Lexer.h revision e5956bd2730c051835f9acd9e957c5d79f99e7c3
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 "llvm/ADT/SmallVector.h" 21#include <string> 22#include <vector> 23#include <cassert> 24 25namespace clang { 26class Diagnostic; 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 { 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 Preprocessor &PP; // Preprocessor object controlling lexing. 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 /// ParsingPreprocessorDirective - This is true when parsing #XXX. This turns 48 /// '\n' into a tok::eom token. 49 bool ParsingPreprocessorDirective; 50 51 /// ParsingFilename - True after #include: this turns <xx> into a 52 /// tok::angle_string_literal token. 53 bool ParsingFilename; 54 55 /// LexingRawMode - True if in raw mode: This flag disables interpretation of 56 /// tokens and is a far faster mode to lex in than non-raw-mode. This flag: 57 /// 1. If EOF of the current lexer is found, the include stack isn't popped. 58 /// 2. Identifier information is not looked up for identifier tokens. As an 59 /// effect of this, implicit macro expansion is naturally disabled. 60 /// 3. "#" tokens at the start of a line are treated as normal tokens, not 61 /// implicitly transformed by the lexer. 62 /// 4. All diagnostic messages are disabled, except for unterminated /*. 63 /// 5. The only callback made into the preprocessor is to report a hard error 64 /// on an unterminated '/*' comment. 65 bool LexingRawMode; 66 67 /// KeepCommentMode - The lexer can optionally keep C & BCPL-style comments, 68 /// and return them as tokens. This is used for -C and -CC modes. 69 bool KeepCommentMode; 70 71 //===--------------------------------------------------------------------===// 72 // Context that changes as the file is lexed. 73 // NOTE: any state that mutates when in raw mode must have save/restore code 74 // in Lexer::isNextPPTokenLParen. 75 76 // BufferPtr - Current pointer into the buffer. This is the next character 77 // to be lexed. 78 const char *BufferPtr; 79 80 // IsAtStartOfLine - True if the next lexed token should get the "start of 81 // line" flag set on it. 82 bool IsAtStartOfLine; 83 84 /// MIOpt - This is a state machine that detects the #ifndef-wrapping a file 85 /// idiom for the multiple-include optimization. 86 MultipleIncludeOpt MIOpt; 87 88 /// ConditionalStack - Information about the set of #if/#ifdef/#ifndef blocks 89 /// we are currently in. 90 std::vector<PPConditionalInfo> ConditionalStack; 91 92 Lexer(const Lexer&); // DO NOT IMPLEMENT 93 void operator=(const Lexer&); // DO NOT IMPLEMENT 94 friend class Preprocessor; 95public: 96 97 /// Lexer constructor - Create a new lexer object for the specified buffer 98 /// with the specified preprocessor managing the lexing process. This lexer 99 /// assumes that the associated MemoryBuffer and Preprocessor objects will 100 /// outlive it, so it doesn't take ownership of either of them. 101 Lexer(SourceLocation FileLoc, Preprocessor &PP, 102 const char *BufStart = 0, const char *BufEnd = 0); 103 104 /// getFeatures - Return the language features currently enabled. NOTE: this 105 /// lexer modifies features as a file is parsed! 106 const LangOptions &getFeatures() const { return Features; } 107 108 /// getFileLoc - Return the File Location for the file we are lexing out of. 109 /// The physical location encodes the location where the characters come from, 110 /// the virtual location encodes where we should *claim* the characters came 111 /// from. Currently this is only used by _Pragma handling. 112 SourceLocation getFileLoc() const { return FileLoc; } 113 114 /// Lex - Return the next token in the file. If this is the end of file, it 115 /// return the tok::eof token. Return true if an error occurred and 116 /// compilation should terminate, false if normal. This implicitly involves 117 /// the preprocessor. 118 void Lex(Token &Result) { 119 // Start a new token. 120 Result.startToken(); 121 122 // NOTE, any changes here should also change code after calls to 123 // Preprocessor::HandleDirective 124 if (IsAtStartOfLine) { 125 Result.setFlag(Token::StartOfLine); 126 IsAtStartOfLine = false; 127 } 128 129 // Get a token. Note that this may delete the current lexer if the end of 130 // file is reached. 131 LexTokenInternal(Result); 132 } 133 134 /// LexRawToken - Switch the lexer to raw mode, lex a token into Result and 135 /// switch it back. Return true if the 'next character to read' pointer 136 /// points and the end of the lexer buffer, false otherwise. 