInfo.h" 18#include "clang/Lex/Lexer.h" 19#include "clang/Lex/PTHLexer.h" 20#include "clang/Lex/PPCallbacks.h" 21#include "clang/Lex/TokenLexer.h" 22#include "clang/Lex/PTHManager.h" 23#include "clang/Basic/Builtins.h" 24#include "clang/Basic/Diagnostic.h" 25#include "clang/Basic/IdentifierTable.h" 26#include "clang/Basic/SourceLocation.h" 27#include "llvm/ADT/DenseMap.h" 28#include "llvm/ADT/OwningPtr.h" 29#include "llvm/ADT/SmallVector.h" 30#include "llvm/Support/Allocator.h" 31#include <vector> 32 33namespace clang { 34 35class SourceManager; 36class ExternalPreprocessorSource; 37class FileManager; 38class FileSystemOptions; 39class FileEntry; 40class HeaderSearch; 41class PragmaNamespace; 42class PragmaHandler; 43class CommentHandler; 44class ScratchBuffer; 45class TargetInfo; 46class PPCallbacks; 47class CodeCompletionHandler; 48class DirectoryLookup; 49class PreprocessingRecord; 50 51/// Preprocessor - This object engages in a tight little dance with the lexer to 52/// efficiently preprocess tokens. Lexers know only about tokens within a 53/// single source file, and don't know anything about preprocessor-level issues 54/// like the #include stack, token expansion, etc. 55/// 56class Preprocessor { 57 Diagnostic *Diags; 58 LangOptions Features; 59 const TargetInfo &Target; 60 FileManager &FileMgr; 61 const FileSystemOptions &FileSystemOpts; 62 SourceManager &SourceMgr; 63 ScratchBuffer *ScratchBuf; 64 HeaderSearch &HeaderInfo; 65 66 /// \brief External source of macros. 67 ExternalPreprocessorSource *ExternalSource; 68 69 /// PTH - An optional PTHManager object used for getting tokens from 70 /// a token cache rather than lexing the original source file. 71 llvm::OwningPtr<PTHManager> PTH; 72 73 /// BP - A BumpPtrAllocator object used to quickly allocate and release 74 /// objects internal to the Preprocessor. 75 llvm::BumpPtrAllocator BP; 76 77 /// Identifiers for builtin macros and other builtins. 78 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 79 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 80 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 81 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 82 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 83 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 84 IdentifierInfo *Ident_Pragma, *Ident__pragma; // _Pragma, __pragma 85 IdentifierInfo *Ident__VA_ARGS__; // __VA_ARGS__ 86 IdentifierInfo *Ident__has_feature; // __has_feature 87 IdentifierInfo *Ident__has_builtin; // __has_builtin 88 IdentifierInfo *Ident__has_attribute; // __has_attribute 89 IdentifierInfo *Ident__has_include; // __has_include 90 IdentifierInfo *Ident__has_include_next; // __has_include_next 91 92 SourceLocation DATELoc, TIMELoc; 93 unsigned CounterValue; // Next __COUNTER__ value. 94 95 enum { 96 /// MaxIncludeStackDepth - Maximum depth of #includes. 97 MaxAllowedIncludeStackDepth = 200 98 }; 99 100 // State that is set before the preprocessor begins. 101 bool KeepComments : 1; 102 bool KeepMacroComments : 1; 103 104 // State that changes while the preprocessor runs: 105 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 106 107 /// Whether the preprocessor owns the header search object. 108 bool OwnsHeaderSearch : 1; 109 110 /// DisableMacroExpansion - True if macro expansion is disabled. 111 bool DisableMacroExpansion : 1; 112 113 /// \brief Whether we have already loaded macros from the external source. 114 mutable bool ReadMacrosFromExternalSource : 1; 115 116 /// Identifiers - This is mapping/lookup information for all identifiers in 117 /// the program, including program keywords. 118 mutable IdentifierTable Identifiers; 119 120 /// Selectors - This table contains all the selectors in the program. Unlike 121 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 122 /// It is declared/instantiated here because it's role/lifetime is 123 /// conceptually similar the IdentifierTable. In addition, the current control 124 /// flow (in clang::ParseAST()), make it convenient to put here. 125 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 126 /// the lifetime of the preprocessor. 127 SelectorTable Selectors; 128 129 /// BuiltinInfo - Information about builtins. 130 Builtin::Context BuiltinInfo; 131 132 /// PragmaHandlers - This tracks all of the pragmas that the client registered 133 /// with this preprocessor. 134 PragmaNamespace *PragmaHandlers; 135 136 /// \brief Tracks all of the comment handlers that the client registered 137 /// with this preprocessor. 138 std::vector<CommentHandler *> CommentHandlers; 139 140 /// \brief The code-completion handler. 141 CodeCompletionHandler *CodeComplete; 142 143 /// \brief The file that we're performing code-completion for, if any. 144 const FileEntry *CodeCompletionFile; 145 146 /// \brief The number of bytes that we will initially skip when entering the 147 /// main file, which is used when loading a precompiled preamble, along 148 /// with a flag that indicates whether skipping this number of bytes will 149 /// place the lexer at the start of a line. 150 std::pair<unsigned, bool> SkipMainFilePreamble; 151 152 /// CurLexer - This is the current top of the stack that we're lexing from if 153 /// not expanding a macro and we are lexing directly from source code. 154 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 155 llvm::OwningPtr<Lexer> CurLexer; 156 157 /// CurPTHLexer - This is the current top of stack that we're lexing from if 158 /// not expanding from a macro and we are lexing from a PTH cache. 