Preprocessor.h revision 9b36c3f0de0105e903130bbda3c4aea7d792c0af
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 ExternalPreprocessorSource; 36class FileManager; 37class FileEntry; 38class HeaderSearch; 39class PragmaNamespace; 40class PragmaHandler; 41class CommentHandler; 42class ScratchBuffer; 43class TargetInfo; 44class PPCallbacks; 45class DirectoryLookup; 46class PreprocessingRecord; 47 48/// Preprocessor - This object engages in a tight little dance with the lexer to 49/// efficiently preprocess tokens. Lexers know only about tokens within a 50/// single source file, and don't know anything about preprocessor-level issues 51/// like the #include stack, token expansion, etc. 52/// 53class Preprocessor { 54 Diagnostic *Diags; 55 LangOptions Features; 56 const TargetInfo &Target; 57 FileManager &FileMgr; 58 SourceManager &SourceMgr; 59 ScratchBuffer *ScratchBuf; 60 HeaderSearch &HeaderInfo; 61 62 /// \brief External source of macros. 63 ExternalPreprocessorSource *ExternalSource; 64 65 /// PTH - An optional PTHManager object used for getting tokens from 66 /// a token cache rather than lexing the original source file. 67 llvm::OwningPtr<PTHManager> PTH; 68 69 /// BP - A BumpPtrAllocator object used to quickly allocate and release 70 /// objects internal to the Preprocessor. 71 llvm::BumpPtrAllocator BP; 72 73 /// Identifiers for builtin macros and other builtins. 74 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 75 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 76 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 77 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 78 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 79 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 80 IdentifierInfo *Ident_Pragma, *Ident__VA_ARGS__; // _Pragma, __VA_ARGS__ 81 IdentifierInfo *Ident__has_feature; // __has_feature 82 IdentifierInfo *Ident__has_builtin; // __has_builtin 83 IdentifierInfo *Ident__has_include; // __has_include 84 IdentifierInfo *Ident__has_include_next; // __has_include_next 85 86 SourceLocation DATELoc, TIMELoc; 87 unsigned CounterValue; // Next __COUNTER__ value. 88 89 enum { 90 /// MaxIncludeStackDepth - Maximum depth of #includes. 91 MaxAllowedIncludeStackDepth = 200 92 }; 93 94 // State that is set before the preprocessor begins. 95 bool KeepComments : 1; 96 bool KeepMacroComments : 1; 97 98 // State that changes while the preprocessor runs: 99 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 100 101 /// Whether the preprocessor owns the header search object. 102 bool OwnsHeaderSearch : 1; 103 104 /// DisableMacroExpansion - True if macro expansion is disabled. 105 bool DisableMacroExpansion : 1; 106 107 /// \brief Whether we have already loaded macros from the external source. 108 mutable bool ReadMacrosFromExternalSource : 1; 109 110 /// Identifiers - This is mapping/lookup information for all identifiers in 111 /// the program, including program keywords. 112 mutable IdentifierTable Identifiers; 113 114 /// Selectors - This table contains all the selectors in the program. Unlike 115 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 116 /// It is declared/instantiated here because it's role/lifetime is 117 /// conceptually similar the IdentifierTable. In addition, the current control 118 /// flow (in clang::ParseAST()), make it convenient to put here. 119 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 120 /// the lifetime fo the preprocessor. 121 SelectorTable Selectors; 122 123 /// BuiltinInfo - Information about builtins. 124 Builtin::Context BuiltinInfo; 125 126 /// PragmaHandlers - This tracks all of the pragmas that the client registered 127 /// with this preprocessor. 128 PragmaNamespace *PragmaHandlers; 129 130 /// \brief Tracks all of the comment handlers that the client registered 131 /// with this preprocessor. 132 std::vector<CommentHandler *> CommentHandlers; 133 134 /// \brief The file that we're performing code-completion for, if any. 135 const FileEntry *CodeCompletionFile; 136 137 /// CurLexer - This is the current top of the stack that we're lexing from if 138 /// not expanding a macro and we are lexing directly from source code. 139 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 140 llvm::OwningPtr<Lexer> CurLexer; 141 142 /// CurPTHLexer - This is the current top of stack that we're lexing from if 143 /// not expanding from a macro and we are lexing from a PTH cache. 144 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 145 llvm::OwningPtr<PTHLexer> CurPTHLexer; 146 147 /// CurPPLexer - This is the current top of the stack what we're lexing from 148 /// if not expanding a macro. This is an alias for either CurLexer or 149 /// CurPTHLexer. 150 PreprocessorLexer *CurPPLexer; 151 152 /// CurLookup - The DirectoryLookup structure used to find the current 153 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 154 /// implement #include_next and find directory-specific properties. 