Preprocessor.h revision 94dc8f640ebea52241412512ed48601626edbc58
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 llvm::OwningPtr<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. Note 294 /// 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(const char *Namespace, PragmaHandler *Handler); 344 345 /// RemovePragmaHandler - Remove the specific pragma handler from 346 /// the preprocessor. If \arg Namespace is non-null, then it should 347 /// be the namespace that \arg Handler was added to. It is an error 348 /// to remove a handler that has not been registered. 349 void RemovePragmaHandler(const char *Namespace, PragmaHandler *Handler); 350 351 /// \brief Add the specified comment handler to the preprocessor. 352 void AddCommentHandler(CommentHandler *Handler); 353 354 /// \brief Remove the specified comment handler. 355 /// 356 /// It is an error to remove a handler that has not been registered. 357 void RemoveCommentHandler(CommentHandler *Handler); 358 359 /// \brief Retrieve the preprocessing record, or NULL if there is no 360 /// preprocessing record. 361 PreprocessingRecord *getPreprocessingRecord() const { return Record.get(); } 362 363 /// \brief Create a new preprocessing record, which will keep track of 364 /// all macro expansions, macro definitions, etc. 365 void createPreprocessingRecord(); 366 367 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 368 /// which implicitly adds the builtin defines etc. 369 bool EnterMainSourceFile(); 370 371 /// EnterSourceFile - Add a source file to the top of the include stack and 372 /// start lexing tokens from it instead of the current buffer. Return true 373 /// and fill in ErrorStr with the error information on failure. 374 bool EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 375 std::string &ErrorStr); 376 377 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 378 /// tokens from it instead of the current buffer. Args specifies the 379 /// tokens input to a function-like macro. 380 /// 381 /// ILEnd specifies the location of the ')' for a function-like macro or the 382 /// identifier for an object-like macro. 383 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 384 385 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 386 /// which will cause the lexer to start returning the specified tokens. 387 /// 388 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 389 /// not be subject to further macro expansion. Otherwise, these tokens will 390 /// be re-macro-expanded when/if expansion is enabled. 391 /// 392 /// If OwnsTokens is false, this method assumes that the specified stream of 393 /// tokens has a permanent owner somewhere, so they do not need to be copied. 394 /// If it is true, it assumes the array of tokens is allocated with new[] and 395 /// must be freed. 396 /// 397 void EnterTokenStream(const Token *Toks, unsigned NumToks, 398 bool DisableMacroExpansion, bool OwnsTokens); 399 400 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 401 /// lexer stack. This should only be used in situations where the current 402 /// state of the top-of-stack lexer is known. 403 void RemoveTopOfLexerStack(); 404 405 /// EnableBacktrackAtThisPos - From the point that this method is called, and 406 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 407 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 408 /// make the Preprocessor re-lex the same tokens. 409 /// 410 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 411 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 412 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 413 /// 414 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 415 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 416 /// tokens will continue indefinitely. 417 /// 418 void EnableBacktrackAtThisPos(); 419 420 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 421 void CommitBacktrackedTokens(); 422 423 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 424 /// EnableBacktrackAtThisPos() was previously called. 425 void Backtrack(); 426 427 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 428 /// caching of tokens is on. 429 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 430 431 /// Lex - To lex a token from the preprocessor, just pull a token from the 432 /// current lexer or macro object. 433 void Lex(Token &Result) { 434 if (CurLexer) 435 CurLexer->Lex(Result); 436 else if (CurPTHLexer) 437 CurPTHLexer->Lex(Result); 438 else if (CurTokenLexer) 439 CurTokenLexer->Lex(Result); 440 else 441 CachingLex(Result); 442 } 443 444 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 445 /// something not a comment. This is useful in -E -C mode where comments 446 /// would foul up preprocessor directive handling. 447 void LexNonComment(Token &Result) { 448 do 449 Lex(Result); 450 while (Result.getKind() == tok::comment); 451 } 452 453 /// LexUnexpandedToken - This is just like Lex, but this disables macro 454 /// expansion of identifier tokens. 455 void LexUnexpandedToken(Token &Result) { 456 // Disable macro expansion. 457 bool OldVal = DisableMacroExpansion; 458 DisableMacroExpansion = true; 459 // Lex the token. 