159 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 160 llvm::OwningPtr<PTHLexer> CurPTHLexer; 161 162 /// CurPPLexer - This is the current top of the stack what we're lexing from 163 /// if not expanding a macro. This is an alias for either CurLexer or 164 /// CurPTHLexer. 165 PreprocessorLexer *CurPPLexer; 166 167 /// CurLookup - The DirectoryLookup structure used to find the current 168 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 169 /// implement #include_next and find directory-specific properties. 170 const DirectoryLookup *CurDirLookup; 171 172 /// CurTokenLexer - This is the current macro we are expanding, if we are 173 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 174 llvm::OwningPtr<TokenLexer> CurTokenLexer; 175 176 /// IncludeMacroStack - This keeps track of the stack of files currently 177 /// #included, and macros currently being expanded from, not counting 178 /// CurLexer/CurTokenLexer. 179 struct IncludeStackInfo { 180 Lexer *TheLexer; 181 PTHLexer *ThePTHLexer; 182 PreprocessorLexer *ThePPLexer; 183 TokenLexer *TheTokenLexer; 184 const DirectoryLookup *TheDirLookup; 185 186 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 187 TokenLexer* TL, const DirectoryLookup *D) 188 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 189 TheDirLookup(D) {} 190 }; 191 std::vector<IncludeStackInfo> IncludeMacroStack; 192 193 /// Callbacks - These are actions invoked when some preprocessor activity is 194 /// encountered (e.g. a file is #included, etc). 195 PPCallbacks *Callbacks; 196 197 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 198 /// to the actual definition of the macro. 199 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 200 201 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 202 /// reused for quick allocation. 203 MacroArgs *MacroArgCache; 204 friend class MacroArgs; 205 206 /// PragmaPushMacroInfo - For each IdentifierInfo used in a #pragma 207 /// push_macro directive, we keep a MacroInfo stack used to restore 208 /// previous macro value. 209 llvm::DenseMap<IdentifierInfo*, std::vector<MacroInfo*> > PragmaPushMacroInfo; 210 211 // Various statistics we track for performance analysis. 212 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 213 unsigned NumIf, NumElse, NumEndif; 214 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 215 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 216 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 217 unsigned NumSkipped; 218 219 /// Predefines - This string is the predefined macros that preprocessor 220 /// should use from the command line etc. 221 std::string Predefines; 222 223 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 224 enum { TokenLexerCacheSize = 8 }; 225 unsigned NumCachedTokenLexers; 226 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 227 228 /// \brief A record of the macro definitions and instantiations that 229 /// occurred during preprocessing. 230 /// 231 /// This is an optional side structure that can be enabled with 232 /// \c createPreprocessingRecord() prior to preprocessing. 233 PreprocessingRecord *Record; 234 235private: // Cached tokens state. 236 typedef llvm::SmallVector<Token, 1> CachedTokensTy; 237 238 /// CachedTokens - Cached tokens are stored here when we do backtracking or 239 /// lookahead. They are "lexed" by the CachingLex() method. 240 CachedTokensTy CachedTokens; 241 242 /// CachedLexPos - The position of the cached token that CachingLex() should 243 /// "lex" next. If it points beyond the CachedTokens vector, it means that 244 /// a normal Lex() should be invoked. 245 CachedTokensTy::size_type CachedLexPos; 246 247 /// BacktrackPositions - Stack of backtrack positions, allowing nested 248 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 249 /// indicate where CachedLexPos should be set when the BackTrack() method is 250 /// invoked (at which point the last position is popped). 251 std::vector<CachedTokensTy::size_type> BacktrackPositions; 252 253 struct MacroInfoChain { 254 MacroInfo MI; 255 MacroInfoChain *Next; 256 MacroInfoChain *Prev; 257 }; 258 259 /// MacroInfos are managed as a chain for easy disposal. This is the head 260 /// of that list. 261 MacroInfoChain *MIChainHead; 262 263 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 264 /// allocation. 265 MacroInfoChain *MICache; 266 267 MacroInfo *getInfoForMacro(IdentifierInfo *II) const; 268 269public: 270 Preprocessor(Diagnostic &diags, const LangOptions &opts, 271 const TargetInfo &target, 272 SourceManager &SM, HeaderSearch &Headers, 273 IdentifierInfoLookup *IILookup = 0, 274 bool OwnsHeaderSearch = false); 275 276 ~Preprocessor(); 277 278 Diagnostic &getDiagnostics() const { return *Diags; } 279 void setDiagnostics(Diagnostic &D) { Diags = &D; } 280 281 const LangOptions &getLangOptions() const { return Features; } 282 const TargetInfo &getTargetInfo() const { return Target; } 283 FileManager &getFileManager() const { return FileMgr; } 284 const FileSystemOptions &getFileSystemOpts() const { return FileSystemOpts; } 285 SourceManager &getSourceManager() const { return SourceMgr; } 286 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 287 288 IdentifierTable &getIdentifierTable() { return Identifiers; } 289 SelectorTable &getSelectorTable() { return Selectors; } 290 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 291 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 292 293 void setPTHManager(PTHManager* pm); 294 295 PTHManager *getPTHManager() { return PTH.get(); } 296 297 void setExternalSource(ExternalPreprocessorSource *Source) { 298 ExternalSource = Source; 299 } 300 301 ExternalPreprocessorSource *getExternalSource() const { 302 return ExternalSource; 303 } 304 305 /// SetCommentRetentionState - Control whether or not the preprocessor retains 306 /// comments in output. 