Preprocessor.h revision aa39197431a0a0b1326ecf6b3be6a11f6e2f8503
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/PPCallbacks.h" 19#include "clang/Lex/TokenLexer.h" 20#include "clang/Basic/IdentifierTable.h" 21#include "clang/Basic/SourceLocation.h" 22#include "llvm/ADT/DenseMap.h" 23 24namespace clang { 25 26class SourceManager; 27class FileManager; 28class FileEntry; 29class HeaderSearch; 30class PragmaNamespace; 31class PragmaHandler; 32class ScratchBuffer; 33class TargetInfo; 34class PPCallbacks; 35class DirectoryLookup; 36 37/// Preprocessor - This object engages in a tight little dance with the lexer to 38/// efficiently preprocess tokens. Lexers know only about tokens within a 39/// single source file, and don't know anything about preprocessor-level issues 40/// like the #include stack, token expansion, etc. 41/// 42class Preprocessor { 43 Diagnostic &Diags; 44 const LangOptions &Features; 45 TargetInfo &Target; 46 FileManager &FileMgr; 47 SourceManager &SourceMgr; 48 ScratchBuffer *ScratchBuf; 49 HeaderSearch &HeaderInfo; 50 51 /// Identifiers for builtin macros and other builtins. 52 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 53 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 54 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 55 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 56 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 57 IdentifierInfo *Ident_Pragma, *Ident__VA_ARGS__; // _Pragma, __VA_ARGS__ 58 59 SourceLocation DATELoc, TIMELoc; 60 61 enum { 62 /// MaxIncludeStackDepth - Maximum depth of #includes. 63 MaxAllowedIncludeStackDepth = 200 64 }; 65 66 // State that is set before the preprocessor begins. 67 bool KeepComments : 1; 68 bool KeepMacroComments : 1; 69 70 // State that changes while the preprocessor runs: 71 bool DisableMacroExpansion : 1; // True if macro expansion is disabled. 72 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 73 74 /// Identifiers - This is mapping/lookup information for all identifiers in 75 /// the program, including program keywords. 76 IdentifierTable Identifiers; 77 78 /// Selectors - This table contains all the selectors in the program. Unlike 79 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 80 /// It is declared/instantiated here because it's role/lifetime is 81 /// conceptually similar the IdentifierTable. In addition, the current control 82 /// flow (in clang::ParseAST()), make it convenient to put here. 83 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 84 /// the lifetime fo the preprocessor. 85 SelectorTable Selectors; 86 87 /// PragmaHandlers - This tracks all of the pragmas that the client registered 88 /// with this preprocessor. 89 PragmaNamespace *PragmaHandlers; 90 91 /// CurLexer - This is the current top of the stack that we're lexing from if 92 /// not expanding a macro. One of CurLexer and CurTokenLexer must be null. 93 Lexer *CurLexer; 94 95 /// CurLookup - The DirectoryLookup structure used to find the current 96 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 97 /// implement #include_next and find directory-specific properties. 98 const DirectoryLookup *CurDirLookup; 99 100 /// CurTokenLexer - This is the current macro we are expanding, if we are 101 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 102 TokenLexer *CurTokenLexer; 103 104 /// IncludeMacroStack - This keeps track of the stack of files currently 105 /// #included, and macros currently being expanded from, not counting 106 /// CurLexer/CurTokenLexer. 107 struct IncludeStackInfo { 108 Lexer *TheLexer; 109 const DirectoryLookup *TheDirLookup; 110 TokenLexer *TheTokenLexer; 111 IncludeStackInfo(Lexer *L, const DirectoryLookup *D, TokenLexer *TL) 112 : TheLexer(L), TheDirLookup(D), TheTokenLexer(TL) { 113 } 114 }; 115 std::vector<IncludeStackInfo> IncludeMacroStack; 116 117 /// Callbacks - These are actions invoked when some preprocessor activity is 118 /// encountered (e.g. a file is #included, etc). 119 PPCallbacks *Callbacks; 120 121 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 122 /// to the actual definition of the macro. 123 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 124 125 // Various statistics we track for performance analysis. 