Lexer.h revision d217773f106856a11879ec79dc468efefaf2ee75
1//===--- Lexer.h - C Language Family Lexer ----------------------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file was developed by Chris Lattner and is distributed under 6// the University of Illinois Open Source License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// This file defines the Lexer interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_LEXER_H 15#define LLVM_CLANG_LEXER_H 16 17#include "clang/Lex/Token.h" 18#include "clang/Lex/MultipleIncludeOpt.h" 19#include "clang/Basic/LangOptions.h" 20#include <string> 21#include <vector> 22#include <cassert> 23 24namespace llvm { 25 class MemoryBuffer; 26} 27 28namespace clang { 29class Diagnostic; 30class Preprocessor; 31 32/// Lexer - This provides a simple interface that turns a text buffer into a 33/// stream of tokens. This provides no support for file reading or buffering, 34/// or buffering/seeking of tokens, only forward lexing is supported. It relies 35/// on the specified Preprocessor object to handle preprocessor directives, etc. 36class Lexer { 37 //===--------------------------------------------------------------------===// 38 // Constant configuration values for this lexer. 39 const char *BufferEnd; // End of the buffer. 40 const llvm::MemoryBuffer *InputFile; // The file we are reading from. 41 SourceLocation FileLoc; // Location for start of file. 42 Preprocessor &PP; // Preprocessor object controlling lexing. 43 LangOptions Features; // Features enabled by this language (cache). 44 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 45 bool IsMainFile; // True if top-level file. 46 47 //===--------------------------------------------------------------------===// 48 // Context-specific lexing flags set by the preprocessor. 49 // 50 51 /// ParsingPreprocessorDirective - This is true when parsing #XXX. This turns 52 /// '\n' into a tok::eom token. 53 bool ParsingPreprocessorDirective; 54 55 /// ParsingFilename - True after #include: this turns <xx> into a 56 /// tok::angle_string_literal token. 57 bool ParsingFilename; 58 59 /// LexingRawMode - True if in raw mode: This flag disables interpretation of 60 /// tokens and is a far faster mode to lex in than non-raw-mode. This flag: 61 /// 1. If EOF of the current lexer is found, the include stack isn't popped. 62 /// 2. Identifier information is not looked up for identifier tokens. As an 63 /// effect of this, implicit macro expansion is naturally disabled. 64 /// 3. "#" tokens at the start of a line are treated as normal tokens, not 65 /// implicitly transformed by the lexer. 66 /// 4. All diagnostic messages are disabled, except for unterminated /*. 67 /// 5. The only callback made into the preprocessor is to report a hard error 68 /// on an unterminated '/*' comment. 69 bool LexingRawMode; 70 71 /// KeepCommentMode - The lexer can optionally keep C & BCPL-style comments, 72 /// and return them as tokens. This is used for -C and -CC modes. 73 bool KeepCommentMode; 74 75 //===--------------------------------------------------------------------===// 76 // Context that changes as the file is lexed. 77 // NOTE: any state that mutates when in raw mode must have save/restore code 78 // in Lexer::isNextPPTokenLParen. 79 80 // BufferPtr - Current pointer into the buffer. This is the next character 81 // to be lexed. 82 const char *BufferPtr; 83 84 // IsAtStartOfLine - True if the next lexed token should get the "start of 85 // line" flag set on it. 86 bool IsAtStartOfLine; 87 88 /// MIOpt - This is a state machine that detects the #ifndef-wrapping a file 89 /// idiom for the multiple-include optimization. 90 MultipleIncludeOpt MIOpt; 91 92 /// ConditionalStack - Information about the set of #if/#ifdef/#ifndef blocks 93 /// we are currently in. 94 std::vector<PPConditionalInfo> ConditionalStack; 95 96 Lexer(const Lexer&); // DO NOT IMPLEMENT 97 void operator=(const Lexer&); // DO NOT IMPLEMENT 98 friend class Preprocessor; 99public: 100 101 /// Lexer constructor - Create a new lexer object for the specified buffer 102 /// with the specified preprocessor managing the lexing process. This lexer 103 /// assumes that the associated MemoryBuffer and Preprocessor objects will 104 /// outlive it, so it doesn't take ownership of either of them. 105 Lexer(SourceLocation FileLoc, Preprocessor &PP, 106 const char *BufStart = 0, const char *BufEnd = 0); 107 108 /// getFeatures - Return the language features currently enabled. NOTE: this 109 /// lexer modifies features as a file is parsed! 110 const LangOptions &getFeatures() const { return Features; } 111 112 /// getFileLoc - Return the File Location for the file we are lexing out of. 113 /// The physical location encodes the location where the characters come from, 114 /// the virtual location encodes where we should *claim* the characters came 115 /// from. Currently this is only used by _Pragma handling. 