307 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 308 this->KeepComments = KeepComments | KeepMacroComments; 309 this->KeepMacroComments = KeepMacroComments; 310 } 311 312 bool getCommentRetentionState() const { return KeepComments; } 313 314 /// isCurrentLexer - Return true if we are lexing directly from the specified 315 /// lexer. 316 bool isCurrentLexer(const PreprocessorLexer *L) const { 317 return CurPPLexer == L; 318 } 319 320 /// getCurrentLexer - Return the current lexer being lexed from. Note 321 /// that this ignores any potentially active macro expansions and _Pragma 322 /// expansions going on at the time. 323 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 324 325 /// getCurrentFileLexer - Return the current file lexer being lexed from. 326 /// Note that this ignores any potentially active macro expansions and _Pragma 327 /// expansions going on at the time. 328 PreprocessorLexer *getCurrentFileLexer() const; 329 330 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 331 /// Note that this class takes ownership of any PPCallbacks object given to 332 /// it. 333 PPCallbacks *getPPCallbacks() const { return Callbacks; } 334 void addPPCallbacks(PPCallbacks *C) { 335 if (Callbacks) 336 C = new PPChainedCallbacks(C, Callbacks); 337 Callbacks = C; 338 } 339 340 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 341 /// or null if it isn't #define'd. 342 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 343 if (!II->hasMacroDefinition()) 344 return 0; 345 346 return getInfoForMacro(II); 347 } 348 349 /// setMacroInfo - Specify a macro for this identifier. 350 /// 351 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 352 353 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 354 /// state of the macro table. This visits every currently-defined macro. 355 typedef llvm::DenseMap<IdentifierInfo*, 356 MacroInfo*>::const_iterator macro_iterator; 357 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 358 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 359 360 const std::string &getPredefines() const { return Predefines; } 361 /// setPredefines - Set the predefines for this Preprocessor. These 362 /// predefines are automatically injected when parsing the main file. 363 void setPredefines(const char *P) { Predefines = P; } 364 void setPredefines(const std::string &P) { Predefines = P; } 365 366 /// getIdentifierInfo - Return information about the specified preprocessor 367 /// identifier token. The version of this method that takes two character 368 /// pointers is preferred unless the identifier is already available as a 369 /// string (this avoids allocation and copying of memory to construct an 370 /// std::string). 371 IdentifierInfo *getIdentifierInfo(llvm::StringRef Name) const { 372 return &Identifiers.get(Name); 373 } 374 375 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 376 /// If 'Namespace' is non-null, then it is a token required to exist on the 377 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 378 void AddPragmaHandler(llvm::StringRef Namespace, PragmaHandler *Handler); 379 void AddPragmaHandler(PragmaHandler *Handler) { 380 AddPragmaHandler(llvm::StringRef(), Handler); 381 } 382 383 /// RemovePragmaHandler - Remove the specific pragma handler from 384 /// the preprocessor. If \arg Namespace is non-null, then it should 385 /// be the namespace that \arg Handler was added to. It is an error 386 /// to remove a handler that has not been registered. 387 void RemovePragmaHandler(llvm::StringRef Namespace, PragmaHandler *Handler); 388 void RemovePragmaHandler(PragmaHandler *Handler) { 389 RemovePragmaHandler(llvm::StringRef(), Handler); 390 } 391 392 /// \brief Add the specified comment handler to the preprocessor. 393 void AddCommentHandler(CommentHandler *Handler); 394 395 /// \brief Remove the specified comment handler. 396 /// 397 /// It is an error to remove a handler that has not been registered. 398 void RemoveCommentHandler(CommentHandler *Handler); 399 400 /// \brief Set the code completion handler to the given object. 401 void setCodeCompletionHandler(CodeCompletionHandler &Handler) { 402 CodeComplete = &Handler; 403 } 404 405 /// \brief Retrieve the current code-completion handler. 406 CodeCompletionHandler *getCodeCompletionHandler() const { 407 return CodeComplete; 408 } 409 410 /// \brief Clear out the code completion handler. 411 void clearCodeCompletionHandler() { 412 CodeComplete = 0; 413 } 414 415 /// \brief Hook used by the lexer to invoke the "natural language" code 416 /// completion point. 