Lexer.h revision d6a2e7d71e9a910c45a3eb0268a5d94635003d58
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; 28 29/// Lexer - This provides a simple interface that turns a text buffer into a 30/// stream of tokens. This provides no support for file reading or buffering, 31/// or buffering/seeking of tokens, only forward lexing is supported. It relies 32/// on the specified Preprocessor object to handle preprocessor directives, etc. 33class Lexer : public PreprocessorLexer { 34 //===--------------------------------------------------------------------===// 35 // Constant configuration values for this lexer. 36 const char *BufferStart; // Start of the buffer. 37 const char *BufferEnd; // End of the buffer. 38 SourceLocation FileLoc; // Location for start of file. 39 LangOptions Features; // Features enabled by this language (cache). 40 bool Is_PragmaLexer; // True if lexer for _Pragma handling. 41 42 //===--------------------------------------------------------------------===// 43 // Context-specific lexing flags set by the preprocessor. 44 // 45 46 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace 47 /// and return them as tokens. This is used for -C and -CC modes, and 48 /// whitespace preservation can be useful for some clients that want to lex 49 /// the file in raw mode and get every character from the file. 50 /// 51 /// When this is set to 2 it returns comments and whitespace. When set to 1 52 /// it returns comments, when it is set to 0 it returns normal tokens only. 53 unsigned char ExtendedTokenMode; 54 55 //===--------------------------------------------------------------------===// 56 // Context that changes as the file is lexed. 57 // NOTE: any state that mutates when in raw mode must have save/restore code 58 // in Lexer::isNextPPTokenLParen. 59 60 // BufferPtr - Current pointer into the buffer. This is the next character 61 // to be lexed. 62 const char *BufferPtr; 63 64 // IsAtStartOfLine - True if the next lexed token should get the "start of 65 // line" flag set on it. 66 bool IsAtStartOfLine; 67 68 Lexer(const Lexer&); // DO NOT IMPLEMENT 69 void operator=(const Lexer&); // DO NOT IMPLEMENT 70 friend class Preprocessor; 71public: 72 73 /// Lexer constructor - Create a new lexer object for the specified buffer 74 /// with the specified preprocessor managing the lexing process. This lexer 75 /// assumes that the associated file buffer and Preprocessor objects will 76 /// outlive it, so it doesn't take ownership of either of them. 77 Lexer(SourceLocation FileLoc, Preprocessor &PP, 78 const char *BufStart = 0, const char *BufEnd = 0); 79 80 /// Lexer constructor - Create a new raw lexer object. This object is only 81 /// suitable for calls to 'LexRawToken'. This lexer assumes that the text 82 /// range will outlive it, so it doesn't take ownership of it. 83 Lexer(SourceLocation FileLoc, const LangOptions &Features, 84 const char *BufStart, const char *BufEnd, 85 const llvm::MemoryBuffer *FromFile = 0); 86 87 /// getFeatures - Return the language features currently enabled. NOTE: this 88 /// lexer modifies features as a file is parsed! 89 const LangOptions &getFeatures() const { return Features; } 90 91 /// getFileLoc - Return the File Location for the file we are lexing out of. 92 /// The physical location encodes the location where the characters come from, 93 /// the virtual location encodes where we should *claim* the characters came 94 /// from. Currently this is only used by _Pragma handling. 95 SourceLocation getFileLoc() const { return FileLoc; } 96 97 /// Lex - Return the next token in the file. If this is the end of file, it 98 /// return the tok::eof token. Return true if an error occurred and 99 /// compilation should terminate, false if normal. This implicitly involves 100 /// the preprocessor. 101 void Lex(Token &Result) { 102 // Start a new token. 103 Result.startToken(); 104 105 // NOTE, any changes here should also change code after calls to 106 // Preprocessor::HandleDirective 107 if (IsAtStartOfLine) { 108 Result.setFlag(Token::StartOfLine); 109 IsAtStartOfLine = false; 110 } 111 112 // Get a token. Note that this may delete the current lexer if the end of 113 // file is reached. 114 LexTokenInternal(Result); 115 } 116 117 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 118 /// the PreprocessorLexer interface. 119 void IndirectLex(Token &Result) { Lex(Result); } 120 121 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 122 /// associated preprocessor object. Return true if the 'next character to 123 /// read' pointer points and the end of the lexer buffer, false otherwise. 