Preprocessor.h revision ec6c574478a22008847d7ebc2498ef3336752096
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 pointer, if non-null, are the predefined macros that 134 /// preprocessor should use from the command line etc. 135 const char *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 void setPredefines(const char *P) { 200 Predefines = P; 201 } 202 203 /// getIdentifierInfo - Return information about the specified preprocessor 204 /// identifier token. The version of this method that takes two character 205 /// pointers is preferred unless the identifier is already available as a 206 /// string (this avoids allocation and copying of memory to construct an 207 /// std::string). 208 IdentifierInfo *getIdentifierInfo(const char *NameStart, 209 const char *NameEnd) { 210 return &Identifiers.get(NameStart, NameEnd); 211 } 212 IdentifierInfo *getIdentifierInfo(const char *NameStr) { 213 return getIdentifierInfo(NameStr, NameStr+strlen(NameStr)); 214 } 215 216 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 217 /// If 'Namespace' is non-null, then it is a token required to exist on the 218 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 219 void AddPragmaHandler(const char *Namespace, PragmaHandler *Handler); 220 221 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 222 /// which implicitly adds the builtin defines etc. 223 void EnterMainSourceFile(); 224 225 /// EnterSourceFile - Add a source file to the top of the include stack and 226 /// start lexing tokens from it instead of the current buffer. If isMainFile 227 /// is true, this is the main file for the translation unit. 228 void EnterSourceFile(unsigned CurFileID, const DirectoryLookup *Dir); 229 230 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 231 /// tokens from it instead of the current buffer. Args specifies the 232 /// tokens input to a function-like macro. 233 void EnterMacro(Token &Identifier, MacroArgs *Args); 234 235 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 236 /// which will cause the lexer to start returning the specified tokens. 237 /// 238 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 239 /// not be subject to further macro expansion. Otherwise, these tokens will 240 /// be re-macro-expanded when/if expansion is enabled. 241 /// 242 /// If OwnsTokens is false, this method assumes that the specified stream of 243 /// tokens has a permanent owner somewhere, so they do not need to be copied. 244 /// If it is true, it assumes the array of tokens is allocated with new[] and 245 /// must be freed. 246 /// 247 void EnterTokenStream(const Token *Toks, unsigned NumToks, 248 bool DisableMacroExpansion, bool OwnsTokens); 249 250 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 251 /// lexer stack. This should only be used in situations where the current 252 /// state of the top-of-stack lexer is known. 253 void RemoveTopOfLexerStack(); 254 255 /// Lex - To lex a token from the preprocessor, just pull a token from the 256 /// current lexer or macro object. 257 void Lex(Token &Result) { 258 if (CurLexer) 259 CurLexer->Lex(Result); 260 else 261 CurTokenLexer->Lex(Result); 262 } 263 264 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 265 /// something not a comment. This is useful in -E -C mode where comments 266 /// would foul up preprocessor directive handling. 267 void LexNonComment(Token &Result) { 268 do 269 Lex(Result); 270 while (Result.getKind() == tok::comment); 271 } 272 273 /// LexUnexpandedToken - This is just like Lex, but this disables macro 274 /// expansion of identifier tokens. 275 void LexUnexpandedToken(Token &Result) { 276 // Disable macro expansion. 277 bool OldVal = DisableMacroExpansion; 278 DisableMacroExpansion = true; 279 // Lex the token. 280 Lex(Result); 281 282 // Reenable it. 283 DisableMacroExpansion = OldVal; 284 } 285 286 /// LookAhead - This peeks ahead N tokens and returns that token without 287 /// consuming any tokens. LookAhead(0) returns the next token that would be 288 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 289 /// returns normal tokens after phase 5. As such, it is equivalent to using 290 /// 'Lex', not 'LexUnexpandedToken'. 291 /// 292 /// NOTE: is a relatively expensive method, so it should not be used in common 293 /// code paths if possible! 294 /// 295 Token LookAhead(unsigned N); 296 297 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 298 /// the specified Token's location, translating the token's start 299 /// position in the current buffer into a SourcePosition object for rendering. 300 void Diag(SourceLocation Loc, unsigned DiagID); 301 void Diag(SourceLocation Loc, unsigned DiagID, const std::string &Msg); 302 void Diag(const Token &Tok, unsigned DiagID) { 303 Diag(Tok.getLocation(), DiagID); 304 } 305 void Diag(const Token &Tok, unsigned DiagID, const std::string &Msg) { 306 Diag(Tok.getLocation(), DiagID, Msg); 307 } 308 309 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 310 /// token is the characters used to represent the token in the source file 311 /// after trigraph expansion and escaped-newline folding. In particular, this 312 /// wants to get the true, uncanonicalized, spelling of things like digraphs 313 /// UCNs, etc. 314 std::string getSpelling(const Token &Tok) const; 315 316 /// getSpelling - This method is used to get the spelling of a token into a 317 /// preallocated buffer, instead of as an std::string. The caller is required 318 /// to allocate enough space for the token, which is guaranteed to be at least 319 /// Tok.getLength() bytes long. The length of the actual result is returned. 