126 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 127 unsigned NumIf, NumElse, NumEndif; 128 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 129 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 130 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 131 unsigned NumSkipped; 132 133 /// Predefines - This string is the predefined macros that preprocessor 134 /// should use from the command line etc. 135 std::string Predefines; 136 137 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 138 enum { TokenLexerCacheSize = 8 }; 139 unsigned NumCachedTokenLexers; 140 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 141public: 142 Preprocessor(Diagnostic &diags, const LangOptions &opts, TargetInfo &target, 143 SourceManager &SM, HeaderSearch &Headers); 144 ~Preprocessor(); 145 146 Diagnostic &getDiagnostics() const { return Diags; } 147 const LangOptions &getLangOptions() const { return Features; } 148 TargetInfo &getTargetInfo() const { return Target; } 149 FileManager &getFileManager() const { return FileMgr; } 150 SourceManager &getSourceManager() const { return SourceMgr; } 151 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 152 153 IdentifierTable &getIdentifierTable() { return Identifiers; } 154 SelectorTable &getSelectorTable() { return Selectors; } 155 156 inline FullSourceLoc getFullLoc(SourceLocation Loc) const { 157 return FullSourceLoc(Loc, getSourceManager()); 158 } 159 160 /// SetCommentRetentionState - Control whether or not the preprocessor retains 161 /// comments in output. 162 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 163 this->KeepComments = KeepComments | KeepMacroComments; 164 this->KeepMacroComments = KeepMacroComments; 165 } 166 167 bool getCommentRetentionState() const { return KeepComments; } 168 169 /// isCurrentLexer - Return true if we are lexing directly from the specified 170 /// lexer. 171 bool isCurrentLexer(const Lexer *L) const { 172 return CurLexer == L; 173 } 174 175 /// getCurrentLexer - Return the current file lexer being lexed from. Note 176 /// that this ignores any potentially active macro expansions and _Pragma 177 /// expansions going on at the time. 178 Lexer *getCurrentFileLexer() const; 179 180 /// getPPCallbacks/setPPCallbacks - Accessors for preprocessor callbacks. 181 /// Note that this class takes ownership of any PPCallbacks object given to 182 /// it. 183 PPCallbacks *getPPCallbacks() const { return Callbacks; } 184 void setPPCallbacks(PPCallbacks *C) { 185 delete Callbacks; 186 Callbacks = C; 187 } 188 189 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 190 /// or null if it isn't #define'd. 191 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 192 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 193 } 194 195 /// setMacroInfo - Specify a macro for this identifier. 196 /// 197 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 198 199 /// setPredefines - Set the predefines for this Preprocessor. These 200 /// predefines are automatically injected when parsing the main file. 201 void setPredefines(const char *P) { Predefines = P; } 202 void setPredefines(const std::string &P) { Predefines = P; } 203 204 /// getIdentifierInfo - Return information about the specified preprocessor 205 /// identifier token. The version of this method that takes two character 206 /// pointers is preferred unless the identifier is already available as a 207 /// string (this avoids allocation and copying of memory to construct an 208 /// std::string). 209 IdentifierInfo *getIdentifierInfo(const char *NameStart, 210 const char *NameEnd) { 211 return &Identifiers.get(NameStart, NameEnd); 212 } 213 IdentifierInfo *getIdentifierInfo(const char *NameStr) { 214 return getIdentifierInfo(NameStr, NameStr+strlen(NameStr)); 215 } 216 217 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 218 /// If 'Namespace' is non-null, then it is a token required to exist on the 219 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 220 void AddPragmaHandler(const char *Namespace, PragmaHandler *Handler); 221 222 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 223 /// which implicitly adds the builtin defines etc. 224 void EnterMainSourceFile(); 225 226 /// EnterSourceFile - Add a source file to the top of the include stack and 227 /// start lexing tokens from it instead of the current buffer. If isMainFile 228 /// is true, this is the main file for the translation unit. 229 void EnterSourceFile(unsigned CurFileID, const DirectoryLookup *Dir); 230 231 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 232 /// tokens from it instead of the current buffer. Args specifies the 233 /// tokens input to a function-like macro. 