124 bool LexFromRawLexer(Token &Result) { 125 assert(LexingRawMode && "Not already in raw mode!"); 126 Lex(Result); 127 // Note that lexing to the end of the buffer doesn't implicitly delete the 128 // lexer when in raw mode. 129 return BufferPtr == BufferEnd; 130 } 131 132 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 133 /// every character in the file, including whitespace and comments. This 134 /// should only be used in raw mode, as the preprocessor is not prepared to 135 /// deal with the excess tokens. 136 bool isKeepWhitespaceMode() const { 137 return ExtendedTokenMode > 1; 138 } 139 140 /// SetKeepWhitespaceMode - This method lets clients enable or disable 141 /// whitespace retention mode. 142 void SetKeepWhitespaceMode(bool Val) { 143 assert((!Val || LexingRawMode) && 144 "Can only enable whitespace retention in raw mode"); 145 ExtendedTokenMode = Val ? 2 : 0; 146 } 147 148 /// inKeepCommentMode - Return true if the lexer should return comments as 149 /// tokens. 150 bool inKeepCommentMode() const { 151 return ExtendedTokenMode > 0; 152 } 153 154 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 155 /// to the specified mode. This is really only useful when lexing in raw 156 /// mode, because otherwise the lexer needs to manage this. 157 void SetCommentRetentionState(bool Mode) { 158 assert(!isKeepWhitespaceMode() && 159 "Can't play with comment retention state when retaining whitespace"); 160 ExtendedTokenMode = Mode ? 1 : 0; 161 } 162 163 164 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 165 /// uninterpreted string. This switches the lexer out of directive mode. 166 std::string ReadToEndOfLine(); 167 168 169 /// Diag - Forwarding function for diagnostics. This translate a source 170 /// position in the current buffer into a SourceLocation object for rendering. 171 void Diag(const char *Loc, unsigned DiagID, 172 const std::string &Msg = std::string()) const; 173 174 /// getSourceLocation - Return a source location identifier for the specified 175 /// offset in the current file. 176 SourceLocation getSourceLocation(const char *Loc) const; 177 178 /// Stringify - Convert the specified string into a C string by escaping '\' 179 /// and " characters. This does not add surrounding ""'s to the string. 180 /// If Charify is true, this escapes the ' character instead of ". 181 static std::string Stringify(const std::string &Str, bool Charify = false); 182 183 /// Stringify - Convert the specified string into a C string by escaping '\' 184 /// and " characters. This does not add surrounding ""'s to the string. 185 static void Stringify(llvm::SmallVectorImpl<char> &Str); 186 187 /// MeasureTokenLength - Relex the token at the specified location and return 188 /// its length in bytes in the input file. If the token needs cleaning (e.g. 189 /// includes a trigraph or an escaped newline) then this count includes bytes 190 /// that are part of that. 191 static unsigned MeasureTokenLength(SourceLocation Loc, 192 const SourceManager &SM); 193 194 //===--------------------------------------------------------------------===// 195 // Internal implementation interfaces. 196private: 197 198 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 199 /// by Lex. 200 /// 201 void LexTokenInternal(Token &Result); 202 203 /// FormTokenWithChars - When we lex a token, we have identified a span 204 /// starting at BufferPtr, going to TokEnd that forms the token. This method 205 /// takes that range and assigns it to the token as its location and size. In 206 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 207 /// TokEnd. 208 void FormTokenWithChars(Token &Result, const char *TokEnd, 209 tok::TokenKind Kind) { 210 Result.setLocation(getSourceLocation(BufferPtr)); 211 Result.setLength(TokEnd-BufferPtr); 212 Result.setKind(Kind); 213 BufferPtr = TokEnd; 214 } 215 216 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 217 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 218 /// tokens in the buffer controlled by this lexer. 219 unsigned isNextPPTokenLParen(); 220 221 //===--------------------------------------------------------------------===// 222 // Lexer character reading interfaces. 223public: 224 225 // This lexer is built on two interfaces for reading characters, both of which 226 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 227 // when we know that we will be reading a character from the input buffer and 228 // that this character will be part of the result token. This occurs in (f.e.) 229 // string processing, because we know we need to read until we find the 230 // closing '"' character. 