137 bool LexRawToken(Token &Result) { 138 assert(!LexingRawMode && "Already in raw mode!"); 139 LexingRawMode = true; 140 Lex(Result); 141 LexingRawMode = false; 142 // Note that lexing to the end of the buffer doesn't implicitly delete the 143 // lexer when in raw mode. 144 return BufferPtr == BufferEnd; 145 } 146 147 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an 148 /// uninterpreted string. This switches the lexer out of directive mode. 149 std::string ReadToEndOfLine(); 150 151 152 /// Diag - Forwarding function for diagnostics. This translate a source 153 /// position in the current buffer into a SourceLocation object for rendering. 154 void Diag(const char *Loc, unsigned DiagID, 155 const std::string &Msg = std::string()) const; 156 void Diag(SourceLocation Loc, unsigned DiagID, 157 const std::string &Msg = std::string()) const; 158 159 /// getSourceLocation - Return a source location identifier for the specified 160 /// offset in the current file. 161 SourceLocation getSourceLocation(const char *Loc) const; 162 163 /// Stringify - Convert the specified string into a C string by escaping '\' 164 /// and " characters. This does not add surrounding ""'s to the string. 165 /// If Charify is true, this escapes the ' character instead of ". 166 static std::string Stringify(const std::string &Str, bool Charify = false); 167 168 /// Stringify - Convert the specified string into a C string by escaping '\' 169 /// and " characters. This does not add surrounding ""'s to the string. 170 static void Stringify(llvm::SmallVectorImpl<char> &Str); 171 172 //===--------------------------------------------------------------------===// 173 // Internal implementation interfaces. 174private: 175 176 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called 177 /// by Lex. 178 /// 179 void LexTokenInternal(Token &Result); 180 181 /// FormTokenWithChars - When we lex a token, we have identified a span 182 /// starting at BufferPtr, going to TokEnd that forms the token. This method 183 /// takes that range and assigns it to the token as its location and size. In 184 /// addition, since tokens cannot overlap, this also updates BufferPtr to be 185 /// TokEnd. 186 void FormTokenWithChars(Token &Result, const char *TokEnd) { 187 Result.setLocation(getSourceLocation(BufferPtr)); 188 Result.setLength(TokEnd-BufferPtr); 189 BufferPtr = TokEnd; 190 } 191 192 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a 193 /// tok::l_paren token, 0 if it is something else and 2 if there are no more 194 /// tokens in the buffer controlled by this lexer. 195 unsigned isNextPPTokenLParen(); 196 197 //===--------------------------------------------------------------------===// 198 // Lexer character reading interfaces. 199public: 200 201 // This lexer is built on two interfaces for reading characters, both of which 202 // automatically provide phase 1/2 translation. getAndAdvanceChar is used 203 // when we know that we will be reading a character from the input buffer and 204 // that this character will be part of the result token. This occurs in (f.e.) 205 // string processing, because we know we need to read until we find the 206 // closing '"' character. 207 // 208 // The second interface is the combination of PeekCharAndSize with 209 // ConsumeChar. PeekCharAndSize reads a phase 1/2 translated character, 210 // returning it and its size. If the lexer decides that this character is 211 // part of the current token, it calls ConsumeChar on it. This two stage 212 // approach allows us to emit diagnostics for characters (e.g. warnings about 213 // trigraphs), knowing that they only are emitted if the character is 214 // consumed. 215 216 /// isObviouslySimpleCharacter - Return true if the specified character is 217 /// obviously the same in translation phase 1 and translation phase 3. This 218 /// can return false for characters that end up being the same, but it will 219 /// never return true for something that needs to be mapped. 220 static bool isObviouslySimpleCharacter(char C) { 221 return C != '?' && C != '\\'; 222 } 223 224 /// getAndAdvanceChar - Read a single 'character' from the specified buffer, 225 /// advance over it, and return it. This is tricky in several cases. Here we 226 /// just handle the trivial case and fall-back to the non-inlined 227 /// getCharAndSizeSlow method to handle the hard case. 228 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) { 229 // If this is not a trigraph and not a UCN or escaped newline, return 230 // quickly. 