460 Lex(Result); 461 462 // Reenable it. 463 DisableMacroExpansion = OldVal; 464 } 465 466 /// LookAhead - This peeks ahead N tokens and returns that token without 467 /// consuming any tokens. LookAhead(0) returns the next token that would be 468 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 469 /// returns normal tokens after phase 5. As such, it is equivalent to using 470 /// 'Lex', not 'LexUnexpandedToken'. 471 const Token &LookAhead(unsigned N) { 472 if (CachedLexPos + N < CachedTokens.size()) 473 return CachedTokens[CachedLexPos+N]; 474 else 475 return PeekAhead(N+1); 476 } 477 478 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 479 /// this allows to revert a specific number of tokens. 480 /// Note that the number of tokens being reverted should be up to the last 481 /// backtrack position, not more. 482 void RevertCachedTokens(unsigned N) { 483 assert(isBacktrackEnabled() && 484 "Should only be called when tokens are cached for backtracking"); 485 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 486 && "Should revert tokens up to the last backtrack position, not more"); 487 assert(signed(CachedLexPos) - signed(N) >= 0 && 488 "Corrupted backtrack positions ?"); 489 CachedLexPos -= N; 490 } 491 492 /// EnterToken - Enters a token in the token stream to be lexed next. If 493 /// BackTrack() is called afterwards, the token will remain at the insertion 494 /// point. 495 void EnterToken(const Token &Tok) { 496 EnterCachingLexMode(); 497 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 498 } 499 500 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 501 /// tokens (because backtrack is enabled) it should replace the most recent 502 /// cached tokens with the given annotation token. This function has no effect 503 /// if backtracking is not enabled. 504 /// 505 /// Note that the use of this function is just for optimization; so that the 506 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 507 /// invoked. 508 void AnnotateCachedTokens(const Token &Tok) { 509 assert(Tok.isAnnotation() && "Expected annotation token"); 510 if (CachedLexPos != 0 && isBacktrackEnabled()) 511 AnnotatePreviousCachedTokens(Tok); 512 } 513 514 /// \brief Replace the last token with an annotation token. 515 /// 516 /// Like AnnotateCachedTokens(), this routine replaces an 517 /// already-parsed (and resolved) token with an annotation 518 /// token. However, this routine only replaces the last token with 519 /// the annotation token; it does not affect any other cached 520 /// tokens. This function has no effect if backtracking is not 521 /// enabled. 522 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 523 assert(Tok.isAnnotation() && "Expected annotation token"); 524 if (CachedLexPos != 0 && isBacktrackEnabled()) 525 CachedTokens[CachedLexPos-1] = Tok; 526 } 527 528 /// \brief Specify the point at which code-completion will be performed. 529 /// 530 /// \param File the file in which code completion should occur. If 531 /// this file is included multiple times, code-completion will 532 /// perform completion the first time it is included. If NULL, this 533 /// function clears out the code-completion point. 534 /// 535 /// \param Line the line at which code completion should occur 536 /// (1-based). 537 /// 538 /// \param Column the column at which code completion should occur 539 /// (1-based). 540 /// 541 /// \returns true if an error occurred, false otherwise. 542 bool SetCodeCompletionPoint(const FileEntry *File, 543 unsigned Line, unsigned Column); 544 545 /// \brief Determine if this source location refers into the file 546 /// for which we are performing code completion. 547 bool isCodeCompletionFile(SourceLocation FileLoc) const; 548 549 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 550 /// the specified Token's location, translating the token's start 551 /// position in the current buffer into a SourcePosition object for rendering. 552 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 553 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 554 } 555 556 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 557 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 558 DiagID); 559 } 560 561 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 562 /// token is the characters used to represent the token in the source file 563 /// after trigraph expansion and escaped-newline folding. In particular, this 564 /// wants to get the true, uncanonicalized, spelling of things like digraphs 565 /// UCNs, etc. 566 /// 567 /// \param Invalid If non-NULL, will be set \c true if an error occurs. 568 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const; 569 570 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 571 /// token is the characters used to represent the token in the source file 572 /// after trigraph expansion and escaped-newline folding. In particular, this 573 /// wants to get the true, uncanonicalized, spelling of things like digraphs 574 /// UCNs, etc. 575 static std::string getSpelling(const Token &Tok, 576 const SourceManager &SourceMgr, 577 const LangOptions &Features, 578 bool *Invalid = 0); 579 580 /// getSpelling - This method is used to get the spelling of a token into a 581 /// preallocated buffer, instead of as an std::string. The caller is required 582 /// to allocate enough space for the token, which is guaranteed to be at least 583 /// Tok.getLength() bytes long. The length of the actual result is returned. 