Lexer.h revision 42e00d19b0dac64732eb5449d52a076282fcbf77
1//===--- Lexer.h - C Language Family Lexer ----------------------*- 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 Lexer interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_LEXER_H 15#define LLVM_CLANG_LEXER_H 16 17#include "clang/Lex/PreprocessorLexer.h" 18#include "clang/Basic/LangOptions.h" 19#include "llvm/ADT/SmallVector.h" 20#include <string> 21#include <vector> 22#include <cassert> 23 24namespace clang { 25class Diagnostic; 26class SourceManager; 27class Preprocessor; 28class DiagnosticBuilder; 29 30/// Lexer - This provides a simple interface that turns a text buffer into a 31/// stream of tokens. This provides no support for file reading or buffering, 32/// or buffering/seeking of tokens, only forward lexing is supported. It relies 33/// on the specified Preprocessor object to handle preprocessor directives, etc. 34class Lexer : public PreprocessorLexer { 35 //===--------------------------------------------------------------------===// 36 // Constant configuration values for this lexer. 37 const char *BufferStart; // Start of the buffer. 38 const char *BufferEnd; // End of the buffer. 39 SourceLocation FileLoc; // Location for start of file. 40 LangOptions Features; // Features enabled by this language (cache). 41 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 42 43 //===--------------------------------------------------------------------===// 44 // Context-specific lexing flags set by the preprocessor. 45 // 46 47 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace 48 /// and return them as tokens. This is used for -C and -CC modes, and 49 /// whitespace preservation can be useful for some clients that want to lex 50 /// the file in raw mode and get every character from the file. 51 /// 52 /// When this is set to 2 it returns comments and whitespace. When set to 1 53 /// it returns comments, when it is set to 0 it returns normal tokens only. 54 unsigned char ExtendedTokenMode; 55 56 //===--------------------------------------------------------------------===// 57 // Context that changes as the file is lexed. 58 // NOTE: any state that mutates when in raw mode must have save/restore code 59 // in Lexer::isNextPPTokenLParen. 60 61 // BufferPtr - Current pointer into the buffer. This is the next character 62 // to be lexed. 63 const char *BufferPtr; 64 65 // IsAtStartOfLine - True if the next lexed token should get the "start of 66 // line" flag set on it. 67 bool IsAtStartOfLine; 68 69 Lexer(const Lexer&); // DO NOT IMPLEMENT 70 void operator=(const Lexer&); // DO NOT IMPLEMENT 71 friend class Preprocessor; 72 73 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd); 74public: 75 76 /// Lexer constructor - Create a new lexer object for the specified buffer 77 /// with the specified preprocessor managing the lexing process. This lexer 78 /// assumes that the associated file buffer and Preprocessor objects will 79 /// outlive it, so it doesn't take ownership of either of them. 80 Lexer(FileID FID, Preprocessor &PP); 81 82 /// Lexer constructor - Create a new raw lexer object. This object is only 83 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 84 /// range will outlive it, so it doesn't take ownership of it. 85 Lexer(SourceLocation FileLoc, const LangOptions &Features, 86 const char *BufStart, const char *BufPtr, const char *BufEnd); 87 88 /// Lexer constructor - Create a new raw lexer object. This object is only 89 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 90 /// range will outlive it, so it doesn't take ownership of it. 91 Lexer(FileID FID, const SourceManager &SM, const LangOptions &Features); 92 93 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for 94 /// _Pragma expansion. This has a variety of magic semantics that this method 95 /// sets up. It returns a new'd Lexer that must be delete'd when done. 96 static Lexer *Create_PragmaLexer(SourceLocation TokStartLoc, unsigned TokLen, 97 Preprocessor &PP); 98 99 100 /// getFeatures - Return the language features currently enabled. NOTE: this 101 /// lexer modifies features as a file is parsed! 102 const LangOptions &getFeatures() const { return Features; } 103 104 /// getFileLoc - Return the File Location for the file we are lexing out of. 105 /// The physical location encodes the location where the characters come from, 106 /// the virtual location encodes where we should *claim* the characters came 107 /// from. Currently this is only used by _Pragma handling. 108 SourceLocation getFileLoc() const { return FileLoc; } 109 110 /// Lex - Return the next token in the file. If this is the end of file, it 111 /// return the tok::eof token. Return true if an error occurred and 112 /// compilation should terminate, false if normal. This implicitly involves 113 /// the preprocessor. 