417 void CodeCompleteNaturalLanguage(); 418 419 /// \brief Retrieve the preprocessing record, or NULL if there is no 420 /// preprocessing record. 421 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 422 423 /// \brief Create a new preprocessing record, which will keep track of 424 /// all macro expansions, macro definitions, etc. 425 void createPreprocessingRecord(); 426 427 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 428 /// which implicitly adds the builtin defines etc. 429 void EnterMainSourceFile(); 430 431 /// EndSourceFile - Inform the preprocessor callbacks that processing is 432 /// complete. 433 void EndSourceFile(); 434 435 /// EnterSourceFile - Add a source file to the top of the include stack and 436 /// start lexing tokens from it instead of the current buffer. Emit an error 437 /// and don't enter the file on error. 438 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 439 SourceLocation Loc); 440 441 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 442 /// tokens from it instead of the current buffer. Args specifies the 443 /// tokens input to a function-like macro. 444 /// 445 /// ILEnd specifies the location of the ')' for a function-like macro or the 446 /// identifier for an object-like macro. 447 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 448 449 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 450 /// which will cause the lexer to start returning the specified tokens. 451 /// 452 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 453 /// not be subject to further macro expansion. Otherwise, these tokens will 454 /// be re-macro-expanded when/if expansion is enabled. 455 /// 456 /// If OwnsTokens is false, this method assumes that the specified stream of 457 /// tokens has a permanent owner somewhere, so they do not need to be copied. 458 /// If it is true, it assumes the array of tokens is allocated with new[] and 459 /// must be freed. 460 /// 461 void EnterTokenStream(const Token *Toks, unsigned NumToks, 462 bool DisableMacroExpansion, bool OwnsTokens); 463 464 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 465 /// lexer stack. This should only be used in situations where the current 466 /// state of the top-of-stack lexer is known. 467 void RemoveTopOfLexerStack(); 468 469 /// EnableBacktrackAtThisPos - From the point that this method is called, and 470 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 471 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 472 /// make the Preprocessor re-lex the same tokens. 473 /// 474 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 475 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 476 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 477 /// 478 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 479 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 480 /// tokens will continue indefinitely. 481 /// 482 void EnableBacktrackAtThisPos(); 483 484 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 485 void CommitBacktrackedTokens(); 486 487 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 488 /// EnableBacktrackAtThisPos() was previously called. 489 void Backtrack(); 490 491 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 492 /// caching of tokens is on. 493 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 494 495 /// Lex - To lex a token from the preprocessor, just pull a token from the 496 /// current lexer or macro object. 497 void Lex(Token &Result) { 498 if (CurLexer) 499 CurLexer->Lex(Result); 500 else if (CurPTHLexer) 501 CurPTHLexer->Lex(Result); 502 else if (CurTokenLexer) 503 CurTokenLexer->Lex(Result); 504 else 505 CachingLex(Result); 506 } 507 508 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 509 /// something not a comment. This is useful in -E -C mode where comments 510 /// would foul up preprocessor directive handling. 511 void LexNonComment(Token &Result) { 512 do 513 Lex(Result); 514 while (Result.getKind() == tok::comment); 515 } 516 517 /// LexUnexpandedToken - This is just like Lex, but this disables macro 518 /// expansion of identifier tokens. 519 void LexUnexpandedToken(Token &Result) { 520 // Disable macro expansion. 521 bool OldVal = DisableMacroExpansion; 522 DisableMacroExpansion = true; 523 // Lex the token. 524 Lex(Result); 525 526 // Reenable it. 527 DisableMacroExpansion = OldVal; 528 } 529 530 /// LookAhead - This peeks ahead N tokens and returns that token without 531 /// consuming any tokens. LookAhead(0) returns the next token that would be 532 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 533 /// returns normal tokens after phase 5. As such, it is equivalent to using 534 /// 'Lex', not 'LexUnexpandedToken'. 535 const Token &LookAhead(unsigned N) { 536 if (CachedLexPos + N < CachedTokens.size()) 537 return CachedTokens[CachedLexPos+N]; 538 else 539 return PeekAhead(N+1); 540 } 541 542 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 543 /// this allows to revert a specific number of tokens. 