231 // 232 // The second interface is the combination of PeekCharAndSize with 233 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 234 // returning it and its size. If the lexer decides that this character is 235 // part of the current token, it calls ConsumeChar on it. This two stage 236 // approach allows us to emit diagnostics for characters (e.g. warnings about 237 // trigraphs), knowing that they only are emitted if the character is 238 // consumed. 239 240 /// isObviouslySimpleCharacter - Return true if the specified character is 241 /// obviously the same in translation phase 1 and translation phase 3. This 242 /// can return false for characters that end up being the same, but it will 243 /// never return true for something that needs to be mapped. 244 static bool isObviouslySimpleCharacter(char C) { 245 return C != '?' && C != '\\'; 246 } 247 248 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 249 /// advance over it, and return it. This is tricky in several cases. Here we 250 /// just handle the trivial case and fall-back to the non-inlined 251 /// getCharAndSizeSlow method to handle the hard case. 252 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 253 // If this is not a trigraph and not a UCN or escaped newline, return 254 // quickly. 255 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 256 257 unsigned Size = 0; 258 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 259 Ptr += Size; 260 return C; 261 } 262 263private: 264 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 265 /// and added to a given token, check to see if there are diagnostics that 266 /// need to be emitted or flags that need to be set on the token. If so, do 267 /// it. 268 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 269 // Normal case, we consumed exactly one token. Just return it. 270 if (Size == 1) 271 return Ptr+Size; 272 273 // Otherwise, re-lex the character with a current token, allowing 274 // diagnostics to be emitted and flags to be set. 275 Size = 0; 276 getCharAndSizeSlow(Ptr, Size, &Tok); 277 return Ptr+Size; 278 } 279 280 /// getCharAndSize - Peek a single 'character' from the specified buffer, 281 /// get its size, and return it. This is tricky in several cases. Here we 282 /// just handle the trivial case and fall-back to the non-inlined 283 /// getCharAndSizeSlow method to handle the hard case. 284 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 285 // If this is not a trigraph and not a UCN or escaped newline, return 286 // quickly. 287 if (isObviouslySimpleCharacter(Ptr[0])) { 288 Size = 1; 289 return *Ptr; 290 } 291 292 Size = 0; 293 return getCharAndSizeSlow(Ptr, Size); 294 } 295 296 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 297 /// method. 298 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 299 300 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 301 /// emit a warning. 302 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 303 const LangOptions &Features) { 304 // If this is not a trigraph and not a UCN or escaped newline, return 305 // quickly. 306 if (isObviouslySimpleCharacter(Ptr[0])) { 307 Size = 1; 308 return *Ptr; 309 } 310 311 Size = 0; 312 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 313 } 314 315 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 316 /// diagnostic. 317 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 318 const LangOptions &Features); 319 320 //===--------------------------------------------------------------------===// 321 // Other lexer functions. 322 323 // Helper functions to lex the remainder of a token of the specific type. 324 void LexIdentifier (Token &Result, const char *CurPtr); 325 void LexNumericConstant (Token &Result, const char *CurPtr); 326 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 327 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 328 void LexCharConstant (Token &Result, const char *CurPtr); 329 bool LexEndOfFile (Token &Result, const char *CurPtr); 330 331 bool SkipWhitespace (Token &Result, const char *CurPtr); 332 bool SkipBCPLComment (Token &Result, const char *CurPtr); 333 bool SkipBlockComment (Token &Result, const char *CurPtr); 334 bool SaveBCPLComment (Token &Result, const char *CurPtr); 335}; 336 337 338} // end namespace clang 339 340#endif 341