234 void EnterMacro(Token &Identifier, MacroArgs *Args); 235 236 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 237 /// which will cause the lexer to start returning the specified tokens. 238 /// 239 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 240 /// not be subject to further macro expansion. Otherwise, these tokens will 241 /// be re-macro-expanded when/if expansion is enabled. 242 /// 243 /// If OwnsTokens is false, this method assumes that the specified stream of 244 /// tokens has a permanent owner somewhere, so they do not need to be copied. 245 /// If it is true, it assumes the array of tokens is allocated with new[] and 246 /// must be freed. 247 /// 248 void EnterTokenStream(const Token *Toks, unsigned NumToks, 249 bool DisableMacroExpansion, bool OwnsTokens); 250 251 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 252 /// lexer stack. This should only be used in situations where the current 253 /// state of the top-of-stack lexer is known. 254 void RemoveTopOfLexerStack(); 255 256 /// Lex - To lex a token from the preprocessor, just pull a token from the 257 /// current lexer or macro object. 258 void Lex(Token &Result) { 259 if (CurLexer) 260 CurLexer->Lex(Result); 261 else 262 CurTokenLexer->Lex(Result); 263 } 264 265 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 266 /// something not a comment. This is useful in -E -C mode where comments 267 /// would foul up preprocessor directive handling. 268 void LexNonComment(Token &Result) { 269 do 270 Lex(Result); 271 while (Result.getKind() == tok::comment); 272 } 273 274 /// LexUnexpandedToken - This is just like Lex, but this disables macro 275 /// expansion of identifier tokens. 276 void LexUnexpandedToken(Token &Result) { 277 // Disable macro expansion. 278 bool OldVal = DisableMacroExpansion; 279 DisableMacroExpansion = true; 280 // Lex the token. 281 Lex(Result); 282 283 // Reenable it. 284 DisableMacroExpansion = OldVal; 285 } 286 287 /// LookAhead - This peeks ahead N tokens and returns that token without 288 /// consuming any tokens. LookAhead(0) returns the next token that would be 289 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 290 /// returns normal tokens after phase 5. As such, it is equivalent to using 291 /// 'Lex', not 'LexUnexpandedToken'. 292 /// 293 /// NOTE: is a relatively expensive method, so it should not be used in common 294 /// code paths if possible! 295 /// 296 Token LookAhead(unsigned N); 297 298 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 299 /// the specified Token's location, translating the token's start 300 /// position in the current buffer into a SourcePosition object for rendering. 301 void Diag(SourceLocation Loc, unsigned DiagID); 302 void Diag(SourceLocation Loc, unsigned DiagID, const std::string &Msg); 303 void Diag(const Token &Tok, unsigned DiagID) { 304 Diag(Tok.getLocation(), DiagID); 305 } 306 void Diag(const Token &Tok, unsigned DiagID, const std::string &Msg) { 307 Diag(Tok.getLocation(), DiagID, Msg); 308 } 309 310 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 311 /// token is the characters used to represent the token in the source file 312 /// after trigraph expansion and escaped-newline folding. In particular, this 313 /// wants to get the true, uncanonicalized, spelling of things like digraphs 314 /// UCNs, etc. 315 std::string getSpelling(const Token &Tok) const; 316 317 /// getSpelling - This method is used to get the spelling of a token into a 318 /// preallocated buffer, instead of as an std::string. The caller is required 319 /// to allocate enough space for the token, which is guaranteed to be at least 320 /// Tok.getLength() bytes long. The length of the actual result is returned. 321 /// 322 /// Note that this method may do two possible things: it may either fill in 323 /// the buffer specified with characters, or it may *change the input pointer* 324 /// to point to a constant buffer with the data already in it (avoiding a 325 /// copy). The caller is not allowed to modify the returned buffer pointer 326 /// if an internal buffer is returned. 327 unsigned getSpelling(const Token &Tok, const char *&Buffer) const; 328 329 330 /// CreateString - Plop the specified string into a scratch buffer and return 331 /// a location for it. If specified, the source location provides a source 332 /// location for the token. 333 SourceLocation CreateString(const char *Buf, unsigned Len, 334 SourceLocation SourceLoc = SourceLocation()); 335 336 /// DumpToken - Print the token to stderr, used for debugging. 