544 /// Note that the number of tokens being reverted should be up to the last 545 /// backtrack position, not more. 546 void RevertCachedTokens(unsigned N) { 547 assert(isBacktrackEnabled() && 548 "Should only be called when tokens are cached for backtracking"); 549 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 550 && "Should revert tokens up to the last backtrack position, not more"); 551 assert(signed(CachedLexPos) - signed(N) >= 0 && 552 "Corrupted backtrack positions ?"); 553 CachedLexPos -= N; 554 } 555 556 /// EnterToken - Enters a token in the token stream to be lexed next. If 557 /// BackTrack() is called afterwards, the token will remain at the insertion 558 /// point. 559 void EnterToken(const Token &Tok) { 560 EnterCachingLexMode(); 561 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 562 } 563 564 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 565 /// tokens (because backtrack is enabled) it should replace the most recent 566 /// cached tokens with the given annotation token. This function has no effect 567 /// if backtracking is not enabled. 568 /// 569 /// Note that the use of this function is just for optimization; so that the 570 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 571 /// invoked. 572 void AnnotateCachedTokens(const Token &Tok) { 573 assert(Tok.isAnnotation() && "Expected annotation token"); 574 if (CachedLexPos != 0 && isBacktrackEnabled()) 575 AnnotatePreviousCachedTokens(Tok); 576 } 577 578 /// \brief Replace the last token with an annotation token. 579 /// 580 /// Like AnnotateCachedTokens(), this routine replaces an 581 /// already-parsed (and resolved) token with an annotation 582 /// token. However, this routine only replaces the last token with 583 /// the annotation token; it does not affect any other cached 584 /// tokens. This function has no effect if backtracking is not 585 /// enabled. 586 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 587 assert(Tok.isAnnotation() && "Expected annotation token"); 588 if (CachedLexPos != 0 && isBacktrackEnabled()) 589 CachedTokens[CachedLexPos-1] = Tok; 590 } 591 592 /// \brief Specify the point at which code-completion will be performed. 593 /// 594 /// \param File the file in which code completion should occur. If 595 /// this file is included multiple times, code-completion will 596 /// perform completion the first time it is included. If NULL, this 597 /// function clears out the code-completion point. 598 /// 599 /// \param Line the line at which code completion should occur 600 /// (1-based). 601 /// 602 /// \param Column the column at which code completion should occur 603 /// (1-based). 604 /// 605 /// \returns true if an error occurred, false otherwise. 606 bool SetCodeCompletionPoint(const FileEntry *File, 607 unsigned Line, unsigned Column); 608 609 /// \brief Determine if this source location refers into the file 610 /// for which we are performing code completion. 611 bool isCodeCompletionFile(SourceLocation FileLoc) const; 612 613 /// \brief Instruct the preprocessor to skip part of the main 614 /// the main source file. 615 /// 616 /// \brief Bytes The number of bytes in the preamble to skip. 617 /// 618 /// \brief StartOfLine Whether skipping these bytes puts the lexer at the 619 /// start of a line. 620 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 621 SkipMainFilePreamble.first = Bytes; 622 SkipMainFilePreamble.second = StartOfLine; 623 } 624 625 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 626 /// the specified Token's location, translating the token's start 627 /// position in the current buffer into a SourcePosition object for rendering. 628 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 629 return Diags->Report(Loc, DiagID); 630 } 631 632 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 633 return Diags->Report(Tok.getLocation(), DiagID); 634 } 635 636 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 637 /// token is the characters used to represent the token in the source file 638 /// after trigraph expansion and escaped-newline folding. In particular, this 639 /// wants to get the true, uncanonicalized, spelling of things like digraphs 640 /// UCNs, etc. 641 /// 642 /// \param Invalid If non-NULL, will be set \c true if an error occurs. 643 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const { 644 return Lexer::getSpelling(Tok, SourceMgr, Features, Invalid); 645 } 646 647 /// getSpelling - This method is used to get the spelling of a token into a 648 /// preallocated buffer, instead of as an std::string. The caller is required 649 /// to allocate enough space for the token, which is guaranteed to be at least 650 /// Tok.getLength() bytes long. The length of the actual result is returned. 651 /// 652 /// Note that this method may do two possible things: it may either fill in 653 /// the buffer specified with characters, or it may *change the input pointer* 654 /// to point to a constant buffer with the data already in it (avoiding a 655 /// copy). The caller is not allowed to modify the returned buffer pointer 656 /// if an internal buffer is returned. 657 unsigned getSpelling(const Token &Tok, const char *&Buffer, 658 bool *Invalid = 0) const { 659 return Lexer::getSpelling(Tok, Buffer, SourceMgr, Features, Invalid); 660 } 661 662 /// getSpelling - This method is used to get the spelling of a token into a 663 /// SmallVector. Note that the returned StringRef may not point to the 664 /// supplied buffer if a copy can be avoided. 