Lexer.h revision 4b391087f9c5582d448ab66ccbc7028f7673f380
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 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma. 118 bool isPragmaLexer() const { return Is_PragmaLexer; } 119 120 /// IndirectLex - An indirect call to 'Lex' that can be invoked via 121 /// the PreprocessorLexer interface. 122 void IndirectLex(Token &Result) { Lex(Result); } 123 124 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no 125 /// associated preprocessor object. Return true if the 'next character to 126 /// read' pointer points and the end of the lexer buffer, false otherwise. 127 bool LexFromRawLexer(Token &Result) { 128 assert(LexingRawMode && "Not already in raw mode!"); 129 Lex(Result); 130 // Note that lexing to the end of the buffer doesn't implicitly delete the 131 // lexer when in raw mode. 132 return BufferPtr == BufferEnd; 133 } 134 135 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for 136 /// every character in the file, including whitespace and comments. This 137 /// should only be used in raw mode, as the preprocessor is not prepared to 138 /// deal with the excess tokens. 139 bool isKeepWhitespaceMode() const { 140 return ExtendedTokenMode > 1; 141 } 142 143 /// SetKeepWhitespaceMode - This method lets clients enable or disable 144 /// whitespace retention mode. 145 void SetKeepWhitespaceMode(bool Val) { 146 assert((!Val || LexingRawMode) && 147 "Can only enable whitespace retention in raw mode"); 148 ExtendedTokenMode = Val ? 2 : 0; 149 } 150 151 /// inKeepCommentMode - Return true if the lexer should return comments as 152 /// tokens. 153 bool inKeepCommentMode() const { 154 return ExtendedTokenMode > 0; 155 } 156 157 /// SetCommentRetentionMode - Change the comment retention mode of the lexer 158 /// to the specified mode. This is really only useful when lexing in raw 159 /// mode, because otherwise the lexer needs to manage this. 160 void SetCommentRetentionState(bool Mode) { 161 assert(!isKeepWhitespaceMode() && 162 "Can't play with comment retention state when retaining whitespace"); 163 ExtendedTokenMode = Mode ? 1 : 0; 164 } 165 166 167 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 168 /// uninterpreted string. This switches the lexer out of directive mode. 169 std::string ReadToEndOfLine(); 170 171 172 /// Diag - Forwarding function for diagnostics. This translate a source 173 /// position in the current buffer into a SourceLocation object for rendering. 174 void Diag(const char *Loc, unsigned DiagID, 175 const std::string &Msg = std::string()) const; 176 177 /// getSourceLocation - Return a source location identifier for the specified 178 /// offset in the current file. 179 SourceLocation getSourceLocation(const char *Loc) const; 180 181 /// Stringify - Convert the specified string into a C string by escaping '\' 182 /// and " characters. This does not add surrounding ""'s to the string. 183 /// If Charify is true, this escapes the ' character instead of ". 184 static std::string Stringify(const std::string &Str, bool Charify = false); 185 186 /// Stringify - Convert the specified string into a C string by escaping '\' 187 /// and " characters. This does not add surrounding ""'s to the string. 188 static void Stringify(llvm::SmallVectorImpl<char> &Str); 189 190 /// MeasureTokenLength - Relex the token at the specified location and return 191 /// its length in bytes in the input file. If the token needs cleaning (e.g. 192 /// includes a trigraph or an escaped newline) then this count includes bytes 193 /// that are part of that. 194 static unsigned MeasureTokenLength(SourceLocation Loc, 195 const SourceManager &SM); 196 197 //===--------------------------------------------------------------------===// 198 // Internal implementation interfaces. 199private: 200 201 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 202 /// by Lex. 203 /// 204 void LexTokenInternal(Token &Result); 205 206 /// FormTokenWithChars - When we lex a token, we have identified a span 207 /// starting at BufferPtr, going to TokEnd that forms the token. This method 208 /// takes that range and assigns it to the token as its location and size. In 209 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 210 /// TokEnd. 211 void FormTokenWithChars(Token &Result, const char *TokEnd, 212 tok::TokenKind Kind) { 213 Result.setLocation(getSourceLocation(BufferPtr)); 214 Result.setLength(TokEnd-BufferPtr); 215 Result.setKind(Kind); 216 BufferPtr = TokEnd; 217 } 218 219 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 220 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 221 /// tokens in the buffer controlled by this lexer. 