155 const DirectoryLookup *CurDirLookup; 156 157 /// CurTokenLexer - This is the current macro we are expanding, if we are 158 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 159 llvm::OwningPtr<TokenLexer> CurTokenLexer; 160 161 /// IncludeMacroStack - This keeps track of the stack of files currently 162 /// #included, and macros currently being expanded from, not counting 163 /// CurLexer/CurTokenLexer. 164 struct IncludeStackInfo { 165 Lexer *TheLexer; 166 PTHLexer *ThePTHLexer; 167 PreprocessorLexer *ThePPLexer; 168 TokenLexer *TheTokenLexer; 169 const DirectoryLookup *TheDirLookup; 170 171 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 172 TokenLexer* TL, const DirectoryLookup *D) 173 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 174 TheDirLookup(D) {} 175 }; 176 std::vector<IncludeStackInfo> IncludeMacroStack; 177 178 /// Callbacks - These are actions invoked when some preprocessor activity is 179 /// encountered (e.g. a file is #included, etc). 180 PPCallbacks *Callbacks; 181 182 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 183 /// to the actual definition of the macro. 184 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 185 186 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 187 /// allocation. 188 /// FIXME: why not use a singly linked list? 189 std::vector<MacroInfo*> MICache; 190 191 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 192 /// reused for quick allocation. 193 MacroArgs *MacroArgCache; 194 friend class MacroArgs; 195 196 // Various statistics we track for performance analysis. 197 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 198 unsigned NumIf, NumElse, NumEndif; 199 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 200 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 201 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 202 unsigned NumSkipped; 203 204 /// Predefines - This string is the predefined macros that preprocessor 205 /// should use from the command line etc. 206 std::string Predefines; 207 208 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 209 enum { TokenLexerCacheSize = 8 }; 210 unsigned NumCachedTokenLexers; 211 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 212 213 /// \brief A record of the macro definitions and instantiations that 214 /// occurred during preprocessing. 215 /// 216 /// This is an optional side structure that can be enabled with 217 /// \c createPreprocessingRecord() prior to preprocessing. 218 PreprocessingRecord *Record; 219 220private: // Cached tokens state. 221 typedef llvm::SmallVector<Token, 1> CachedTokensTy; 222 223 /// CachedTokens - Cached tokens are stored here when we do backtracking or 224 /// lookahead. They are "lexed" by the CachingLex() method. 225 CachedTokensTy CachedTokens; 226 227 /// CachedLexPos - The position of the cached token that CachingLex() should 228 /// "lex" next. If it points beyond the CachedTokens vector, it means that 229 /// a normal Lex() should be invoked. 230 CachedTokensTy::size_type CachedLexPos; 231 232 /// BacktrackPositions - Stack of backtrack positions, allowing nested 233 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 234 /// indicate where CachedLexPos should be set when the BackTrack() method is 235 /// invoked (at which point the last position is popped). 236 std::vector<CachedTokensTy::size_type> BacktrackPositions; 237 238public: 239 Preprocessor(Diagnostic &diags, const LangOptions &opts, 240 const TargetInfo &target, 241 SourceManager &SM, HeaderSearch &Headers, 242 IdentifierInfoLookup *IILookup = 0, 243 bool OwnsHeaderSearch = false); 244 245 ~Preprocessor(); 246 247 Diagnostic &getDiagnostics() const { return *Diags; } 248 void setDiagnostics(Diagnostic &D) { Diags = &D; } 249 250 const LangOptions &getLangOptions() const { return Features; } 251 const TargetInfo &getTargetInfo() const { return Target; } 252 FileManager &getFileManager() const { return FileMgr; } 253 SourceManager &getSourceManager() const { return SourceMgr; } 254 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 255 256 IdentifierTable &getIdentifierTable() { return Identifiers; } 257 SelectorTable &getSelectorTable() { return Selectors; } 258 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 259 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 260 261 void setPTHManager(PTHManager* pm); 262 263 PTHManager *getPTHManager() { return PTH.get(); } 264 265 void setExternalSource(ExternalPreprocessorSource *Source) { 266 ExternalSource = Source; 267 } 268 269 ExternalPreprocessorSource *getExternalSource() const { 270 return ExternalSource; 271 } 272 273 /// SetCommentRetentionState - Control whether or not the preprocessor retains 274 /// comments in output. 