1 //===- AsmLexer.cpp - Lexer for Assembly Files ----------------------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This class implements the lexer for assembly files.
12 //===----------------------------------------------------------------------===//
15 #include "llvm/Support/SourceMgr.h"
16 #include "llvm/Support/MemoryBuffer.h"
17 #include "llvm/Config/config.h" // for strtoull.
18 #include "llvm/MC/MCAsmInfo.h"
24 AsmLexer::AsmLexer(SourceMgr &SM, const MCAsmInfo &_MAI) : SrcMgr(SM),
27 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
28 CurPtr = CurBuf->getBufferStart();
32 AsmLexer::~AsmLexer() {
35 SMLoc AsmLexer::getLoc() const {
36 return SMLoc::getFromPointer(TokStart);
39 void AsmLexer::PrintMessage(SMLoc Loc, const std::string &Msg,
40 const char *Type) const {
41 SrcMgr.PrintMessage(Loc, Msg, Type);
44 /// ReturnError - Set the error to the specified string at the specified
45 /// location. This is defined to always return AsmToken::Error.
46 AsmToken AsmLexer::ReturnError(const char *Loc, const std::string &Msg) {
47 SrcMgr.PrintMessage(SMLoc::getFromPointer(Loc), Msg, "error");
48 return AsmToken(AsmToken::Error, StringRef(Loc, 0));
51 /// EnterIncludeFile - Enter the specified file. This prints an error and
52 /// returns true on failure.
53 bool AsmLexer::EnterIncludeFile(const std::string &Filename) {
54 int NewBuf = SrcMgr.AddIncludeFile(Filename, SMLoc::getFromPointer(CurPtr));
58 // Save the line number and lex buffer of the includer.
60 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
61 CurPtr = CurBuf->getBufferStart();
66 int AsmLexer::getNextChar() {
67 char CurChar = *CurPtr++;
70 return (unsigned char)CurChar;
72 // A nul character in the stream is either the end of the current buffer or
73 // a random nul in the file. Disambiguate that here.
74 if (CurPtr-1 != CurBuf->getBufferEnd())
75 return 0; // Just whitespace.
77 // If this is the end of an included file, pop the parent file off the
79 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
80 if (ParentIncludeLoc != SMLoc()) {
81 CurBuffer = SrcMgr.FindBufferContainingLoc(ParentIncludeLoc);
82 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
83 CurPtr = ParentIncludeLoc.getPointer();
85 // Reset the token start pointer to the start of the new file.
91 // Otherwise, return end of file.
92 --CurPtr; // Another call to lex will return EOF again.
98 /// LexIdentifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
99 AsmToken AsmLexer::LexIdentifier() {
100 while (isalnum(*CurPtr) || *CurPtr == '_' || *CurPtr == '$' ||
101 *CurPtr == '.' || *CurPtr == '@')
103 return AsmToken(AsmToken::Identifier, StringRef(TokStart, CurPtr - TokStart));
106 /// LexSlash: Slash: /
107 /// C-Style Comment: /* ... */
108 AsmToken AsmLexer::LexSlash() {
110 case '*': break; // C style comment.
111 case '/': return ++CurPtr, LexLineComment();
112 default: return AsmToken(AsmToken::Slash, StringRef(CurPtr, 1));
116 ++CurPtr; // skip the star.
118 int CurChar = getNextChar();
121 return ReturnError(TokStart, "unterminated comment");
123 // End of the comment?
124 if (CurPtr[0] != '/') break;
126 ++CurPtr; // End the */.
132 /// LexLineComment: Comment: #[^\n]*
134 AsmToken AsmLexer::LexLineComment() {
135 // FIXME: This is broken if we happen to a comment at the end of a file, which
136 // was .included, and which doesn't end with a newline.
137 int CurChar = getNextChar();
138 while (CurChar != '\n' && CurChar != '\n' && CurChar != EOF)
139 CurChar = getNextChar();
142 return AsmToken(AsmToken::Eof, StringRef(CurPtr, 0));
143 return AsmToken(AsmToken::EndOfStatement, StringRef(CurPtr, 0));
147 /// LexDigit: First character is [0-9].
148 /// Local Label: [0-9][:]
149 /// Forward/Backward Label: [0-9][fb]
150 /// Binary integer: 0b[01]+
151 /// Octal integer: 0[0-7]+
152 /// Hex integer: 0x[0-9a-fA-F]+
153 /// Decimal integer: [1-9][0-9]*
154 /// TODO: FP literal.
155 AsmToken AsmLexer::LexDigit() {
157 return ReturnError(TokStart, "FIXME: local label not implemented");
158 if (*CurPtr == 'f' || *CurPtr == 'b')
159 return ReturnError(TokStart, "FIXME: directional label not implemented");
161 // Decimal integer: [1-9][0-9]*
162 if (CurPtr[-1] != '0') {
163 while (isdigit(*CurPtr))
165 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
166 strtoll(TokStart, 0, 10));
169 if (*CurPtr == 'b') {
171 const char *NumStart = CurPtr;
172 while (CurPtr[0] == '0' || CurPtr[0] == '1')
175 // Requires at least one binary digit.
176 if (CurPtr == NumStart)
177 return ReturnError(CurPtr-2, "Invalid binary number");
178 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
179 strtoll(NumStart, 0, 2));
182 if (*CurPtr == 'x') {
184 const char *NumStart = CurPtr;
185 while (isxdigit(CurPtr[0]))
188 // Requires at least one hex digit.