222 unsigned isNextPPTokenLParen(); 223 224 //===--------------------------------------------------------------------===// 225 // Lexer character reading interfaces. 226public: 227 228 // This lexer is built on two interfaces for reading characters, both of which 229 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 230 // when we know that we will be reading a character from the input buffer and 231 // that this character will be part of the result token. This occurs in (f.e.) 232 // string processing, because we know we need to read until we find the 233 // closing '"' character. 234 // 235 // The second interface is the combination of PeekCharAndSize with 236 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 237 // returning it and its size. If the lexer decides that this character is 238 // part of the current token, it calls ConsumeChar on it. This two stage 239 // approach allows us to emit diagnostics for characters (e.g. warnings about 240 // trigraphs), knowing that they only are emitted if the character is 241 // consumed. 242 243 /// isObviouslySimpleCharacter - Return true if the specified character is 244 /// obviously the same in translation phase 1 and translation phase 3. This 245 /// can return false for characters that end up being the same, but it will 246 /// never return true for something that needs to be mapped. 247 static bool isObviouslySimpleCharacter(char C) { 248 return C != '?' && C != '\\'; 249 } 250 251 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 252 /// advance over it, and return it. This is tricky in several cases. Here we 253 /// just handle the trivial case and fall-back to the non-inlined 254 /// getCharAndSizeSlow method to handle the hard case. 255 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 256 // If this is not a trigraph and not a UCN or escaped newline, return 257 // quickly. 258 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 259 260 unsigned Size = 0; 261 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 262 Ptr += Size; 263 return C; 264 } 265 266private: 267 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 268 /// and added to a given token, check to see if there are diagnostics that 269 /// need to be emitted or flags that need to be set on the token. If so, do 270 /// it. 271 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 272 // Normal case, we consumed exactly one token. Just return it. 273 if (Size == 1) 274 return Ptr+Size; 275 276 // Otherwise, re-lex the character with a current token, allowing 277 // diagnostics to be emitted and flags to be set. 278 Size = 0; 279 getCharAndSizeSlow(Ptr, Size, &Tok); 280 return Ptr+Size; 281 } 282 283 /// getCharAndSize - Peek a single 'character' from the specified buffer, 284 /// get its size, and return it. This is tricky in several cases. Here we 285 /// just handle the trivial case and fall-back to the non-inlined 286 /// getCharAndSizeSlow method to handle the hard case. 287 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 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 getCharAndSizeSlow(Ptr, Size); 297 } 298 299 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 300 /// method. 301 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 302 303 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 304 /// emit a warning. 305 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 306 const LangOptions &Features) { 307 // If this is not a trigraph and not a UCN or escaped newline, return 308 // quickly. 309 if (isObviouslySimpleCharacter(Ptr[0])) { 310 Size = 1; 311 return *Ptr; 312 } 313 314 Size = 0; 315 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 316 } 317 318 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 319 /// diagnostic. 320 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 321 const LangOptions &Features); 322 323 //===--------------------------------------------------------------------===// 324 // Other lexer functions. 325 326 // Helper functions to lex the remainder of a token of the specific type. 327 void LexIdentifier (Token &Result, const char *CurPtr); 328 void LexNumericConstant (Token &Result, const char *CurPtr); 329 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 330 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 331 void LexCharConstant (Token &Result, const char *CurPtr); 332 bool LexEndOfFile (Token &Result, const char *CurPtr); 333 334 bool SkipWhitespace (Token &Result, const char *CurPtr); 335 bool SkipBCPLComment (Token &Result, const char *CurPtr); 336 bool SkipBlockComment (Token &Result, const char *CurPtr); 337 bool SaveBCPLComment (Token &Result, const char *CurPtr); 338}; 339 340 341} // end namespace clang 342 343#endif 344