665 llvm::StringRef getSpelling(const Token &Tok, 666 llvm::SmallVectorImpl<char> &Buffer, 667 bool *Invalid = 0) const; 668 669 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 670 /// with length 1, return the character. 671 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 672 bool *Invalid = 0) const { 673 assert(Tok.is(tok::numeric_constant) && 674 Tok.getLength() == 1 && "Called on unsupported token"); 675 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 676 677 // If the token is carrying a literal data pointer, just use it. 678 if (const char *D = Tok.getLiteralData()) 679 return *D; 680 681 // Otherwise, fall back on getCharacterData, which is slower, but always 682 // works. 683 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 684 } 685 686 /// CreateString - Plop the specified string into a scratch buffer and set the 687 /// specified token's location and length to it. If specified, the source 688 /// location provides a location of the instantiation point of the token. 689 void CreateString(const char *Buf, unsigned Len, 690 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 691 692 /// \brief Computes the source location just past the end of the 693 /// token at this source location. 694 /// 695 /// This routine can be used to produce a source location that 696 /// points just past the end of the token referenced by \p Loc, and 697 /// is generally used when a diagnostic needs to point just after a 698 /// token where it expected something different that it received. If 699 /// the returned source location would not be meaningful (e.g., if 700 /// it points into a macro), this routine returns an invalid 701 /// source location. 702 /// 703 /// \param Offset an offset from the end of the token, where the source 704 /// location should refer to. The default offset (0) produces a source 705 /// location pointing just past the end of the token; an offset of 1 produces 706 /// a source location pointing to the last character in the token, etc. 707 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0) { 708 return Lexer::getLocForEndOfToken(Loc, Offset, SourceMgr, Features); 709 } 710 711 /// DumpToken - Print the token to stderr, used for debugging. 712 /// 713 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 714 void DumpLocation(SourceLocation Loc) const; 715 void DumpMacro(const MacroInfo &MI) const; 716 717 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 718 /// token, return a new location that specifies a character within the token. 719 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart, 720 unsigned Char) const { 721 return Lexer::AdvanceToTokenCharacter(TokStart, Char, SourceMgr, Features); 722 } 723 724 /// IncrementPasteCounter - Increment the counters for the number of token 725 /// paste operations performed. If fast was specified, this is a 'fast paste' 726 /// case we handled. 727 /// 728 void IncrementPasteCounter(bool isFast) { 729 if (isFast) 730 ++NumFastTokenPaste; 731 else 732 ++NumTokenPaste; 733 } 734 735 void PrintStats(); 736 737 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 738 /// comment (/##/) in microsoft mode, this method handles updating the current 739 /// state, returning the token on the next source line. 740 void HandleMicrosoftCommentPaste(Token &Tok); 741 742 //===--------------------------------------------------------------------===// 743 // Preprocessor callback methods. These are invoked by a lexer as various 744 // directives and events are found. 745 746 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 747 /// identifier information for the token and install it into the token. 748 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 749 const char *BufPtr = 0) const; 750 751 /// HandleIdentifier - This callback is invoked when the lexer reads an 752 /// identifier and has filled in the tokens IdentifierInfo member. This 753 /// callback potentially macro expands it or turns it into a named token (like 754 /// 'for'). 755 void HandleIdentifier(Token &Identifier); 756 757 758 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 759 /// the current file. This either returns the EOF token and returns true, or 760 /// pops a level off the include stack and returns false, at which point the 761 /// client should call lex again. 762 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 763 764 /// HandleEndOfTokenLexer - This callback is invoked when the current 765 /// TokenLexer hits the end of its token stream. 766 bool HandleEndOfTokenLexer(Token &Result); 767 768 /// HandleDirective - This callback is invoked when the lexer sees a # token 769 /// at the start of a line. This consumes the directive, modifies the 770 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 771 /// read is the correct one. 772 void HandleDirective(Token &Result); 773 774 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 775 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 776 /// true, then we consider macros that expand to zero tokens as being ok. 777 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 778 779 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 780 /// current line until the tok::eom token is found. 