tools/llvm-mc/AsmLexer.cpp

   1 //===- AsmLexer.cpp - Lexer for Assembly Files ----------------------------===//
   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 class implements the lexer for assembly files.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "AsmLexer.h"
  15 #include "llvm/ADT/StringSet.h"
  16 #include "llvm/Support/SourceMgr.h"
  17 #include "llvm/Support/MemoryBuffer.h"
  18 #include "llvm/Config/config.h"  // for strtoull.
  19 #include <cerrno>
  20 #include <cstdio>
  21 #include <cstdlib>
  22 using namespace llvm;
  23
  24 static StringSet<> &getSS(void *TheSS) {
  25   return *(StringSet<>*)TheSS;
  26 }
  27
  28 AsmLexer::AsmLexer(SourceMgr &SM) : SrcMgr(SM) {
  29   CurBuffer = 0;
  30   CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
  31   CurPtr = CurBuf->getBufferStart();
  32   TokStart = 0;
  33
  34   TheStringSet = new StringSet<>();
  35 }
  36
  37 AsmLexer::~AsmLexer() {
  38   delete &getSS(TheStringSet);
  39 }
  40
  41 SMLoc AsmLexer::getLoc() const {
  42   return SMLoc::getFromPointer(TokStart);
  43 }
  44
  45 void AsmLexer::PrintMessage(SMLoc Loc, const std::string &Msg,
  46                             const char *Type) const {
  47   SrcMgr.PrintMessage(Loc, Msg, Type);
  48 }
  49
  50 /// ReturnError - Set the error to the specified string at the specified
  51 /// location.  This is defined to always return asmtok::Error.
  52 asmtok::TokKind AsmLexer::ReturnError(const char *Loc, const std::string &Msg) {
  53   SrcMgr.PrintMessage(SMLoc::getFromPointer(Loc), Msg, "error");
  54   return asmtok::Error;
  55 }
  56
  57 /// EnterIncludeFile - Enter the specified file.  This prints an error and
  58 /// returns true on failure.
  59 bool AsmLexer::EnterIncludeFile(const std::string &Filename) {
  60   int NewBuf = SrcMgr.AddIncludeFile(Filename, SMLoc::getFromPointer(CurPtr));
  61   if (NewBuf == -1)
  62     return true;
  63
  64   // Save the line number and lex buffer of the includer.
  65   CurBuffer = NewBuf;
  66   CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
  67   CurPtr = CurBuf->getBufferStart();
  68   return false;
  69 }
  70
  71
  72 int AsmLexer::getNextChar() {
  73   char CurChar = *CurPtr++;
  74   switch (CurChar) {
  75   default:
  76     return (unsigned char)CurChar;
  77   case 0: {
  78     // A nul character in the stream is either the end of the current buffer or
  79     // a random nul in the file.  Disambiguate that here.
  80     if (CurPtr-1 != CurBuf->getBufferEnd())
  81       return 0;  // Just whitespace.
  82
  83     // If this is the end of an included file, pop the parent file off the
  84     // include stack.
  85     SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
  86     if (ParentIncludeLoc != SMLoc()) {
  87       CurBuffer = SrcMgr.FindBufferContainingLoc(ParentIncludeLoc);
  88       CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
  89       CurPtr = ParentIncludeLoc.getPointer();
  90
  91       // Reset the token start pointer to the start of the new file.
  92       TokStart = CurPtr;
  93
  94       return getNextChar();
  95     }
  96
  97     // Otherwise, return end of file.
  98     --CurPtr;  // Another call to lex will return EOF again.
  99     return EOF;
 100   }
 101   }
 102 }
 103
 104 /// LexIdentifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
 105 asmtok::TokKind AsmLexer::LexIdentifier() {
 106   while (isalnum(*CurPtr) || *CurPtr == '_' || *CurPtr == '$' ||
 107          *CurPtr == '.' || *CurPtr == '@')
 108     ++CurPtr;
 109   // Unique string.
 110   CurStrVal =
 111     getSS(TheStringSet).GetOrCreateValue(TokStart, CurPtr, 0).getKeyData();
 112   return asmtok::Identifier;
 113 }
 114
 115 /// LexPercent: Register: %[a-zA-Z0-9]+
 116 asmtok::TokKind AsmLexer::LexPercent() {
 117   if (!isalnum(*CurPtr))
 118     return asmtok::Percent;  // Single %.