231 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++; 232 233 unsigned Size = 0; 234 char C = getCharAndSizeSlow(Ptr, Size, &Tok); 235 Ptr += Size; 236 return C; 237 } 238 239private: 240 /// ConsumeChar - When a character (identified by PeekCharAndSize) is consumed 241 /// and added to a given token, check to see if there are diagnostics that 242 /// need to be emitted or flags that need to be set on the token. If so, do 243 /// it. 244 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) { 245 // Normal case, we consumed exactly one token. Just return it. 246 if (Size == 1) 247 return Ptr+Size; 248 249 // Otherwise, re-lex the character with a current token, allowing 250 // diagnostics to be emitted and flags to be set. 251 Size = 0; 252 getCharAndSizeSlow(Ptr, Size, &Tok); 253 return Ptr+Size; 254 } 255 256 /// getCharAndSize - Peek a single 'character' from the specified buffer, 257 /// get its size, and return it. This is tricky in several cases. Here we 258 /// just handle the trivial case and fall-back to the non-inlined 259 /// getCharAndSizeSlow method to handle the hard case. 260 inline char getCharAndSize(const char *Ptr, unsigned &Size) { 261 // If this is not a trigraph and not a UCN or escaped newline, return 262 // quickly. 263 if (isObviouslySimpleCharacter(Ptr[0])) { 264 Size = 1; 265 return *Ptr; 266 } 267 268 Size = 0; 269 return getCharAndSizeSlow(Ptr, Size); 270 } 271 272 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize 273 /// method. 274 char getCharAndSizeSlow(const char *Ptr, unsigned &Size, Token *Tok = 0); 275 276 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever 277 /// emit a warning. 278 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size, 279 const LangOptions &Features) { 280 // If this is not a trigraph and not a UCN or escaped newline, return 281 // quickly. 282 if (isObviouslySimpleCharacter(Ptr[0])) { 283 Size = 1; 284 return *Ptr; 285 } 286 287 Size = 0; 288 return getCharAndSizeSlowNoWarn(Ptr, Size, Features); 289 } 290 291 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a 292 /// diagnostic. 293 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size, 294 const LangOptions &Features); 295 296 //===--------------------------------------------------------------------===// 297 // #if directive handling. 298 299 /// pushConditionalLevel - When we enter a #if directive, this keeps track of 300 /// what we are currently in for diagnostic emission (e.g. #if with missing 301 /// #endif). 302 void pushConditionalLevel(SourceLocation DirectiveStart, bool WasSkipping, 303 bool FoundNonSkip, bool FoundElse) { 304 PPConditionalInfo CI; 305 CI.IfLoc = DirectiveStart; 306 CI.WasSkipping = WasSkipping; 307 CI.FoundNonSkip = FoundNonSkip; 308 CI.FoundElse = FoundElse; 309 ConditionalStack.push_back(CI); 310 } 311 void pushConditionalLevel(const PPConditionalInfo &CI) { 312 ConditionalStack.push_back(CI); 313 } 314 315 /// popConditionalLevel - Remove an entry off the top of the conditional 316 /// stack, returning information about it. If the conditional stack is empty, 317 /// this returns true and does not fill in the arguments. 318 bool popConditionalLevel(PPConditionalInfo &CI) { 319 if (ConditionalStack.empty()) return true; 320 CI = ConditionalStack.back(); 321 ConditionalStack.pop_back(); 322 return false; 323 } 324 325 /// peekConditionalLevel - Return the top of the conditional stack. This 326 /// requires that there be a conditional active. 327 PPConditionalInfo &peekConditionalLevel() { 328 assert(!ConditionalStack.empty() && "No conditionals active!"); 329 return ConditionalStack.back(); 330 } 331 332 unsigned getConditionalStackDepth() const { return ConditionalStack.size(); } 333 334 //===--------------------------------------------------------------------===// 335 // Other lexer functions. 336 337 // Helper functions to lex the remainder of a token of the specific type. 338 void LexIdentifier (Token &Result, const char *CurPtr); 339 void LexNumericConstant (Token &Result, const char *CurPtr); 340 void LexStringLiteral (Token &Result, const char *CurPtr,bool Wide); 341 void LexAngledStringLiteral(Token &Result, const char *CurPtr); 342 void LexCharConstant (Token &Result, const char *CurPtr); 343 bool LexEndOfFile (Token &Result, const char *CurPtr); 344 345 void SkipWhitespace (Token &Result, const char *CurPtr); 346 bool SkipBCPLComment (Token &Result, const char *CurPtr); 347 bool SkipBlockComment (Token &Result, const char *CurPtr); 348 bool SaveBCPLComment (Token &Result, const char *CurPtr); 349 350 /// LexIncludeFilename - After the preprocessor has parsed a #include, lex and 351 /// (potentially) macro expand the filename. If the sequence parsed is not 352 /// lexically legal, emit a diagnostic and return a result EOM token. 353 void LexIncludeFilename(Token &Result); 354}; 355 356 357} // end namespace clang 358 359#endif 360