116 SourceLocation getFileLoc() const { return FileLoc; } 117 118 /// setIsMainFile - Mark this lexer as being the lexer for the top-level 119 /// source file. 120 void setIsMainFile() { 121 IsMainFile = true; 122 } 123 124 /// isMainFile - Return true if this is the top-level file. 125 /// 126 bool isMainFile() const { return IsMainFile; } 127 128 /// Lex - Return the next token in the file. If this is the end of file, it 129 /// return the tok::eof token. Return true if an error occurred and 130 /// compilation should terminate, false if normal. This implicitly involves 131 /// the preprocessor. 132 void Lex(Token &Result) { 133 // Start a new token. 134 Result.startToken(); 135 136 // NOTE, any changes here should also change code after calls to 137 // Preprocessor::HandleDirective 138 if (IsAtStartOfLine) { 139 Result.setFlag(Token::StartOfLine); 140 IsAtStartOfLine = false; 141 } 142 143 // Get a token. Note that this may delete the current lexer if the end of 144 // file is reached. 145 LexTokenInternal(Result); 146 } 147 148 /// LexRawToken - Switch the lexer to raw mode, lex a token into Result and 149 /// switch it back. Return true if the 'next character to read' pointer 150 /// points and the end of the lexer buffer, false otherwise. 151 bool LexRawToken(Token &Result) { 152 assert(!LexingRawMode && "Already in raw mode!"); 153 LexingRawMode = true; 154 Lex(Result); 155 LexingRawMode = false; 156 return BufferPtr == BufferEnd; 157 } 158 159 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 160 /// uninterpreted string. This switches the lexer out of directive mode. 161 std::string ReadToEndOfLine(); 162 163 164 /// Diag - Forwarding function for diagnostics. This translate a source 165 /// position in the current buffer into a SourceLocation object for rendering. 166 void Diag(const char *Loc, unsigned DiagID, 167 const std::string &Msg = std::string()) const; 168 void Diag(SourceLocation Loc, unsigned DiagID, 169 const std::string &Msg = std::string()) const; 170 171 /// getSourceLocation - Return a source location identifier for the specified 172 /// offset in the current file. 173 SourceLocation getSourceLocation(const char *Loc) const; 174 175 /// Stringify - Convert the specified string into a C string by escaping '\' 176 /// and " characters. This does not add surrounding ""'s to the string. 177 /// If Charify is true, this escapes the ' character instead of ". 178 static std::string Stringify(const std::string &Str, bool Charify = false); 179 180 //===--------------------------------------------------------------------===// 181 // Internal implementation interfaces. 182private: 183 184 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 185 /// by Lex. 186 /// 187 void LexTokenInternal(Token &Result); 188 189 /// FormTokenWithChars - When we lex a token, we have identified a span 190 /// starting at BufferPtr, going to TokEnd that forms the token. This method 191 /// takes that range and assigns it to the token as its location and size. In 192 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 193 /// TokEnd. 194 void FormTokenWithChars(Token &Result, const char *TokEnd) { 195 Result.setLocation(getSourceLocation(BufferPtr)); 196 Result.setLength(TokEnd-BufferPtr); 197 BufferPtr = TokEnd; 198 } 199 200 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 201 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 202 /// tokens in the buffer controlled by this lexer. 203 unsigned isNextPPTokenLParen(); 204 205 //===--------------------------------------------------------------------===// 206 // Lexer character reading interfaces. 207public: 208 209 // This lexer is built on two interfaces for reading characters, both of which 210 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 211 // when we know that we will be reading a character from the input buffer and 212 // that this character will be part of the result token. This occurs in (f.e.) 213 // string processing, because we know we need to read until we find the 214 // closing '"' character. 215 // 216 // The second interface is the combination of PeekCharAndSize with 217 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 218 // returning it and its size. If the lexer decides that this character is 219 // part of the current token, it calls ConsumeChar on it. This two stage 220 // approach allows us to emit diagnostics for characters (e.g. warnings about 221 // trigraphs), knowing that they only are emitted if the character is 222 // consumed. 223 224 /// isObviouslySimpleCharacter - Return true if the specified character is 225 /// obviously the same in translation phase 1 and translation phase 3. This 226 /// can return false for characters that end up being the same, but it will 227 /// never return true for something that needs to be mapped. 228 static bool isObviouslySimpleCharacter(char C) { 229 return C != '?' && C != '\\'; 230 } 231 232 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 233 /// advance over it, and return it. This is tricky in several cases. Here we 234 /// just handle the trivial case and fall-back to the non-inlined 235 /// getCharAndSizeSlow method to handle the hard case. 236 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 237 // If this is not a trigraph and not a UCN or escaped newline, return 238 // quickly. 