 119
 120   while (isalnum(*CurPtr))
 121     ++CurPtr;
 122
 123   // Unique string.
 124   CurStrVal =
 125     getSS(TheStringSet).GetOrCreateValue(TokStart, CurPtr, 0).getKeyData();
 126   return asmtok::Register;
 127 }
 128
 129 /// LexSlash: Slash: /
 130 ///           C-Style Comment: /* ... */
 131 asmtok::TokKind AsmLexer::LexSlash() {
 132   switch (*CurPtr) {
 133   case '*': break; // C style comment.
 134   case '/': return ++CurPtr, LexLineComment();
 135   default:  return asmtok::Slash;
 136   }
 137
 138   // C Style comment.
 139   ++CurPtr;  // skip the star.
 140   while (1) {
 141     int CurChar = getNextChar();
 142     switch (CurChar) {
 143     case EOF:
 144       return ReturnError(TokStart, "unterminated comment");
 145     case '*':
 146       // End of the comment?
 147       if (CurPtr[0] != '/') break;
 148
 149       ++CurPtr;   // End the */.
 150       return LexToken();
 151     }
 152   }
 153 }
 154
 155 /// LexLineComment: Comment: #[^\n]*
 156 ///                        : //[^\n]*
 157 asmtok::TokKind AsmLexer::LexLineComment() {
 158   int CurChar = getNextChar();
 159   while (CurChar != '\n' && CurChar != '\n' && CurChar != EOF)
 160     CurChar = getNextChar();
 161
 162   if (CurChar == EOF)
 163     return asmtok::Eof;
 164   return asmtok::EndOfStatement;
 165 }
 166
 167
 168 /// LexDigit: First character is [0-9].
 169 ///   Local Label: [0-9][:]
 170 ///   Forward/Backward Label: [0-9][fb]
 171 ///   Binary integer: 0b[01]+
 172 ///   Octal integer: 0[0-7]+
 173 ///   Hex integer: 0x[0-9a-fA-F]+
 174 ///   Decimal integer: [1-9][0-9]*
 175 /// TODO: FP literal.
 176 asmtok::TokKind AsmLexer::LexDigit() {
 177   if (*CurPtr == ':')
 178     return ReturnError(TokStart, "FIXME: local label not implemented");
 179   if (*CurPtr == 'f' || *CurPtr == 'b')
 180     return ReturnError(TokStart, "FIXME: directional label not implemented");
 181
 182   // Decimal integer: [1-9][0-9]*
 183   if (CurPtr[-1] != '0') {
 184     while (isdigit(*CurPtr))
 185       ++CurPtr;
 186     CurIntVal = strtoll(TokStart, 0, 10);
 187     return asmtok::IntVal;
 188   }
 189
 190   if (*CurPtr == 'b') {
 191     ++CurPtr;
 192     const char *NumStart = CurPtr;
 193     while (CurPtr[0] == '0' || CurPtr[0] == '1')
 194       ++CurPtr;
 195
 196     // Requires at least one binary digit.
 197     if (CurPtr == NumStart)
 198       return ReturnError(CurPtr-2, "Invalid binary number");
 199     CurIntVal = strtoll(NumStart, 0, 2);
 200     return asmtok::IntVal;
 201   }
 202
 203   if (*CurPtr == 'x') {
 204     ++CurPtr;
 205     const char *NumStart = CurPtr;
 206     while (isxdigit(CurPtr[0]))
 207       ++CurPtr;
 208
 209     // Requires at least one hex digit.
 210     if (CurPtr == NumStart)
 211       return ReturnError(CurPtr-2, "Invalid hexadecimal number");
 212
 213     errno = 0;
 214     CurIntVal = strtoll(NumStart, 0, 16);
 215     if (errno == EINVAL)
 216       return ReturnError(CurPtr-2, "Invalid hexadecimal number");
 217     if (errno == ERANGE) {
 218       errno = 0;
 219       CurIntVal = (int64_t)strtoull(NumStart, 0, 16);
 220       if (errno == EINVAL)
 221         return ReturnError(CurPtr-2, "Invalid hexadecimal number");
 222       if (errno == ERANGE)
 223         return ReturnError(CurPtr-2, "Hexadecimal number out of range");
 224     }
 225     return asmtok::IntVal;
 226   }
 227
 228   // Must be an octal number, it starts with 0.