275 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 276 this->KeepComments = KeepComments | KeepMacroComments; 277 this->KeepMacroComments = KeepMacroComments; 278 } 279 280 bool getCommentRetentionState() const { return KeepComments; } 281 282 /// isCurrentLexer - Return true if we are lexing directly from the specified 283 /// lexer. 284 bool isCurrentLexer(const PreprocessorLexer *L) const { 285 return CurPPLexer == L; 286 } 287 288 /// getCurrentLexer - Return the current lexer being lexed from. Note 289 /// that this ignores any potentially active macro expansions and _Pragma 290 /// expansions going on at the time. 291 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 292 293 /// getCurrentFileLexer - Return the current file lexer being lexed from. 294 /// Note that this ignores any potentially active macro expansions and _Pragma 295 /// expansions going on at the time. 296 PreprocessorLexer *getCurrentFileLexer() const; 297 298 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 299 /// Note that this class takes ownership of any PPCallbacks object given to 300 /// it. 301 PPCallbacks *getPPCallbacks() const { return Callbacks; } 302 void addPPCallbacks(PPCallbacks *C) { 303 if (Callbacks) 304 C = new PPChainedCallbacks(C, Callbacks); 305 Callbacks = C; 306 } 307 308 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 309 /// or null if it isn't #define'd. 310 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 311 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 312 } 313 314 /// setMacroInfo - Specify a macro for this identifier. 315 /// 316 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 317 318 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 319 /// state of the macro table. This visits every currently-defined macro. 320 typedef llvm::DenseMap<IdentifierInfo*, 321 MacroInfo*>::const_iterator macro_iterator; 322 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 323 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 324 325 const std::string &getPredefines() const { return Predefines; } 326 /// setPredefines - Set the predefines for this Preprocessor. These 327 /// predefines are automatically injected when parsing the main file. 328 void setPredefines(const char *P) { Predefines = P; } 329 void setPredefines(const std::string &P) { Predefines = P; } 330 331 /// getIdentifierInfo - Return information about the specified preprocessor 332 /// identifier token. The version of this method that takes two character 333 /// pointers is preferred unless the identifier is already available as a 334 /// string (this avoids allocation and copying of memory to construct an 335 /// std::string). 336 IdentifierInfo *getIdentifierInfo(llvm::StringRef Name) const { 337 return &Identifiers.get(Name); 338 } 339 340 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 341 /// If 'Namespace' is non-null, then it is a token required to exist on the 342 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 343 void AddPragmaHandler(llvm::StringRef Namespace, PragmaHandler *Handler); 344 void AddPragmaHandler(PragmaHandler *Handler) { 345 AddPragmaHandler(llvm::StringRef(), Handler); 346 } 347 348 /// RemovePragmaHandler - Remove the specific pragma handler from 349 /// the preprocessor. If \arg Namespace is non-null, then it should 350 /// be the namespace that \arg Handler was added to. It is an error 351 /// to remove a handler that has not been registered. 352 void RemovePragmaHandler(llvm::StringRef Namespace, PragmaHandler *Handler); 353 void RemovePragmaHandler(PragmaHandler *Handler) { 354 RemovePragmaHandler(llvm::StringRef(), Handler); 355 } 356 357 /// \brief Add the specified comment handler to the preprocessor. 358 void AddCommentHandler(CommentHandler *Handler); 359 360 /// \brief Remove the specified comment handler. 361 /// 362 /// It is an error to remove a handler that has not been registered. 363 void RemoveCommentHandler(CommentHandler *Handler); 364 365 /// \brief Retrieve the preprocessing record, or NULL if there is no 366 /// preprocessing record. 367 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 368 369 /// \brief Create a new preprocessing record, which will keep track of 370 /// all macro expansions, macro definitions, etc. 371 void createPreprocessingRecord(); 372 373 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 374 /// which implicitly adds the builtin defines etc. 375 void EnterMainSourceFile(); 376 377 /// EndSourceFile - Inform the preprocessor callbacks that processing is 378 /// complete. 379 void EndSourceFile(); 380 381 /// EnterSourceFile - Add a source file to the top of the include stack and 382 /// start lexing tokens from it instead of the current buffer. Emit an error 383 /// and don't enter the file on error. 384 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 385 SourceLocation Loc); 386 387 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 388 /// tokens from it instead of the current buffer. Args specifies the 389 /// tokens input to a function-like macro. 