189 if (CurPtr == NumStart)
190 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
194 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
195 if (errno == ERANGE) {
198 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
200 return ReturnError(CurPtr-2, "Hexadecimal number out of range");
202 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
203 (int64_t) strtoull(NumStart, 0, 16));
206 // Must be an octal number, it starts with 0.
207 while (*CurPtr >= '0' && *CurPtr <= '7')
209 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
210 strtoll(TokStart, 0, 8));
213 /// LexQuote: String: "..."
214 AsmToken AsmLexer::LexQuote() {
215 int CurChar = getNextChar();
216 // TODO: does gas allow multiline string constants?
217 while (CurChar != '"') {
218 if (CurChar == '\\') {
220 CurChar = getNextChar();
224 return ReturnError(TokStart, "unterminated string constant");
226 CurChar = getNextChar();
229 return AsmToken(AsmToken::String, StringRef(TokStart, CurPtr - TokStart));
232 StringRef AsmLexer::LexUntilEndOfStatement() {
235 while (!isAtStartOfComment(*CurPtr) && // Start of line comment.
236 *CurPtr != ';' && // End of statement marker.
239 (*CurPtr != 0 || CurPtr != CurBuf->getBufferEnd())) {
242 return StringRef(TokStart, CurPtr-TokStart);
245 bool AsmLexer::isAtStartOfComment(char Char) {
246 // FIXME: This won't work for multi-character comment indicators like "//".
247 return Char == *MAI.getCommentString();
250 AsmToken AsmLexer::LexToken() {
252 // This always consumes at least one character.
253 int CurChar = getNextChar();
255 if (isAtStartOfComment(CurChar))
256 return LexLineComment();
260 // Handle identifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
261 if (isalpha(CurChar) || CurChar == '_' || CurChar == '.')
262 return LexIdentifier();
264 // Unknown character, emit an error.
265 return ReturnError(TokStart, "invalid character in input");
266 case EOF: return AsmToken(AsmToken::Eof, StringRef(TokStart, 0));
270 // Ignore whitespace.
272 case '\n': // FALL THROUGH.
273 case '\r': // FALL THROUGH.
274 case ';': return AsmToken(AsmToken::EndOfStatement, StringRef(TokStart, 1));
275 case ':': return AsmToken(AsmToken::Colon, StringRef(TokStart, 1));
276 case '+': return AsmToken(AsmToken::Plus, StringRef(TokStart, 1));
277 case '-': return AsmToken(AsmToken::Minus, StringRef(TokStart, 1));
278 case '~': return AsmToken(AsmToken::Tilde, StringRef(TokStart, 1));
279 case '(': return AsmToken(AsmToken::LParen, StringRef(TokStart, 1));
280 case ')': return AsmToken(AsmToken::RParen, StringRef(TokStart, 1));
281 case '[': return AsmToken(AsmToken::LBrac, StringRef(TokStart, 1));
282 case ']': return AsmToken(AsmToken::RBrac, StringRef(TokStart, 1));
283 case '{': return AsmToken(AsmToken::LCurly, StringRef(TokStart, 1));
284 case '}': return AsmToken(AsmToken::RCurly, StringRef(TokStart, 1));
285 case '*': return AsmToken(AsmToken::Star, StringRef(TokStart, 1));
286 case ',': return AsmToken(AsmToken::Comma, StringRef(TokStart, 1));
287 case '$': return AsmToken(AsmToken::Dollar, StringRef(TokStart, 1));
290 return ++CurPtr, AsmToken(AsmToken::EqualEqual, StringRef(TokStart, 2));
291 return AsmToken(AsmToken::Equal, StringRef(TokStart, 1));
294 return ++CurPtr, AsmToken(AsmToken::PipePipe, StringRef(TokStart, 2));
295 return AsmToken(AsmToken::Pipe, StringRef(TokStart, 1));
296 case '^': return AsmToken(AsmToken::Caret, StringRef(TokStart, 1));
299 return ++CurPtr, AsmToken(AsmToken::AmpAmp, StringRef(TokStart, 2));
300 return AsmToken(AsmToken::Amp, StringRef(TokStart, 1));
303 return ++CurPtr, AsmToken(AsmToken::ExclaimEqual, StringRef(TokStart, 2));
304 return AsmToken(AsmToken::Exclaim, StringRef(TokStart, 1));
305 case '%': return AsmToken(AsmToken::Percent, StringRef(TokStart, 1));
306 case '/': return LexSlash();
307 case '#': return AsmToken(AsmToken::Hash, StringRef(TokStart, 1));
308 case '"': return LexQuote();
309 case '0': case '1': case '2': case '3': case '4':
310 case '5': case '6': case '7': case '8': case '9':
314 case '<': return ++CurPtr, AsmToken(AsmToken::LessLess,
315 StringRef(TokStart, 2));
316 case '=': return ++CurPtr, AsmToken(AsmToken::LessEqual,
317 StringRef(TokStart, 2));
318 case '>': return ++CurPtr, AsmToken(AsmToken::LessGreater,
319 StringRef(TokStart, 2));
320 default: return AsmToken(AsmToken::Less, StringRef(TokStart, 1));
324 case '>': return ++CurPtr, AsmToken(AsmToken::GreaterGreater,
325 StringRef(TokStart, 2));
326 case '=': return ++CurPtr, AsmToken(AsmToken::GreaterEqual,
327 StringRef(TokStart, 2));
328 default: return AsmToken(AsmToken::Greater, StringRef(TokStart, 1));
331 // TODO: Quoted identifiers (objc methods etc)
332 // local labels: [0-9][:]
333 // Forward/backward labels: [0-9][fb]
334 // Integers, fp constants, character constants.