114 void Lex(Token &Result) { 115 // Start a new token. 116 Result.startToken(); 117 118 // NOTE, any changes here should also change code after calls to 119 // Preprocessor::HandleDirective 120 if (IsAtStartOfLine) { 121 Result.setFlag(Token::StartOfLine); 122 IsAtStartOfLine = false; 123 } 124 125 // Get a token. Note that this may delete the current lexer if the end of 126 // file is reached. 127 LexTokenInternal(Result); 128 } 129 130 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 131 bool isPragmaLexer() const { return Is_PragmaLexer; } 132 133 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 134 /// the PreprocessorLexer interface. 135 void IndirectLex(Token &Result) { Lex(Result); } 136 137 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 138 /// associated preprocessor object. Return true if the 'next character to 139 /// read' pointer points and the end of the lexer buffer, false otherwise. 140 bool LexFromRawLexer(Token &Result) { 141 assert(LexingRawMode && "Not already in raw mode!"); 142 Lex(Result); 143 // Note that lexing to the end of the buffer doesn't implicitly delete the 144 // lexer when in raw mode. 145 return BufferPtr == BufferEnd; 146 } 147 148 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 149 /// every character in the file, including whitespace and comments. This 150 /// should only be used in raw mode, as the preprocessor is not prepared to 151 /// deal with the excess tokens. 152 bool isKeepWhitespaceMode() const { 153 return ExtendedTokenMode > 1; 154 } 155 156 /// SetKeepWhitespaceMode - This method lets clients enable or disable 157 /// whitespace retention mode. 158 void SetKeepWhitespaceMode(bool Val) { 159 assert((!Val || LexingRawMode) && 160 "Can only enable whitespace retention in raw mode"); 161 ExtendedTokenMode = Val ? 2 : 0; 162 } 163 164 /// inKeepCommentMode - Return true if the lexer should return comments as 165 /// tokens. 166 bool inKeepCommentMode() const { 167 return ExtendedTokenMode > 0; 168 } 169 170 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 171 /// to the specified mode. This is really only useful when lexing in raw 172 /// mode, because otherwise the lexer needs to manage this. 173 void SetCommentRetentionState(bool Mode) { 174 assert(!isKeepWhitespaceMode() && 175 "Can't play with comment retention state when retaining whitespace"); 176 ExtendedTokenMode = Mode ? 1 : 0; 177 } 178 179 const char *getBufferStart() const { return BufferStart; } 180 181 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 182 /// uninterpreted string. This switches the lexer out of directive mode. 183 std::string ReadToEndOfLine(); 184 185 186 /// Diag - Forwarding function for diagnostics. This translate a source 187 /// position in the current buffer into a SourceLocation object for rendering. 188 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const; 189 190 /// getSourceLocation - Return a source location identifier for the specified 191 /// offset in the current file. 192 SourceLocation getSourceLocation(const char *Loc) const; 193 194 /// getSourceLocation - Return a source location for the next character in 195 /// the current file. 196 SourceLocation getSourceLocation() { return getSourceLocation(BufferPtr); } 197 198 /// Stringify - Convert the specified string into a C string by escaping '\' 199 /// and " characters. This does not add surrounding ""'s to the string. 200 /// If Charify is true, this escapes the ' character instead of ". 201 static std::string Stringify(const std::string &Str, bool Charify = false); 202 203 /// Stringify - Convert the specified string into a C string by escaping '\' 204 /// and " characters. This does not add surrounding ""'s to the string. 205 static void Stringify(llvm::SmallVectorImpl<char> &Str); 206 207 /// MeasureTokenLength - Relex the token at the specified location and return 208 /// its length in bytes in the input file. If the token needs cleaning (e.g. 209 /// includes a trigraph or an escaped newline) then this count includes bytes 210 /// that are part of that. 211 static unsigned MeasureTokenLength(SourceLocation Loc, 212 const SourceManager &SM); 213 214 //===--------------------------------------------------------------------===// 215 // Internal implementation interfaces. 216private: 217 218 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 219 /// by Lex. 220 /// 221 void LexTokenInternal(Token &Result); 222 223 /// FormTokenWithChars - When we lex a token, we have identified a span 224 /// starting at BufferPtr, going to TokEnd that forms the token. This method 225 /// takes that range and assigns it to the token as its location and size. In 226 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 227 /// TokEnd. 