390 /// 391 /// ILEnd specifies the location of the ')' for a function-like macro or the 392 /// identifier for an object-like macro. 393 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 394 395 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 396 /// which will cause the lexer to start returning the specified tokens. 397 /// 398 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 399 /// not be subject to further macro expansion. Otherwise, these tokens will 400 /// be re-macro-expanded when/if expansion is enabled. 401 /// 402 /// If OwnsTokens is false, this method assumes that the specified stream of 403 /// tokens has a permanent owner somewhere, so they do not need to be copied. 404 /// If it is true, it assumes the array of tokens is allocated with new[] and 405 /// must be freed. 406 /// 407 void EnterTokenStream(const Token *Toks, unsigned NumToks, 408 bool DisableMacroExpansion, bool OwnsTokens); 409 410 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 411 /// lexer stack. This should only be used in situations where the current 412 /// state of the top-of-stack lexer is known. 413 void RemoveTopOfLexerStack(); 414 415 /// EnableBacktrackAtThisPos - From the point that this method is called, and 416 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 417 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 418 /// make the Preprocessor re-lex the same tokens. 419 /// 420 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 421 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 422 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 423 /// 424 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 425 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 426 /// tokens will continue indefinitely. 427 /// 428 void EnableBacktrackAtThisPos(); 429 430 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 431 void CommitBacktrackedTokens(); 432 433 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 434 /// EnableBacktrackAtThisPos() was previously called. 435 void Backtrack(); 436 437 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 438 /// caching of tokens is on. 439 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 440 441 /// Lex - To lex a token from the preprocessor, just pull a token from the 442 /// current lexer or macro object. 443 void Lex(Token &Result) { 444 if (CurLexer) 445 CurLexer->Lex(Result); 446 else if (CurPTHLexer) 447 CurPTHLexer->Lex(Result); 448 else if (CurTokenLexer) 449 CurTokenLexer->Lex(Result); 450 else 451 CachingLex(Result); 452 } 453 454 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 455 /// something not a comment. This is useful in -E -C mode where comments 456 /// would foul up preprocessor directive handling. 457 void LexNonComment(Token &Result) { 458 do 459 Lex(Result); 460 while (Result.getKind() == tok::comment); 461 } 462 463 /// LexUnexpandedToken - This is just like Lex, but this disables macro 464 /// expansion of identifier tokens. 465 void LexUnexpandedToken(Token &Result) { 466 // Disable macro expansion. 467 bool OldVal = DisableMacroExpansion; 468 DisableMacroExpansion = true; 469 // Lex the token. 470 Lex(Result); 471 472 // Reenable it. 473 DisableMacroExpansion = OldVal; 474 } 475 476 /// LookAhead - This peeks ahead N tokens and returns that token without 477 /// consuming any tokens. LookAhead(0) returns the next token that would be 478 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 479 /// returns normal tokens after phase 5. As such, it is equivalent to using 480 /// 'Lex', not 'LexUnexpandedToken'. 481 const Token &LookAhead(unsigned N) { 482 if (CachedLexPos + N < CachedTokens.size()) 483 return CachedTokens[CachedLexPos+N]; 484 else 485 return PeekAhead(N+1); 486 } 487 488 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 489 /// this allows to revert a specific number of tokens. 490 /// Note that the number of tokens being reverted should be up to the last 491 /// backtrack position, not more. 492 void RevertCachedTokens(unsigned N) { 493 assert(isBacktrackEnabled() && 494 "Should only be called when tokens are cached for backtracking"); 495 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 496 && "Should revert tokens up to the last backtrack position, not more"); 497 assert(signed(CachedLexPos) - signed(N) >= 0 && 498 "Corrupted backtrack positions ?"); 499 CachedLexPos -= N; 500 } 501 502 /// EnterToken - Enters a token in the token stream to be lexed next. If 503 /// BackTrack() is called afterwards, the token will remain at the insertion 504 /// point. 