239 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 240 241 unsigned Size = 0; 242 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 243 Ptr += Size; 244 return C; 245 } 246 247private: 248 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 249 /// and added to a given token, check to see if there are diagnostics that 250 /// need to be emitted or flags that need to be set on the token. If so, do 251 /// it. 252 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 253 // Normal case, we consumed exactly one token. Just return it. 254 if (Size == 1) 255 return Ptr+Size; 256 257 // Otherwise, re-lex the character with a current token, allowing 258 // diagnostics to be emitted and flags to be set. 259 Size = 0; 260 getCharAndSizeSlow(Ptr, Size, &Tok); 261 return Ptr+Size; 262 } 263 264 /// getCharAndSize - Peek a single 'character' from the specified buffer, 265 /// get its size, and return it. This is tricky in several cases. Here we 266 /// just handle the trivial case and fall-back to the non-inlined 267 /// getCharAndSizeSlow method to handle the hard case. 268 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 269 // If this is not a trigraph and not a UCN or escaped newline, return 270 // quickly. 271 if (isObviouslySimpleCharacter(Ptr[0])) { 272 Size = 1; 273 return *Ptr; 274 } 275 276 Size = 0; 277 return getCharAndSizeSlow(Ptr, Size); 278 } 279 280 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 281 /// method. 282 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 283 284 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 285 /// emit a warning. 286 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 287 const LangOptions &Features) { 288 // If this is not a trigraph and not a UCN or escaped newline, return 289 // quickly. 290 if (isObviouslySimpleCharacter(Ptr[0])) { 291 Size = 1; 292 return *Ptr; 293 } 294 295 Size = 0; 296 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 297 } 298 299 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 300 /// diagnostic. 301 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 302 const LangOptions &Features); 303 304 //===--------------------------------------------------------------------===// 305 // #if directive handling. 306 307 /// pushConditionalLevel - When we enter a #if directive, this keeps track of 308 /// what we are currently in for diagnostic emission (e.g. #if with missing 309 /// #endif). 310 void pushConditionalLevel(SourceLocation DirectiveStart, bool WasSkipping, 311 bool FoundNonSkip, bool FoundElse) { 312 PPConditionalInfo CI; 313 CI.IfLoc = DirectiveStart; 314 CI.WasSkipping = WasSkipping; 315 CI.FoundNonSkip = FoundNonSkip; 316 CI.FoundElse = FoundElse; 317 ConditionalStack.push_back(CI); 318 } 319 void pushConditionalLevel(const PPConditionalInfo &CI) { 320 ConditionalStack.push_back(CI); 321 } 322 323 /// popConditionalLevel - Remove an entry off the top of the conditional 324 /// stack, returning information about it. If the conditional stack is empty, 325 /// this returns true and does not fill in the arguments. 326 bool popConditionalLevel(PPConditionalInfo &CI) { 327 if (ConditionalStack.empty()) return true; 328 CI = ConditionalStack.back(); 329 ConditionalStack.pop_back(); 330 return false; 331 } 332 333 /// peekConditionalLevel - Return the top of the conditional stack. This 334 /// requires that there be a conditional active. 335 PPConditionalInfo &peekConditionalLevel() { 336 assert(!ConditionalStack.empty() && "No conditionals active!"); 337 return ConditionalStack.back(); 338 } 339 340 unsigned getConditionalStackDepth() const { return ConditionalStack.size(); } 341 342 //===--------------------------------------------------------------------===// 343 // Other lexer functions. 344 345 // Helper functions to lex the remainder of a token of the specific type. 346 void LexIdentifier (Token &Result, const char *CurPtr); 347 void LexNumericConstant (Token &Result, const char *CurPtr); 348 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 349 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 350 void LexCharConstant (Token &Result, const char *CurPtr); 351 bool LexEndOfFile (Token &Result, const char *CurPtr); 352 353 void SkipWhitespace (Token &Result, const char *CurPtr); 354 bool SkipBCPLComment (Token &Result, const char *CurPtr); 355 bool SkipBlockComment (Token &Result, const char *CurPtr); 356 bool SaveBCPLComment (Token &Result, const char *CurPtr); 357 358 /// LexIncludeFilename - After the preprocessor has parsed a #include, lex and 359 /// (potentially) macro expand the filename. If the sequence parsed is not 360 /// lexically legal, emit a diagnostic and return a result EOM token. 361 void LexIncludeFilename(Token &Result); 362}; 363 364 365} // end namespace clang 366 367#endif 368