781 void DiscardUntilEndOfDirective(); 782 783 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 784 /// __DATE__ or __TIME__ in the file so far. 785 bool SawDateOrTime() const { 786 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 787 } 788 unsigned getCounterValue() const { return CounterValue; } 789 void setCounterValue(unsigned V) { CounterValue = V; } 790 791 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 792 /// SourceLocation. 793 MacroInfo *AllocateMacroInfo(SourceLocation L); 794 795 /// CloneMacroInfo - Allocate a new MacroInfo object which is clone of MI. 796 MacroInfo *CloneMacroInfo(const MacroInfo &MI); 797 798 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 799 /// checked and spelled filename, e.g. as an operand of #include. This returns 800 /// true if the input filename was in <>'s or false if it were in ""'s. The 801 /// caller is expected to provide a buffer that is large enough to hold the 802 /// spelling of the filename, but is also expected to handle the case when 803 /// this method decides to use a different buffer. 804 bool GetIncludeFilenameSpelling(SourceLocation Loc,llvm::StringRef &Filename); 805 806 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 807 /// return null on failure. isAngled indicates whether the file reference is 808 /// for system #include's or not (i.e. using <> instead of ""). 809 const FileEntry *LookupFile(llvm::StringRef Filename, 810 bool isAngled, const DirectoryLookup *FromDir, 811 const DirectoryLookup *&CurDir); 812 813 /// GetCurLookup - The DirectoryLookup structure used to find the current 814 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 815 /// implement #include_next and find directory-specific properties. 816 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 817 818 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 819 /// #include. 820 bool isInPrimaryFile() const; 821 822 /// ConcatenateIncludeName - Handle cases where the #include name is expanded 823 /// from a macro as multiple tokens, which need to be glued together. This 824 /// occurs for code like: 825 /// #define FOO <a/b.h> 826 /// #include FOO 827 /// because in this case, "<a/b.h>" is returned as 7 tokens, not one. 828 /// 829 /// This code concatenates and consumes tokens up to the '>' token. It 830 /// returns false if the > was found, otherwise it returns true if it finds 831 /// and consumes the EOM marker. 832 bool ConcatenateIncludeName(llvm::SmallString<128> &FilenameBuffer, 833 SourceLocation &End); 834 835private: 836 837 void PushIncludeMacroStack() { 838 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 839 CurPTHLexer.take(), 840 CurPPLexer, 841 CurTokenLexer.take(), 842 CurDirLookup)); 843 CurPPLexer = 0; 844 } 845 846 void PopIncludeMacroStack() { 847 CurLexer.reset(IncludeMacroStack.back().TheLexer); 848 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 849 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 850 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 851 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 852 IncludeMacroStack.pop_back(); 853 } 854 855 /// AllocateMacroInfo - Allocate a new MacroInfo object. 856 MacroInfo *AllocateMacroInfo(); 857 858 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 859 /// be reused for allocating new MacroInfo objects. 860 void ReleaseMacroInfo(MacroInfo* MI); 861 862 /// ReadMacroName - Lex and validate a macro name, which occurs after a 863 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 864 /// and discards the rest of the macro line if the macro name is invalid. 865 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 866 867 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 868 /// definition has just been read. Lex the rest of the arguments and the 869 /// closing ), updating MI with what we learn. Return true if an error occurs 870 /// parsing the arg list. 871 bool ReadMacroDefinitionArgList(MacroInfo *MI); 872 873 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 874 /// decided that the subsequent tokens are in the #if'd out portion of the 875 /// file. Lex the rest of the file, until we see an #endif. If 876 /// FoundNonSkipPortion is true, then we have already emitted code for part of 877 /// this #if directive, so #else/#elif blocks should never be entered. If 878 /// FoundElse is false, then #else directives are ok, if not, then we have 879 /// already seen one so a #else directive is a duplicate. When this returns, 880 /// the caller can lex the first valid token. 881 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 882 bool FoundNonSkipPortion, bool FoundElse); 883 884 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 885 /// SkipExcludedConditionalBlock. 886 void PTHSkipExcludedConditionalBlock(); 887 888 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 889 /// may occur after a #if or #elif directive and return it as a bool. If the 890 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 891 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 892 893 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 894 /// #pragma GCC poison/system_header/dependency and #pragma once. 895 void RegisterBuiltinPragmas(); 896 897 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 898 /// identifier table. 899 void RegisterBuiltinMacros(); 900 901 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 902 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 903 /// the macro should not be expanded return true, otherwise return false. 