584 /// 585 /// Note that this method may do two possible things: it may either fill in 586 /// the buffer specified with characters, or it may *change the input pointer* 587 /// to point to a constant buffer with the data already in it (avoiding a 588 /// copy). The caller is not allowed to modify the returned buffer pointer 589 /// if an internal buffer is returned. 590 unsigned getSpelling(const Token &Tok, const char *&Buffer, 591 bool *Invalid = 0) const; 592 593 /// getSpelling - This method is used to get the spelling of a token into a 594 /// SmallVector. Note that the returned StringRef may not point to the 595 /// supplied buffer if a copy can be avoided. 596 llvm::StringRef getSpelling(const Token &Tok, 597 llvm::SmallVectorImpl<char> &Buffer, 598 bool *Invalid = 0) const; 599 600 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 601 /// with length 1, return the character. 602 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 603 bool *Invalid = 0) const { 604 assert(Tok.is(tok::numeric_constant) && 605 Tok.getLength() == 1 && "Called on unsupported token"); 606 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 607 608 // If the token is carrying a literal data pointer, just use it. 609 if (const char *D = Tok.getLiteralData()) 610 return *D; 611 612 // Otherwise, fall back on getCharacterData, which is slower, but always 613 // works. 614 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 615 } 616 617 /// CreateString - Plop the specified string into a scratch buffer and set the 618 /// specified token's location and length to it. If specified, the source 619 /// location provides a location of the instantiation point of the token. 620 void CreateString(const char *Buf, unsigned Len, 621 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 622 623 /// \brief Computes the source location just past the end of the 624 /// token at this source location. 625 /// 626 /// This routine can be used to produce a source location that 627 /// points just past the end of the token referenced by \p Loc, and 628 /// is generally used when a diagnostic needs to point just after a 629 /// token where it expected something different that it received. If 630 /// the returned source location would not be meaningful (e.g., if 631 /// it points into a macro), this routine returns an invalid 632 /// source location. 633 /// 634 /// \param Offset an offset from the end of the token, where the source 635 /// location should refer to. The default offset (0) produces a source 636 /// location pointing just past the end of the token; an offset of 1 produces 637 /// a source location pointing to the last character in the token, etc. 638 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0); 639 640 /// DumpToken - Print the token to stderr, used for debugging. 641 /// 642 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 643 void DumpLocation(SourceLocation Loc) const; 644 void DumpMacro(const MacroInfo &MI) const; 645 646 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 647 /// token, return a new location that specifies a character within the token. 648 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 649 650 /// IncrementPasteCounter - Increment the counters for the number of token 651 /// paste operations performed. If fast was specified, this is a 'fast paste' 652 /// case we handled. 653 /// 654 void IncrementPasteCounter(bool isFast) { 655 if (isFast) 656 ++NumFastTokenPaste; 657 else 658 ++NumTokenPaste; 659 } 660 661 void PrintStats(); 662 663 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 664 /// comment (/##/) in microsoft mode, this method handles updating the current 665 /// state, returning the token on the next source line. 666 void HandleMicrosoftCommentPaste(Token &Tok); 667 668 //===--------------------------------------------------------------------===// 669 // Preprocessor callback methods. These are invoked by a lexer as various 670 // directives and events are found. 671 672 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 673 /// identifier information for the token and install it into the token. 674 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 675 const char *BufPtr = 0) const; 676 677 /// HandleIdentifier - This callback is invoked when the lexer reads an 678 /// identifier and has filled in the tokens IdentifierInfo member. This 679 /// callback potentially macro expands it or turns it into a named token (like 680 /// 'for'). 681 void HandleIdentifier(Token &Identifier); 682 683 684 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 685 /// the current file. This either returns the EOF token and returns true, or 686 /// pops a level off the include stack and returns false, at which point the 687 /// client should call lex again. 688 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 689 690 /// HandleEndOfTokenLexer - This callback is invoked when the current 691 /// TokenLexer hits the end of its token stream. 