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 SetError(SMLoc::getFromPointer(Loc), Msg);
49 return AsmToken(AsmToken::Error, StringRef(Loc, 0));
52 /// EnterIncludeFile - Enter the specified file. This prints an error and
53 /// returns true on failure.
54 bool AsmLexer::EnterIncludeFile(const std::string &Filename) {
55 int NewBuf = SrcMgr.AddIncludeFile(Filename, SMLoc::getFromPointer(CurPtr));
59 // Save the line number and lex buffer of the includer.
61 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
62 CurPtr = CurBuf->getBufferStart();
67 int AsmLexer::getNextChar() {
68 char CurChar = *CurPtr++;
71 return (unsigned char)CurChar;
73 // A nul character in the stream is either the end of the current buffer or
74 // a random nul in the file. Disambiguate that here.
75 if (CurPtr-1 != CurBuf->getBufferEnd())
76 return 0; // Just whitespace.
78 // If this is the end of an included file, pop the parent file off the
80 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
81 if (ParentIncludeLoc != SMLoc()) {
82 CurBuffer = SrcMgr.FindBufferContainingLoc(ParentIncludeLoc);
83 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
84 CurPtr = ParentIncludeLoc.getPointer();
86 // Reset the token start pointer to the start of the new file.
92 // Otherwise, return end of file.
93 --CurPtr; // Another call to lex will return EOF again.
99 /// LexIdentifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
100 AsmToken AsmLexer::LexIdentifier() {
101 while (isalnum(*CurPtr) || *CurPtr == '_' || *CurPtr == '$' ||
102 *CurPtr == '.' || *CurPtr == '@')
104 return AsmToken(AsmToken::Identifier, StringRef(TokStart, CurPtr - TokStart));
107 /// LexSlash: Slash: /
108 /// C-Style Comment: /* ... */
109 AsmToken AsmLexer::LexSlash() {
111 case '*': break; // C style comment.
112 case '/': return ++CurPtr, LexLineComment();
113 default: return AsmToken(AsmToken::Slash, StringRef(CurPtr, 1));
117 ++CurPtr; // skip the star.
119 int CurChar = getNextChar();
122 return ReturnError(TokStart, "unterminated comment");
124 // End of the comment?
125 if (CurPtr[0] != '/') break;
127 ++CurPtr; // End the */.
133 /// LexLineComment: Comment: #[^\n]*
135 AsmToken AsmLexer::LexLineComment() {
136 // FIXME: This is broken if we happen to a comment at the end of a file, which
137 // was .included, and which doesn't end with a newline.
138 int CurChar = getNextChar();
139 while (CurChar != '\n' && CurChar != '\n' && CurChar != EOF)
140 CurChar = getNextChar();
143 return AsmToken(AsmToken::Eof, StringRef(CurPtr, 0));
144 return AsmToken(AsmToken::EndOfStatement, StringRef(CurPtr, 0));
148 /// LexDigit: First character is [0-9].
149 /// Local Label: [0-9][:]
150 /// Forward/Backward Label: [0-9][fb]
151 /// Binary integer: 0b[01]+
152 /// Octal integer: 0[0-7]+
153 /// Hex integer: 0x[0-9a-fA-F]+
154 /// Decimal integer: [1-9][0-9]*
155 /// TODO: FP literal.
156 AsmToken AsmLexer::LexDigit() {
158 return ReturnError(TokStart, "FIXME: local label not implemented");
159 if (*CurPtr == 'f' || *CurPtr == 'b')
160 return ReturnError(TokStart, "FIXME: directional label not implemented");
162 // Decimal integer: [1-9][0-9]*
163 if (CurPtr[-1] != '0') {
164 while (isdigit(*CurPtr))
166 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
167 strtoll(TokStart, 0, 10));
170 if (*CurPtr == 'b') {
172 const char *NumStart = CurPtr;
173 while (CurPtr[0] == '0' || CurPtr[0] == '1')
176 // Requires at least one binary digit.
177 if (CurPtr == NumStart)
178 return ReturnError(CurPtr-2, "Invalid binary number");
179 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
180 strtoll(NumStart, 0, 2));
183 if (*CurPtr == 'x') {
185 const char *NumStart = CurPtr;
186 while (isxdigit(CurPtr[0]))
189 // Requires at least one hex digit.
190 if (CurPtr == NumStart)
191 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
195 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
196 if (errno == ERANGE) {
199 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
201 return ReturnError(CurPtr-2, "Hexadecimal number out of range");
203 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
204 (int64_t) strtoull(NumStart, 0, 16));
207 // Must be an octal number, it starts with 0.
208 while (*CurPtr >= '0' && *CurPtr <= '7')
210 return AsmToken(AsmToken::Integer, StringRef(TokStart, CurPtr - TokStart),
211 strtoll(TokStart, 0, 8));
214 /// LexQuote: String: "..."
215 AsmToken AsmLexer::LexQuote() {
216 int CurChar = getNextChar();
217 // TODO: does gas allow multiline string constants?