505 void EnterToken(const Token &Tok) { 506 EnterCachingLexMode(); 507 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 508 } 509 510 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 511 /// tokens (because backtrack is enabled) it should replace the most recent 512 /// cached tokens with the given annotation token. This function has no effect 513 /// if backtracking is not enabled. 514 /// 515 /// Note that the use of this function is just for optimization; so that the 516 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 517 /// invoked. 518 void AnnotateCachedTokens(const Token &Tok) { 519 assert(Tok.isAnnotation() && "Expected annotation token"); 520 if (CachedLexPos != 0 && isBacktrackEnabled()) 521 AnnotatePreviousCachedTokens(Tok); 522 } 523 524 /// \brief Replace the last token with an annotation token. 525 /// 526 /// Like AnnotateCachedTokens(), this routine replaces an 527 /// already-parsed (and resolved) token with an annotation 528 /// token. However, this routine only replaces the last token with 529 /// the annotation token; it does not affect any other cached 530 /// tokens. This function has no effect if backtracking is not 531 /// enabled. 532 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 533 assert(Tok.isAnnotation() && "Expected annotation token"); 534 if (CachedLexPos != 0 && isBacktrackEnabled()) 535 CachedTokens[CachedLexPos-1] = Tok; 536 } 537 538 /// \brief Specify the point at which code-completion will be performed. 539 /// 540 /// \param File the file in which code completion should occur. If 541 /// this file is included multiple times, code-completion will 542 /// perform completion the first time it is included. If NULL, this 543 /// function clears out the code-completion point. 544 /// 545 /// \param Line the line at which code completion should occur 546 /// (1-based). 547 /// 548 /// \param Column the column at which code completion should occur 549 /// (1-based). 550 /// 551 /// \returns true if an error occurred, false otherwise. 552 bool SetCodeCompletionPoint(const FileEntry *File, 553 unsigned Line, unsigned Column); 554 555 /// \brief Determine if this source location refers into the file 556 /// for which we are performing code completion. 557 bool isCodeCompletionFile(SourceLocation FileLoc) const; 558 559 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 560 /// the specified Token's location, translating the token's start 561 /// position in the current buffer into a SourcePosition object for rendering. 562 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 563 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 564 } 565 566 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 567 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 568 DiagID); 569 } 570 571 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 572 /// token is the characters used to represent the token in the source file 573 /// after trigraph expansion and escaped-newline folding. In particular, this 574 /// wants to get the true, uncanonicalized, spelling of things like digraphs 575 /// UCNs, etc. 576 /// 577 /// \param Invalid If non-NULL, will be set \c true if an error occurs. 578 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const; 579 580 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 581 /// token is the characters used to represent the token in the source file 582 /// after trigraph expansion and escaped-newline folding. In particular, this 583 /// wants to get the true, uncanonicalized, spelling of things like digraphs 584 /// UCNs, etc. 585 static std::string getSpelling(const Token &Tok, 586 const SourceManager &SourceMgr, 587 const LangOptions &Features, 588 bool *Invalid = 0); 589 590 /// getSpelling - This method is used to get the spelling of a token into a 591 /// preallocated buffer, instead of as an std::string. The caller is required 592 /// to allocate enough space for the token, which is guaranteed to be at least 593 /// Tok.getLength() bytes long. The length of the actual result is returned. 594 /// 595 /// Note that this method may do two possible things: it may either fill in 596 /// the buffer specified with characters, or it may *change the input pointer* 597 /// to point to a constant buffer with the data already in it (avoiding a 598 /// copy). The caller is not allowed to modify the returned buffer pointer 599 /// if an internal buffer is returned. 600 unsigned getSpelling(const Token &Tok, const char *&Buffer, 601 bool *Invalid = 0) const; 602 603 /// getSpelling - This method is used to get the spelling of a token into a 604 /// SmallVector. Note that the returned StringRef may not point to the 605 /// supplied buffer if a copy can be avoided. 606 llvm::StringRef getSpelling(const Token &Tok, 607 llvm::SmallVectorImpl<char> &Buffer, 608 bool *Invalid = 0) const; 609 610 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 611 /// with length 1, return the character. 