320 /// 321 /// Note that this method may do two possible things: it may either fill in 322 /// the buffer specified with characters, or it may *change the input pointer* 323 /// to point to a constant buffer with the data already in it (avoiding a 324 /// copy). The caller is not allowed to modify the returned buffer pointer 325 /// if an internal buffer is returned. 326 unsigned getSpelling(const Token &Tok, const char *&Buffer) const; 327 328 329 /// CreateString - Plop the specified string into a scratch buffer and return 330 /// a location for it. If specified, the source location provides a source 331 /// location for the token. 332 SourceLocation CreateString(const char *Buf, unsigned Len, 333 SourceLocation SourceLoc = SourceLocation()); 334 335 /// DumpToken - Print the token to stderr, used for debugging. 336 /// 337 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 338 void DumpLocation(SourceLocation Loc) const; 339 void DumpMacro(const MacroInfo &MI) const; 340 341 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 342 /// token, return a new location that specifies a character within the token. 343 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 344 345 /// IncrementPasteCounter - Increment the counters for the number of token 346 /// paste operations performed. If fast was specified, this is a 'fast paste' 347 /// case we handled. 348 /// 349 void IncrementPasteCounter(bool isFast) { 350 if (isFast) 351 ++NumFastTokenPaste; 352 else 353 ++NumTokenPaste; 354 } 355 356 void PrintStats(); 357 358 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 359 /// comment (/##/) in microsoft mode, this method handles updating the current 360 /// state, returning the token on the next source line. 361 void HandleMicrosoftCommentPaste(Token &Tok); 362 363 //===--------------------------------------------------------------------===// 364 // Preprocessor callback methods. These are invoked by a lexer as various 365 // directives and events are found. 366 367 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 368 /// identifier information for the token and install it into the token. 369 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 370 const char *BufPtr = 0); 371 372 /// HandleIdentifier - This callback is invoked when the lexer reads an 373 /// identifier and has filled in the tokens IdentifierInfo member. This 374 /// callback potentially macro expands it or turns it into a named token (like 375 /// 'for'). 376 void HandleIdentifier(Token &Identifier); 377 378 379 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 380 /// the current file. This either returns the EOF token and returns true, or 381 /// pops a level off the include stack and returns false, at which point the 382 /// client should call lex again. 383 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 384 385 /// HandleEndOfTokenLexer - This callback is invoked when the current 386 /// TokenLexer hits the end of its token stream. 387 bool HandleEndOfTokenLexer(Token &Result); 388 389 /// HandleDirective - This callback is invoked when the lexer sees a # token 390 /// at the start of a line. This consumes the directive, modifies the 391 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 392 /// read is the correct one. 393 void HandleDirective(Token &Result); 394 395 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 396 /// not, emit a diagnostic and consume up until the eom. 397 void CheckEndOfDirective(const char *Directive); 398private: 399 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 400 /// #include. 401 bool isInPrimaryFile() const; 402 403 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 404 /// current line until the tok::eom token is found. 405 void DiscardUntilEndOfDirective(); 406 407 /// ReadMacroName - Lex and validate a macro name, which occurs after a 408 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 409 /// and discards the rest of the macro line if the macro name is invalid. 410 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 411 412 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 413 /// definition has just been read. Lex the rest of the arguments and the 414 /// closing ), updating MI with what we learn. Return true if an error occurs 415 /// parsing the arg list. 416 bool ReadMacroDefinitionArgList(MacroInfo *MI); 417 418 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 419 /// decided that the subsequent tokens are in the #if'd out portion of the 420 /// file. Lex the rest of the file, until we see an #endif. If 421 /// FoundNonSkipPortion is true, then we have already emitted code for part of 422 /// this #if directive, so #else/#elif blocks should never be entered. If 423 /// FoundElse is false, then #else directives are ok, if not, then we have 424 /// already seen one so a #else directive is a duplicate. When this returns, 425 /// the caller can lex the first valid token. 