692 bool HandleEndOfTokenLexer(Token &Result); 693 694 /// HandleDirective - This callback is invoked when the lexer sees a # token 695 /// at the start of a line. This consumes the directive, modifies the 696 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 697 /// read is the correct one. 698 void HandleDirective(Token &Result); 699 700 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 701 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 702 /// true, then we consider macros that expand to zero tokens as being ok. 703 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 704 705 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 706 /// current line until the tok::eom token is found. 707 void DiscardUntilEndOfDirective(); 708 709 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 710 /// __DATE__ or __TIME__ in the file so far. 711 bool SawDateOrTime() const { 712 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 713 } 714 unsigned getCounterValue() const { return CounterValue; } 715 void setCounterValue(unsigned V) { CounterValue = V; } 716 717 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 718 /// SourceLocation. 719 MacroInfo* AllocateMacroInfo(SourceLocation L); 720 721 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 722 /// checked and spelled filename, e.g. as an operand of #include. This returns 723 /// true if the input filename was in <>'s or false if it were in ""'s. The 724 /// caller is expected to provide a buffer that is large enough to hold the 725 /// spelling of the filename, but is also expected to handle the case when 726 /// this method decides to use a different buffer. 727 bool GetIncludeFilenameSpelling(SourceLocation Loc,llvm::StringRef &Filename); 728 729 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 730 /// return null on failure. isAngled indicates whether the file reference is 731 /// for system #include's or not (i.e. using <> instead of ""). 732 const FileEntry *LookupFile(llvm::StringRef Filename, 733 bool isAngled, const DirectoryLookup *FromDir, 734 const DirectoryLookup *&CurDir); 735 736 /// GetCurLookup - The DirectoryLookup structure used to find the current 737 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 738 /// implement #include_next and find directory-specific properties. 739 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 740 741 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 742 /// #include. 743 bool isInPrimaryFile() const; 744 745 /// ConcatenateIncludeName - Handle cases where the #include name is expanded 746 /// from a macro as multiple tokens, which need to be glued together. This 747 /// occurs for code like: 748 /// #define FOO <a/b.h> 749 /// #include FOO 750 /// because in this case, "<a/b.h>" is returned as 7 tokens, not one. 751 /// 752 /// This code concatenates and consumes tokens up to the '>' token. It returns 753 /// false if the > was found, otherwise it returns true if it finds and consumes 754 /// the EOM marker. 755 bool ConcatenateIncludeName(llvm::SmallString<128> &FilenameBuffer); 756 757private: 758 759 void PushIncludeMacroStack() { 760 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 761 CurPTHLexer.take(), 762 CurPPLexer, 763 CurTokenLexer.take(), 764 CurDirLookup)); 765 CurPPLexer = 0; 766 } 767 768 void PopIncludeMacroStack() { 769 CurLexer.reset(IncludeMacroStack.back().TheLexer); 770 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 771 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 772 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 773 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 774 IncludeMacroStack.pop_back(); 775 } 776 777 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 778 /// be reused for allocating new MacroInfo objects. 779 void ReleaseMacroInfo(MacroInfo* MI); 780 781 /// ReadMacroName - Lex and validate a macro name, which occurs after a 782 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 783 /// and discards the rest of the macro line if the macro name is invalid. 784 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 785 786 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 787 /// definition has just been read. Lex the rest of the arguments and the 788 /// closing ), updating MI with what we learn. Return true if an error occurs 789 /// parsing the arg list. 790 bool ReadMacroDefinitionArgList(MacroInfo *MI); 791 792 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 793 /// decided that the subsequent tokens are in the #if'd out portion of the 794 /// file. Lex the rest of the file, until we see an #endif. If 795 /// FoundNonSkipPortion is true, then we have already emitted code for part of 796 /// this #if directive, so #else/#elif blocks should never be entered. If 797 /// FoundElse is false, then #else directives are ok, if not, then we have 798 /// already seen one so a #else directive is a duplicate. When this returns, 799 /// the caller can lex the first valid token. 800 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 801 bool FoundNonSkipPortion, bool FoundElse); 802 803 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 804 /// SkipExcludedConditionalBlock. 