612 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 613 bool *Invalid = 0) const { 614 assert(Tok.is(tok::numeric_constant) && 615 Tok.getLength() == 1 && "Called on unsupported token"); 616 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 617 618 // If the token is carrying a literal data pointer, just use it. 619 if (const char *D = Tok.getLiteralData()) 620 return *D; 621 622 // Otherwise, fall back on getCharacterData, which is slower, but always 623 // works. 624 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 625 } 626 627 /// CreateString - Plop the specified string into a scratch buffer and set the 628 /// specified token's location and length to it. If specified, the source 629 /// location provides a location of the instantiation point of the token. 630 void CreateString(const char *Buf, unsigned Len, 631 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 632 633 /// \brief Computes the source location just past the end of the 634 /// token at this source location. 635 /// 636 /// This routine can be used to produce a source location that 637 /// points just past the end of the token referenced by \p Loc, and 638 /// is generally used when a diagnostic needs to point just after a 639 /// token where it expected something different that it received. If 640 /// the returned source location would not be meaningful (e.g., if 641 /// it points into a macro), this routine returns an invalid 642 /// source location. 643 /// 644 /// \param Offset an offset from the end of the token, where the source 645 /// location should refer to. The default offset (0) produces a source 646 /// location pointing just past the end of the token; an offset of 1 produces 647 /// a source location pointing to the last character in the token, etc. 648 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0); 649 650 /// DumpToken - Print the token to stderr, used for debugging. 651 /// 652 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 653 void DumpLocation(SourceLocation Loc) const; 654 void DumpMacro(const MacroInfo &MI) const; 655 656 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 657 /// token, return a new location that specifies a character within the token. 658 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 659 660 /// IncrementPasteCounter - Increment the counters for the number of token 661 /// paste operations performed. If fast was specified, this is a 'fast paste' 662 /// case we handled. 663 /// 664 void IncrementPasteCounter(bool isFast) { 665 if (isFast) 666 ++NumFastTokenPaste; 667 else 668 ++NumTokenPaste; 669 } 670 671 void PrintStats(); 672 673 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 674 /// comment (/##/) in microsoft mode, this method handles updating the current 675 /// state, returning the token on the next source line. 676 void HandleMicrosoftCommentPaste(Token &Tok); 677 678 //===--------------------------------------------------------------------===// 679 // Preprocessor callback methods. These are invoked by a lexer as various 680 // directives and events are found. 681 682 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 683 /// identifier information for the token and install it into the token. 684 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 685 const char *BufPtr = 0) const; 686 687 /// HandleIdentifier - This callback is invoked when the lexer reads an 688 /// identifier and has filled in the tokens IdentifierInfo member. This 689 /// callback potentially macro expands it or turns it into a named token (like 690 /// 'for'). 691 void HandleIdentifier(Token &Identifier); 692 693 694 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 695 /// the current file. This either returns the EOF token and returns true, or 696 /// pops a level off the include stack and returns false, at which point the 697 /// client should call lex again. 698 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 699 700 /// HandleEndOfTokenLexer - This callback is invoked when the current 701 /// TokenLexer hits the end of its token stream. 702 bool HandleEndOfTokenLexer(Token &Result); 703 704 /// HandleDirective - This callback is invoked when the lexer sees a # token 705 /// at the start of a line. This consumes the directive, modifies the 706 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 707 /// read is the correct one. 708 void HandleDirective(Token &Result); 709 710 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 711 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 712 /// true, then we consider macros that expand to zero tokens as being ok. 713 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 714 715 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 716 /// current line until the tok::eom token is found. 