228 void FormTokenWithChars(Token &Result, const char *TokEnd, 229 tok::TokenKind Kind) { 230 Result.setLocation(getSourceLocation(BufferPtr)); 231 Result.setLength(TokEnd-BufferPtr); 232 Result.setKind(Kind); 233 BufferPtr = TokEnd; 234 } 235 236 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 237 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 238 /// tokens in the buffer controlled by this lexer. 239 unsigned isNextPPTokenLParen(); 240 241 //===--------------------------------------------------------------------===// 242 // Lexer character reading interfaces. 243public: 244 245 // This lexer is built on two interfaces for reading characters, both of which 246 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 247 // when we know that we will be reading a character from the input buffer and 248 // that this character will be part of the result token. This occurs in (f.e.) 249 // string processing, because we know we need to read until we find the 250 // closing '"' character. 251 // 252 // The second interface is the combination of PeekCharAndSize with 253 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 254 // returning it and its size. If the lexer decides that this character is 255 // part of the current token, it calls ConsumeChar on it. This two stage 256 // approach allows us to emit diagnostics for characters (e.g. warnings about 257 // trigraphs), knowing that they only are emitted if the character is 258 // consumed. 259 260 /// isObviouslySimpleCharacter - Return true if the specified character is 261 /// obviously the same in translation phase 1 and translation phase 3. This 262 /// can return false for characters that end up being the same, but it will 263 /// never return true for something that needs to be mapped. 264 static bool isObviouslySimpleCharacter(char C) { 265 return C != '?' && C != '\\'; 266 } 267 268 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 269 /// advance over it, and return it. This is tricky in several cases. Here we 270 /// just handle the trivial case and fall-back to the non-inlined 271 /// getCharAndSizeSlow method to handle the hard case. 272 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 273 // If this is not a trigraph and not a UCN or escaped newline, return 274 // quickly. 275 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 276 277 unsigned Size = 0; 278 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 279 Ptr += Size; 280 return C; 281 } 282 283private: 284 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 285 /// and added to a given token, check to see if there are diagnostics that 286 /// need to be emitted or flags that need to be set on the token. If so, do 287 /// it. 288 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 289 // Normal case, we consumed exactly one token. Just return it. 290 if (Size == 1) 291 return Ptr+Size; 292 293 // Otherwise, re-lex the character with a current token, allowing 294 // diagnostics to be emitted and flags to be set. 295 Size = 0; 296 getCharAndSizeSlow(Ptr, Size, &Tok); 297 return Ptr+Size; 298 } 299 300 /// getCharAndSize - Peek a single 'character' from the specified buffer, 301 /// get its size, and return it. This is tricky in several cases. Here we 302 /// just handle the trivial case and fall-back to the non-inlined 303 /// getCharAndSizeSlow method to handle the hard case. 304 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 305 // If this is not a trigraph and not a UCN or escaped newline, return 306 // quickly. 307 if (isObviouslySimpleCharacter(Ptr[0])) { 308 Size = 1; 309 return *Ptr; 310 } 311 312 Size = 0; 313 return getCharAndSizeSlow(Ptr, Size); 314 } 315 316 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 317 /// method. 318 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 319public: 320 321 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 322 /// emit a warning. 323 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 324 const LangOptions &Features) { 325 // If this is not a trigraph and not a UCN or escaped newline, return 326 // quickly. 327 if (isObviouslySimpleCharacter(Ptr[0])) { 328 Size = 1; 329 return *Ptr; 330 } 331 332 Size = 0; 333 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 334 } 335private: 336 337 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 338 /// diagnostic. 339 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 340 const LangOptions &Features); 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 bool 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}; 358 359 360} // end namespace clang 361 362#endif 363