805 void PTHSkipExcludedConditionalBlock(); 806 807 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 808 /// may occur after a #if or #elif directive and return it as a bool. If the 809 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 810 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 811 812 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 813 /// #pragma GCC poison/system_header/dependency and #pragma once. 814 void RegisterBuiltinPragmas(); 815 816 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 817 /// identifier table. 818 void RegisterBuiltinMacros(); 819 820 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 821 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 822 /// the macro should not be expanded return true, otherwise return false. 823 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 824 825 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 826 /// lexed is a '('. If so, consume the token and return true, if not, this 827 /// method should have no observable side-effect on the lexed tokens. 828 bool isNextPPTokenLParen(); 829 830 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 831 /// invoked to read all of the formal arguments specified for the macro 832 /// invocation. This returns null on error. 833 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 834 SourceLocation &InstantiationEnd); 835 836 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 837 /// as a builtin macro, handle it and return the next token as 'Tok'. 838 void ExpandBuiltinMacro(Token &Tok); 839 840 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 841 /// return the first token after the directive. The _Pragma token has just 842 /// been read into 'Tok'. 843 void Handle_Pragma(Token &Tok); 844 845 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 846 /// start lexing tokens from it instead of the current buffer. 847 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 848 849 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 850 /// start getting tokens from it using the PTH cache. 851 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 852 853 /// IsFileLexer - Returns true if we are lexing from a file and not a 854 /// pragma or a macro. 855 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 856 return L ? !L->isPragmaLexer() : P != 0; 857 } 858 859 static bool IsFileLexer(const IncludeStackInfo& I) { 860 return IsFileLexer(I.TheLexer, I.ThePPLexer); 861 } 862 863 bool IsFileLexer() const { 864 return IsFileLexer(CurLexer.get(), CurPPLexer); 865 } 866 867 //===--------------------------------------------------------------------===// 868 // Caching stuff. 869 void CachingLex(Token &Result); 870 bool InCachingLexMode() const { return CurPPLexer == 0 && CurTokenLexer == 0;} 871 void EnterCachingLexMode(); 872 void ExitCachingLexMode() { 873 if (InCachingLexMode()) 874 RemoveTopOfLexerStack(); 875 } 876 const Token &PeekAhead(unsigned N); 877 void AnnotatePreviousCachedTokens(const Token &Tok); 878 879 //===--------------------------------------------------------------------===// 880 /// Handle*Directive - implement the various preprocessor directives. These 881 /// should side-effect the current preprocessor object so that the next call 882 /// to Lex() will return the appropriate token next. 883 void HandleLineDirective(Token &Tok); 884 void HandleDigitDirective(Token &Tok); 885 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 886 void HandleIdentSCCSDirective(Token &Tok); 887 888 // File inclusion. 889 void HandleIncludeDirective(Token &Tok, 890 const DirectoryLookup *LookupFrom = 0, 891 bool isImport = false); 892 void HandleIncludeNextDirective(Token &Tok); 893 void HandleIncludeMacrosDirective(Token &Tok); 894 void HandleImportDirective(Token &Tok); 895 896 // Macro handling. 897 void HandleDefineDirective(Token &Tok); 898 void HandleUndefDirective(Token &Tok); 899 // HandleAssertDirective(Token &Tok); 900 // HandleUnassertDirective(Token &Tok); 901 902 // Conditional Inclusion. 903 void HandleIfdefDirective(Token &Tok, bool isIfndef, 904 bool ReadAnyTokensBeforeDirective); 905 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 906 void HandleEndifDirective(Token &Tok); 907 void HandleElseDirective(Token &Tok); 908 void HandleElifDirective(Token &Tok); 909 910 // Pragmas. 911 void HandlePragmaDirective(); 912public: 913 void HandlePragmaOnce(Token &OnceTok); 914 void HandlePragmaMark(); 915 void HandlePragmaPoison(Token &PoisonTok); 916 void HandlePragmaSystemHeader(Token &SysHeaderTok); 917 void HandlePragmaDependency(Token &DependencyTok); 918 void HandlePragmaComment(Token &CommentTok); 919 // Return true and store the first token only if any CommentHandler 920 // has inserted some tokens and getCommentRetentionState() is false. 921 bool HandleComment(Token &Token, SourceRange Comment); 922}; 923 924/// \brief Abstract base class that describes a handler that will receive 925/// source ranges for each of the comments encountered in the source file. 926class CommentHandler { 927public: 928 virtual ~CommentHandler(); 929 930 // The handler shall return true if it has pushed any tokens 931 // to be read using e.g. EnterToken or EnterTokenStream. 932 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 933}; 934 935} // end namespace clang 936 937#endif 938