426 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 427 bool FoundNonSkipPortion, bool FoundElse); 428 429 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 430 /// may occur after a #if or #elif directive and return it as a bool. If the 431 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 432 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 433 434 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 435 /// #pragma GCC poison/system_header/dependency and #pragma once. 436 void RegisterBuiltinPragmas(); 437 438 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 439 /// identifier table. 440 void RegisterBuiltinMacros(); 441 IdentifierInfo *RegisterBuiltinMacro(const char *Name); 442 443 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 444 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 445 /// the macro should not be expanded return true, otherwise return false. 446 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 447 448 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 449 /// lexed is a '('. If so, consume the token and return true, if not, this 450 /// method should have no observable side-effect on the lexed tokens. 451 bool isNextPPTokenLParen(); 452 453 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 454 /// invoked to read all of the formal arguments specified for the macro 455 /// invocation. This returns null on error. 456 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI); 457 458 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 459 /// as a builtin macro, handle it and return the next token as 'Tok'. 460 void ExpandBuiltinMacro(Token &Tok); 461 462 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 463 /// return the first token after the directive. The _Pragma token has just 464 /// been read into 'Tok'. 465 void Handle_Pragma(Token &Tok); 466 467 468 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 469 /// start lexing tokens from it instead of the current buffer. 470 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 471 472 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 473 /// checked and spelled filename, e.g. as an operand of #include. This returns 474 /// true if the input filename was in <>'s or false if it were in ""'s. The 475 /// caller is expected to provide a buffer that is large enough to hold the 476 /// spelling of the filename, but is also expected to handle the case when 477 /// this method decides to use a different buffer. 478 bool GetIncludeFilenameSpelling(SourceLocation Loc, 479 const char *&BufStart, const char *&BufEnd); 480 481 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 482 /// return null on failure. isAngled indicates whether the file reference is 483 /// for system #include's or not (i.e. using <> instead of ""). 484 const FileEntry *LookupFile(const char *FilenameStart,const char *FilenameEnd, 485 bool isAngled, const DirectoryLookup *FromDir, 486 const DirectoryLookup *&CurDir); 487 488 //===--------------------------------------------------------------------===// 489 /// Handle*Directive - implement the various preprocessor directives. These 490 /// should side-effect the current preprocessor object so that the next call 491 /// to Lex() will return the appropriate token next. 492 493 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 494 void HandleIdentSCCSDirective(Token &Tok); 495 496 // File inclusion. 497 void HandleIncludeDirective(Token &Tok, 498 const DirectoryLookup *LookupFrom = 0, 499 bool isImport = false); 500 void HandleIncludeNextDirective(Token &Tok); 501 void HandleImportDirective(Token &Tok); 502 503 // Macro handling. 504 void HandleDefineDirective(Token &Tok); 505 void HandleUndefDirective(Token &Tok); 506 // HandleAssertDirective(Token &Tok); 507 // HandleUnassertDirective(Token &Tok); 508 509 // Conditional Inclusion. 510 void HandleIfdefDirective(Token &Tok, bool isIfndef, 511 bool ReadAnyTokensBeforeDirective); 512 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 513 void HandleEndifDirective(Token &Tok); 514 void HandleElseDirective(Token &Tok); 515 void HandleElifDirective(Token &Tok); 516 517 // Pragmas. 518 void HandlePragmaDirective(); 519public: 520 void HandlePragmaOnce(Token &OnceTok); 521 void HandlePragmaMark(); 522 void HandlePragmaPoison(Token &PoisonTok); 523 void HandlePragmaSystemHeader(Token &SysHeaderTok); 524 void HandlePragmaDependency(Token &DependencyTok); 525}; 526 527/// PreprocessorFactory - A generic factory interface for lazily creating 528/// Preprocessor objects on-demand when they are needed. 529class PreprocessorFactory { 530public: 531 virtual ~PreprocessorFactory(); 532 virtual Preprocessor* CreatePreprocessor() = 0; 533}; 534 535} // end namespace clang 536 537#endif 538