904 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 905 906 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 907 /// lexed is a '('. If so, consume the token and return true, if not, this 908 /// method should have no observable side-effect on the lexed tokens. 909 bool isNextPPTokenLParen(); 910 911 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 912 /// invoked to read all of the formal arguments specified for the macro 913 /// invocation. This returns null on error. 914 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 915 SourceLocation &InstantiationEnd); 916 917 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 918 /// as a builtin macro, handle it and return the next token as 'Tok'. 919 void ExpandBuiltinMacro(Token &Tok); 920 921 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 922 /// return the first token after the directive. The _Pragma token has just 923 /// been read into 'Tok'. 924 void Handle_Pragma(Token &Tok); 925 926 /// HandleMicrosoft__pragma - Like Handle_Pragma except the pragma text 927 /// is not enclosed within a string literal. 928 void HandleMicrosoft__pragma(Token &Tok); 929 930 void Handle_Pragma(unsigned Introducer, const std::string &StrVal, 931 SourceLocation PragmaLoc, SourceLocation RParenLoc); 932 933 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 934 /// start lexing tokens from it instead of the current buffer. 935 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 936 937 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 938 /// start getting tokens from it using the PTH cache. 939 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 940 941 /// IsFileLexer - Returns true if we are lexing from a file and not a 942 /// pragma or a macro. 943 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 944 return L ? !L->isPragmaLexer() : P != 0; 945 } 946 947 static bool IsFileLexer(const IncludeStackInfo& I) { 948 return IsFileLexer(I.TheLexer, I.ThePPLexer); 949 } 950 951 bool IsFileLexer() const { 952 return IsFileLexer(CurLexer.get(), CurPPLexer); 953 } 954 955 //===--------------------------------------------------------------------===// 956 // Caching stuff. 957 void CachingLex(Token &Result); 958 bool InCachingLexMode() const { 959 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 960 // that we are past EOF, not that we are in CachingLex mode. 961 return CurPPLexer == 0 && CurTokenLexer == 0 && CurPTHLexer == 0 && 962 !IncludeMacroStack.empty(); 963 } 964 void EnterCachingLexMode(); 965 void ExitCachingLexMode() { 966 if (InCachingLexMode()) 967 RemoveTopOfLexerStack(); 968 } 969 const Token &PeekAhead(unsigned N); 970 void AnnotatePreviousCachedTokens(const Token &Tok); 971 972 //===--------------------------------------------------------------------===// 973 /// Handle*Directive - implement the various preprocessor directives. These 974 /// should side-effect the current preprocessor object so that the next call 975 /// to Lex() will return the appropriate token next. 976 void HandleLineDirective(Token &Tok); 977 void HandleDigitDirective(Token &Tok); 978 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 979 void HandleIdentSCCSDirective(Token &Tok); 980 981 // File inclusion. 982 void HandleIncludeDirective(SourceLocation HashLoc, 983 Token &Tok, 984 const DirectoryLookup *LookupFrom = 0, 985 bool isImport = false); 986 void HandleIncludeNextDirective(SourceLocation HashLoc, Token &Tok); 987 void HandleIncludeMacrosDirective(SourceLocation HashLoc, Token &Tok); 988 void HandleImportDirective(SourceLocation HashLoc, Token &Tok); 989 990 // Macro handling. 991 void HandleDefineDirective(Token &Tok); 992 void HandleUndefDirective(Token &Tok); 993 994 // Conditional Inclusion. 995 void HandleIfdefDirective(Token &Tok, bool isIfndef, 996 bool ReadAnyTokensBeforeDirective); 997 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 998 void HandleEndifDirective(Token &Tok); 999 void HandleElseDirective(Token &Tok); 1000 void HandleElifDirective(Token &Tok); 1001 1002 // Pragmas. 1003 void HandlePragmaDirective(unsigned Introducer); 1004public: 1005 void HandlePragmaOnce(Token &OnceTok); 1006 void HandlePragmaMark(); 1007 void HandlePragmaPoison(Token &PoisonTok); 1008 void HandlePragmaSystemHeader(Token &SysHeaderTok); 1009 void HandlePragmaDependency(Token &DependencyTok); 1010 void HandlePragmaComment(Token &CommentTok); 1011 void HandlePragmaMessage(Token &MessageTok); 1012 void HandlePragmaPushMacro(Token &Tok); 1013 void HandlePragmaPopMacro(Token &Tok); 1014 IdentifierInfo *ParsePragmaPushOrPopMacro(Token &Tok); 1015 1016 // Return true and store the first token only if any CommentHandler 1017 // has inserted some tokens and getCommentRetentionState() is false. 1018 bool HandleComment(Token &Token, SourceRange Comment); 1019}; 1020 1021/// \brief Abstract base class that describes a handler that will receive 1022/// source ranges for each of the comments encountered in the source file. 1023class CommentHandler { 1024public: 1025 virtual ~CommentHandler(); 1026 1027 // The handler shall return true if it has pushed any tokens 1028 // to be read using e.g. EnterToken or EnterTokenStream. 1029 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 1030}; 1031 1032} // end namespace clang 1033 1034#endif 1035