717 void DiscardUntilEndOfDirective(); 718 719 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 720 /// __DATE__ or __TIME__ in the file so far. 721 bool SawDateOrTime() const { 722 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 723 } 724 unsigned getCounterValue() const { return CounterValue; } 725 void setCounterValue(unsigned V) { CounterValue = V; } 726 727 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 728 /// SourceLocation. 729 MacroInfo* AllocateMacroInfo(SourceLocation L); 730 731 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 732 /// checked and spelled filename, e.g. as an operand of #include. This returns 733 /// true if the input filename was in <>'s or false if it were in ""'s. The 734 /// caller is expected to provide a buffer that is large enough to hold the 735 /// spelling of the filename, but is also expected to handle the case when 736 /// this method decides to use a different buffer. 737 bool GetIncludeFilenameSpelling(SourceLocation Loc,llvm::StringRef &Filename); 738 739 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 740 /// return null on failure. isAngled indicates whether the file reference is 741 /// for system #include's or not (i.e. using <> instead of ""). 742 const FileEntry *LookupFile(llvm::StringRef Filename, 743 bool isAngled, const DirectoryLookup *FromDir, 744 const DirectoryLookup *&CurDir); 745 746 /// GetCurLookup - The DirectoryLookup structure used to find the current 747 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 748 /// implement #include_next and find directory-specific properties. 749 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 750 751 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 752 /// #include. 753 bool isInPrimaryFile() const; 754 755 /// ConcatenateIncludeName - Handle cases where the #include name is expanded 756 /// from a macro as multiple tokens, which need to be glued together. This 757 /// occurs for code like: 758 /// #define FOO <a/b.h> 759 /// #include FOO 760 /// because in this case, "<a/b.h>" is returned as 7 tokens, not one. 761 /// 762 /// This code concatenates and consumes tokens up to the '>' token. It 763 /// returns false if the > was found, otherwise it returns true if it finds 764 /// and consumes the EOM marker. 765 bool ConcatenateIncludeName(llvm::SmallString<128> &FilenameBuffer); 766 767private: 768 769 void PushIncludeMacroStack() { 770 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 771 CurPTHLexer.take(), 772 CurPPLexer, 773 CurTokenLexer.take(), 774 CurDirLookup)); 775 CurPPLexer = 0; 776 } 777 778 void PopIncludeMacroStack() { 779 CurLexer.reset(IncludeMacroStack.back().TheLexer); 780 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 781 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 782 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 783 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 784 IncludeMacroStack.pop_back(); 785 } 786 787 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 788 /// be reused for allocating new MacroInfo objects. 789 void ReleaseMacroInfo(MacroInfo* MI); 790 791 /// ReadMacroName - Lex and validate a macro name, which occurs after a 792 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 793 /// and discards the rest of the macro line if the macro name is invalid. 794 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 795 796 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 797 /// definition has just been read. Lex the rest of the arguments and the 798 /// closing ), updating MI with what we learn. Return true if an error occurs 799 /// parsing the arg list. 800 bool ReadMacroDefinitionArgList(MacroInfo *MI); 801 802 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 803 /// decided that the subsequent tokens are in the #if'd out portion of the 804 /// file. Lex the rest of the file, until we see an #endif. If 805 /// FoundNonSkipPortion is true, then we have already emitted code for part of 806 /// this #if directive, so #else/#elif blocks should never be entered. If 807 /// FoundElse is false, then #else directives are ok, if not, then we have 808 /// already seen one so a #else directive is a duplicate. When this returns, 809 /// the caller can lex the first valid token. 810 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 811 bool FoundNonSkipPortion, bool FoundElse); 812 813 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 814 /// SkipExcludedConditionalBlock. 815 void PTHSkipExcludedConditionalBlock(); 816 817 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 818 /// may occur after a #if or #elif directive and return it as a bool. If the 819 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 820 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 821 822 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 823 /// #pragma GCC poison/system_header/dependency and #pragma once. 824 void RegisterBuiltinPragmas(); 825 826 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 827 /// identifier table. 