 229   while (*CurPtr >= '0' && *CurPtr <= '7')
 230     ++CurPtr;
 231   CurIntVal = strtoll(TokStart, 0, 8);
 232   return asmtok::IntVal;
 233 }
 234
 235 /// LexQuote: String: "..."
 236 asmtok::TokKind AsmLexer::LexQuote() {
 237   int CurChar = getNextChar();
 238   // TODO: does gas allow multiline string constants?
 239   while (CurChar != '"') {
 240     if (CurChar == '\\') {
 241       // Allow \", etc.
 242       CurChar = getNextChar();
 243     }
 244
 245     if (CurChar == EOF)
 246       return ReturnError(TokStart, "unterminated string constant");
 247
 248     CurChar = getNextChar();
 249   }
 250
 251   // Unique string, include quotes for now.
 252   CurStrVal =
 253     getSS(TheStringSet).GetOrCreateValue(TokStart, CurPtr, 0).getKeyData();
 254   return asmtok::String;
 255 }
 256
 257
 258 asmtok::TokKind AsmLexer::LexToken() {
 259   TokStart = CurPtr;
 260   // This always consumes at least one character.
 261   int CurChar = getNextChar();
 262
 263   switch (CurChar) {
 264   default:
 265     // Handle identifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
 266     if (isalpha(CurChar) || CurChar == '_' || CurChar == '.')
 267       return LexIdentifier();
 268
 269     // Unknown character, emit an error.
 270     return ReturnError(TokStart, "invalid character in input");
 271   case EOF: return asmtok::Eof;
 272   case 0:
 273   case ' ':
 274   case '\t':
 275     // Ignore whitespace.
 276     return LexToken();
 277   case '\n': // FALL THROUGH.
 278   case '\r': // FALL THROUGH.
 279   case ';': return asmtok::EndOfStatement;
 280   case ':': return asmtok::Colon;
 281   case '+': return asmtok::Plus;
 282   case '-': return asmtok::Minus;
 283   case '~': return asmtok::Tilde;
 284   case '(': return asmtok::LParen;
 285   case ')': return asmtok::RParen;
 286   case '*': return asmtok::Star;
 287   case ',': return asmtok::Comma;
 288   case '$': return asmtok::Dollar;
 289   case '=':
 290     if (*CurPtr == '=')
 291       return ++CurPtr, asmtok::EqualEqual;
 292     return asmtok::Equal;
 293   case '|':
 294     if (*CurPtr == '|')
 295       return ++CurPtr, asmtok::PipePipe;
 296     return asmtok::Pipe;
 297   case '^': return asmtok::Caret;
 298   case '&':
 299     if (*CurPtr == '&')
 300       return ++CurPtr, asmtok::AmpAmp;
 301     return asmtok::Amp;
 302   case '!':
 303     if (*CurPtr == '=')
 304       return ++CurPtr, asmtok::ExclaimEqual;
 305     return asmtok::Exclaim;
 306   case '%': return LexPercent();
 307   case '/': return LexSlash();
 308   case '#': return LexLineComment();
 309   case '"': return LexQuote();
 310   case '0': case '1': case '2': case '3': case '4':
 311   case '5': case '6': case '7': case '8': case '9':
 312     return LexDigit();
 313   case '<':
 314     switch (*CurPtr) {
 315     case '<': return ++CurPtr, asmtok::LessLess;
 316     case '=': return ++CurPtr, asmtok::LessEqual;
 317     case '>': return ++CurPtr, asmtok::LessGreater;
 318     default: return asmtok::Less;
 319     }
 320   case '>':
 321     switch (*CurPtr) {
 322     case '>': return ++CurPtr, asmtok::GreaterGreater;
 323     case '=': return ++CurPtr, asmtok::GreaterEqual;
 324     default: return asmtok::Greater;
 325     }
 326
 327   // TODO: Quoted identifiers (objc methods etc)
 328   // local labels: [0-9][:]
 329   // Forward/backward labels: [0-9][fb]
 330   // Integers, fp constants, character constants.
 331   }
 332 }