337 /// 338 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 339 void DumpLocation(SourceLocation Loc) const; 340 void DumpMacro(const MacroInfo &MI) const; 341 342 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 343 /// token, return a new location that specifies a character within the token. 344 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 345 346 /// IncrementPasteCounter - Increment the counters for the number of token 347 /// paste operations performed. If fast was specified, this is a 'fast paste' 348 /// case we handled. 349 /// 350 void IncrementPasteCounter(bool isFast) { 351 if (isFast) 352 ++NumFastTokenPaste; 353 else 354 ++NumTokenPaste; 355 } 356 357 void PrintStats(); 358 359 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 360 /// comment (/##/) in microsoft mode, this method handles updating the current 361 /// state, returning the token on the next source line. 362 void HandleMicrosoftCommentPaste(Token &Tok); 363 364 //===--------------------------------------------------------------------===// 365 // Preprocessor callback methods. These are invoked by a lexer as various 366 // directives and events are found. 367 368 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 369 /// identifier information for the token and install it into the token. 370 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 371 const char *BufPtr = 0); 372 373 /// HandleIdentifier - This callback is invoked when the lexer reads an 374 /// identifier and has filled in the tokens IdentifierInfo member. This 375 /// callback potentially macro expands it or turns it into a named token (like 376 /// 'for'). 377 void HandleIdentifier(Token &Identifier); 378 379 380 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 381 /// the current file. This either returns the EOF token and returns true, or 382 /// pops a level off the include stack and returns false, at which point the 383 /// client should call lex again. 384 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 385 386 /// HandleEndOfTokenLexer - This callback is invoked when the current 387 /// TokenLexer hits the end of its token stream. 388 bool HandleEndOfTokenLexer(Token &Result); 389 390 /// HandleDirective - This callback is invoked when the lexer sees a # token 391 /// at the start of a line. This consumes the directive, modifies the 392 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 393 /// read is the correct one. 394 void HandleDirective(Token &Result); 395 396 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 397 /// not, emit a diagnostic and consume up until the eom. 398 void CheckEndOfDirective(const char *Directive); 399private: 400 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 401 /// #include. 402 bool isInPrimaryFile() const; 403 404 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 405 /// current line until the tok::eom token is found. 406 void DiscardUntilEndOfDirective(); 407 408 /// ReadMacroName - Lex and validate a macro name, which occurs after a 409 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 410 /// and discards the rest of the macro line if the macro name is invalid. 411 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 412 413 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 414 /// definition has just been read. Lex the rest of the arguments and the 415 /// closing ), updating MI with what we learn. Return true if an error occurs 416 /// parsing the arg list. 417 bool ReadMacroDefinitionArgList(MacroInfo *MI); 418 419 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 420 /// decided that the subsequent tokens are in the #if'd out portion of the 421 /// file. Lex the rest of the file, until we see an #endif. If 422 /// FoundNonSkipPortion is true, then we have already emitted code for part of 423 /// this #if directive, so #else/#elif blocks should never be entered. If 424 /// FoundElse is false, then #else directives are ok, if not, then we have 425 /// already seen one so a #else directive is a duplicate. When this returns, 426 /// the caller can lex the first valid token. 