218 while (CurChar != '"') {
219 if (CurChar == '\\') {
221 CurChar = getNextChar();
225 return ReturnError(TokStart, "unterminated string constant");
227 CurChar = getNextChar();
230 return AsmToken(AsmToken::String, StringRef(TokStart, CurPtr - TokStart));
233 StringRef AsmLexer::LexUntilEndOfStatement() {
236 while (!isAtStartOfComment(*CurPtr) && // Start of line comment.
237 *CurPtr != ';' && // End of statement marker.
240 (*CurPtr != 0 || CurPtr != CurBuf->getBufferEnd())) {
243 return StringRef(TokStart, CurPtr-TokStart);
246 bool AsmLexer::isAtStartOfComment(char Char) {
247 // FIXME: This won't work for multi-character comment indicators like "//".
248 return Char == *MAI.getCommentString();
251 AsmToken AsmLexer::LexToken() {
253 // This always consumes at least one character.
254 int CurChar = getNextChar();
256 if (isAtStartOfComment(CurChar))
257 return LexLineComment();
261 // Handle identifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
262 if (isalpha(CurChar) || CurChar == '_' || CurChar == '.')
263 return LexIdentifier();
265 // Unknown character, emit an error.
266 return ReturnError(TokStart, "invalid character in input");
267 case EOF: return AsmToken(AsmToken::Eof, StringRef(TokStart, 0));
271 // Ignore whitespace.
273 case '\n': // FALL THROUGH.
274 case '\r': // FALL THROUGH.
275 case ';': return AsmToken(AsmToken::EndOfStatement, StringRef(TokStart, 1));
276 case ':': return AsmToken(AsmToken::Colon, StringRef(TokStart, 1));
277 case '+': return AsmToken(AsmToken::Plus, StringRef(TokStart, 1));
278 case '-': return AsmToken(AsmToken::Minus, StringRef(TokStart, 1));
279 case '~': return AsmToken(AsmToken::Tilde, StringRef(TokStart, 1));
280 case '(': return AsmToken(AsmToken::LParen, StringRef(TokStart, 1));
281 case ')': return AsmToken(AsmToken::RParen, StringRef(TokStart, 1));
282 case '[': return AsmToken(AsmToken::LBrac, StringRef(TokStart, 1));
283 case ']': return AsmToken(AsmToken::RBrac, StringRef(TokStart, 1));
284 case '{': return AsmToken(AsmToken::LCurly, StringRef(TokStart, 1));
285 case '}': return AsmToken(AsmToken::RCurly, StringRef(TokStart, 1));
286 case '*': return AsmToken(AsmToken::Star, StringRef(TokStart, 1));
287 case ',': return AsmToken(AsmToken::Comma, StringRef(TokStart, 1));
288 case '$': return AsmToken(AsmToken::Dollar, StringRef(TokStart, 1));
291 return ++CurPtr, AsmToken(AsmToken::EqualEqual, StringRef(TokStart, 2));
292 return AsmToken(AsmToken::Equal, StringRef(TokStart, 1));
295 return ++CurPtr, AsmToken(AsmToken::PipePipe, StringRef(TokStart, 2));
296 return AsmToken(AsmToken::Pipe, StringRef(TokStart, 1));
297 case '^': return AsmToken(AsmToken::Caret, StringRef(TokStart, 1));
300 return ++CurPtr, AsmToken(AsmToken::AmpAmp, StringRef(TokStart, 2));
301 return AsmToken(AsmToken::Amp, StringRef(TokStart, 1));
304 return ++CurPtr, AsmToken(AsmToken::ExclaimEqual, StringRef(TokStart, 2));
305 return AsmToken(AsmToken::Exclaim, StringRef(TokStart, 1));
306 case '%': return AsmToken(AsmToken::Percent, StringRef(TokStart, 1));
307 case '/': return LexSlash();
308 case '#': return AsmToken(AsmToken::Hash, StringRef(TokStart, 1));
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':
315 case '<': return ++CurPtr, AsmToken(AsmToken::LessLess,
316 StringRef(TokStart, 2));
317 case '=': return ++CurPtr, AsmToken(AsmToken::LessEqual,
318 StringRef(TokStart, 2));
319 case '>': return ++CurPtr, AsmToken(AsmToken::LessGreater,
320 StringRef(TokStart, 2));
321 default: return AsmToken(AsmToken::Less, StringRef(TokStart, 1));
325 case '>': return ++CurPtr, AsmToken(AsmToken::GreaterGreater,
326 StringRef(TokStart, 2));
327 case '=': return ++CurPtr, AsmToken(AsmToken::GreaterEqual,
328 StringRef(TokStart, 2));
329 default: return AsmToken(AsmToken::Greater, StringRef(TokStart, 1));
332 // TODO: Quoted identifiers (objc methods etc)
333 // local labels: [0-9][:]
334 // Forward/backward labels: [0-9][fb]
335 // Integers, fp constants, character constants.