828 void RegisterBuiltinMacros(); 829 830 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 831 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 832 /// the macro should not be expanded return true, otherwise return false. 833 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 834 835 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 836 /// lexed is a '('. If so, consume the token and return true, if not, this 837 /// method should have no observable side-effect on the lexed tokens. 838 bool isNextPPTokenLParen(); 839 840 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 841 /// invoked to read all of the formal arguments specified for the macro 842 /// invocation. This returns null on error. 843 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 844 SourceLocation &InstantiationEnd); 845 846 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 847 /// as a builtin macro, handle it and return the next token as 'Tok'. 848 void ExpandBuiltinMacro(Token &Tok); 849 850 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 851 /// return the first token after the directive. The _Pragma token has just 852 /// been read into 'Tok'. 853 void Handle_Pragma(Token &Tok); 854 855 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 856 /// start lexing tokens from it instead of the current buffer. 857 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 858 859 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 860 /// start getting tokens from it using the PTH cache. 861 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 862 863 /// IsFileLexer - Returns true if we are lexing from a file and not a 864 /// pragma or a macro. 865 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 866 return L ? !L->isPragmaLexer() : P != 0; 867 } 868 869 static bool IsFileLexer(const IncludeStackInfo& I) { 870 return IsFileLexer(I.TheLexer, I.ThePPLexer); 871 } 872 873 bool IsFileLexer() const { 874 return IsFileLexer(CurLexer.get(), CurPPLexer); 875 } 876 877 //===--------------------------------------------------------------------===// 878 // Caching stuff. 879 void CachingLex(Token &Result); 880 bool InCachingLexMode() const { 881 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 882 // that we are past EOF, not that we are in CachingLex mode. 883 return CurPPLexer == 0 && CurTokenLexer == 0 && !IncludeMacroStack.empty(); 884 } 885 void EnterCachingLexMode(); 886 void ExitCachingLexMode() { 887 if (InCachingLexMode()) 888 RemoveTopOfLexerStack(); 889 } 890 const Token &PeekAhead(unsigned N); 891 void AnnotatePreviousCachedTokens(const Token &Tok); 892 893 //===--------------------------------------------------------------------===// 894 /// Handle*Directive - implement the various preprocessor directives. These 895 /// should side-effect the current preprocessor object so that the next call 896 /// to Lex() will return the appropriate token next. 897 void HandleLineDirective(Token &Tok); 898 void HandleDigitDirective(Token &Tok); 899 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 900 void HandleIdentSCCSDirective(Token &Tok); 901 902 // File inclusion. 903 void HandleIncludeDirective(Token &Tok, 904 const DirectoryLookup *LookupFrom = 0, 905 bool isImport = false); 906 void HandleIncludeNextDirective(Token &Tok); 907 void HandleIncludeMacrosDirective(Token &Tok); 908 void HandleImportDirective(Token &Tok); 909 910 // Macro handling. 911 void HandleDefineDirective(Token &Tok); 912 void HandleUndefDirective(Token &Tok); 913 914 // Conditional Inclusion. 915 void HandleIfdefDirective(Token &Tok, bool isIfndef, 916 bool ReadAnyTokensBeforeDirective); 917 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 918 void HandleEndifDirective(Token &Tok); 919 void HandleElseDirective(Token &Tok); 920 void HandleElifDirective(Token &Tok); 921 922 // Pragmas. 923 void HandlePragmaDirective(); 924public: 925 void HandlePragmaOnce(Token &OnceTok); 926 void HandlePragmaMark(); 927 void HandlePragmaPoison(Token &PoisonTok); 928 void HandlePragmaSystemHeader(Token &SysHeaderTok); 929 void HandlePragmaDependency(Token &DependencyTok); 930 void HandlePragmaComment(Token &CommentTok); 931 void HandlePragmaMessage(Token &MessageTok); 932 // Return true and store the first token only if any CommentHandler 933 // has inserted some tokens and getCommentRetentionState() is false. 934 bool HandleComment(Token &Token, SourceRange Comment); 935}; 936 937/// \brief Abstract base class that describes a handler that will receive 938/// source ranges for each of the comments encountered in the source file. 939class CommentHandler { 940public: 941 virtual ~CommentHandler(); 942 943 // The handler shall return true if it has pushed any tokens 944 // to be read using e.g. EnterToken or EnterTokenStream. 945 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 946}; 947 948} // end namespace clang 949 950#endif 951