427 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 428 bool FoundNonSkipPortion, bool FoundElse); 429 430 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 431 /// may occur after a #if or #elif directive and return it as a bool. If the 432 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 433 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 434 435 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 436 /// #pragma GCC poison/system_header/dependency and #pragma once. 437 void RegisterBuiltinPragmas(); 438 439 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 440 /// identifier table. 441 void RegisterBuiltinMacros(); 442 IdentifierInfo *RegisterBuiltinMacro(const char *Name); 443 444 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 445 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 446 /// the macro should not be expanded return true, otherwise return false. 447 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 448 449 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 450 /// lexed is a '('. If so, consume the token and return true, if not, this 451 /// method should have no observable side-effect on the lexed tokens. 452 bool isNextPPTokenLParen(); 453 454 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 455 /// invoked to read all of the formal arguments specified for the macro 456 /// invocation. This returns null on error. 457 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI); 458 459 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 460 /// as a builtin macro, handle it and return the next token as 'Tok'. 461 void ExpandBuiltinMacro(Token &Tok); 462 463 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 464 /// return the first token after the directive. The _Pragma token has just 465 /// been read into 'Tok'. 466 void Handle_Pragma(Token &Tok); 467 468 469 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 470 /// start lexing tokens from it instead of the current buffer. 471 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 472 473 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 474 /// checked and spelled filename, e.g. as an operand of #include. This returns 475 /// true if the input filename was in <>'s or false if it were in ""'s. The 476 /// caller is expected to provide a buffer that is large enough to hold the 477 /// spelling of the filename, but is also expected to handle the case when 478 /// this method decides to use a different buffer. 479 bool GetIncludeFilenameSpelling(SourceLocation Loc, 480 const char *&BufStart, const char *&BufEnd); 481 482 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 483 /// return null on failure. isAngled indicates whether the file reference is 484 /// for system #include's or not (i.e. using <> instead of ""). 485 const FileEntry *LookupFile(const char *FilenameStart,const char *FilenameEnd, 486 bool isAngled, const DirectoryLookup *FromDir, 487 const DirectoryLookup *&CurDir); 488 489 //===--------------------------------------------------------------------===// 490 /// Handle*Directive - implement the various preprocessor directives. These 491 /// should side-effect the current preprocessor object so that the next call 492 /// to Lex() will return the appropriate token next. 493 494 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 495 void HandleIdentSCCSDirective(Token &Tok); 496 497 // File inclusion. 498 void HandleIncludeDirective(Token &Tok, 499 const DirectoryLookup *LookupFrom = 0, 500 bool isImport = false); 501 void HandleIncludeNextDirective(Token &Tok); 502 void HandleImportDirective(Token &Tok); 503 504 // Macro handling. 505 void HandleDefineDirective(Token &Tok); 506 void HandleUndefDirective(Token &Tok); 507 // HandleAssertDirective(Token &Tok); 508 // HandleUnassertDirective(Token &Tok); 509 510 // Conditional Inclusion. 511 void HandleIfdefDirective(Token &Tok, bool isIfndef, 512 bool ReadAnyTokensBeforeDirective); 513 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 514 void HandleEndifDirective(Token &Tok); 515 void HandleElseDirective(Token &Tok); 516 void HandleElifDirective(Token &Tok); 517 518 // Pragmas. 519 void HandlePragmaDirective(); 520public: 521 void HandlePragmaOnce(Token &OnceTok); 522 void HandlePragmaMark(); 523 void HandlePragmaPoison(Token &PoisonTok); 524 void HandlePragmaSystemHeader(Token &SysHeaderTok); 525 void HandlePragmaDependency(Token &DependencyTok); 526}; 527 528/// PreprocessorFactory - A generic factory interface for lazily creating 529/// Preprocessor objects on-demand when they are needed. 530class PreprocessorFactory { 531public: 532 virtual ~PreprocessorFactory(); 533 virtual Preprocessor* CreatePreprocessor() = 0; 534}; 535 536} // end namespace clang 537 538#endif 539