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/ADT/StringSet.h"
16 #include "llvm/Support/SourceMgr.h"
17 #include "llvm/Support/MemoryBuffer.h"
18 #include "llvm/Config/config.h" // for strtoull.
24 static StringSet<> &getSS(void *TheSS) {
25 return *(StringSet<>*)TheSS;
28 AsmLexer::AsmLexer(SourceMgr &SM) : SrcMgr(SM) {
30 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
31 CurPtr = CurBuf->getBufferStart();
34 TheStringSet = new StringSet<>();
37 AsmLexer::~AsmLexer() {
38 delete &getSS(TheStringSet);
41 SMLoc AsmLexer::getLoc() const {
42 return SMLoc::getFromPointer(TokStart);
45 void AsmLexer::PrintMessage(SMLoc Loc, const std::string &Msg,
46 const char *Type) const {
47 SrcMgr.PrintMessage(Loc, Msg, Type);
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");
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));
64 // Save the line number and lex buffer of the includer.
66 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
67 CurPtr = CurBuf->getBufferStart();
72 int AsmLexer::getNextChar() {
73 char CurChar = *CurPtr++;
76 return (unsigned char)CurChar;
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.
83 // If this is the end of an included file, pop the parent file off the
85 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
86 if (ParentIncludeLoc != SMLoc()) {
87 CurBuffer = SrcMgr.FindBufferContainingLoc(ParentIncludeLoc);
88 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
89 CurPtr = ParentIncludeLoc.getPointer();
91 // Reset the token start pointer to the start of the new file.
97 // Otherwise, return end of file.
98 --CurPtr; // Another call to lex will return EOF again.
104 /// LexIdentifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
105 asmtok::TokKind AsmLexer::LexIdentifier() {
106 while (isalnum(*CurPtr) || *CurPtr == '_' || *CurPtr == '$' ||
107 *CurPtr == '.' || *CurPtr == '@')
111 getSS(TheStringSet).GetOrCreateValue(TokStart, CurPtr, 0).getKeyData();
112 return asmtok::Identifier;
115 /// LexPercent: Register: %[a-zA-Z0-9]+
116 asmtok::TokKind AsmLexer::LexPercent() {
117 if (!isalnum(*CurPtr))
118 return asmtok::Percent; // Single %.
120 while (isalnum(*CurPtr))
125 getSS(TheStringSet).GetOrCreateValue(TokStart, CurPtr, 0).getKeyData();
126 return asmtok::Register;
129 /// LexSlash: Slash: /
130 /// C-Style Comment: /* ... */
131 asmtok::TokKind AsmLexer::LexSlash() {
133 case '*': break; // C style comment.
134 case '/': return ++CurPtr, LexLineComment();
135 default: return asmtok::Slash;
139 ++CurPtr; // skip the star.
141 int CurChar = getNextChar();
144 return ReturnError(TokStart, "unterminated comment");
146 // End of the comment?
147 if (CurPtr[0] != '/') break;
149 ++CurPtr; // End the */.
155 /// LexLineComment: Comment: #[^\n]*
157 asmtok::TokKind AsmLexer::LexLineComment() {
158 int CurChar = getNextChar();
159 while (CurChar != '\n' && CurChar != '\n' && CurChar != EOF)
160 CurChar = getNextChar();
164 return asmtok::EndOfStatement;
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() {
178 return ReturnError(TokStart, "FIXME: local label not implemented");
179 if (*CurPtr == 'f' || *CurPtr == 'b')
180 return ReturnError(TokStart, "FIXME: directional label not implemented");
182 // Decimal integer: [1-9][0-9]*
183 if (CurPtr[-1] != '0') {
184 while (isdigit(*CurPtr))
186 CurIntVal = strtoll(TokStart, 0, 10);
187 return asmtok::IntVal;
190 if (*CurPtr == 'b') {
192 const char *NumStart = CurPtr;
193 while (CurPtr[0] == '0' || CurPtr[0] == '1')
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;
203 if (*CurPtr == 'x') {
205 const char *NumStart = CurPtr;
206 while (isxdigit(CurPtr[0]))
209 // Requires at least one hex digit.
210 if (CurPtr == NumStart)
211 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
214 CurIntVal = strtoll(NumStart, 0, 16);
216 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
217 if (errno == ERANGE) {
219 CurIntVal = (int64_t)strtoull(NumStart, 0, 16);
221 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
223 return ReturnError(CurPtr-2, "Hexadecimal number out of range");
225 return asmtok::IntVal;
228 // Must be an octal number, it starts with 0.
229 while (*CurPtr >= '0' && *CurPtr <= '7')
231 CurIntVal = strtoll(TokStart, 0, 8);
232 return asmtok::IntVal;
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 == '\\') {
242 CurChar = getNextChar();
246 return ReturnError(TokStart, "unterminated string constant");
248 CurChar = getNextChar();
251 // Unique string, include quotes for now.
253 getSS(TheStringSet).GetOrCreateValue(TokStart, CurPtr, 0).getKeyData();
254 return asmtok::String;
258 asmtok::TokKind AsmLexer::LexToken() {
260 // This always consumes at least one character.
261 int CurChar = getNextChar();
265 // Handle identifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
266 if (isalpha(CurChar) || CurChar == '_' || CurChar == '.')
267 return LexIdentifier();
269 // Unknown character, emit an error.
270 return ReturnError(TokStart, "invalid character in input");
271 case EOF: return asmtok::Eof;
275 // Ignore whitespace.
277 case '\n': // FALL THROUGH.
278 case '\r': // FALL THROUGH.
279 case ';': return asmtok::EndOfStatement;
280 case ':': return asmtok::Colon;
281 case '+': return asmtok::Plus;
282 case '-': return asmtok::Minus;
283 case '~': return asmtok::Tilde;
284 case '(': return asmtok::LParen;
285 case ')': return asmtok::RParen;
286 case '*': return asmtok::Star;
287 case ',': return asmtok::Comma;
288 case '$': return asmtok::Dollar;
291 return ++CurPtr, asmtok::EqualEqual;
292 return asmtok::Equal;
295 return ++CurPtr, asmtok::PipePipe;
297 case '^': return asmtok::Caret;
300 return ++CurPtr, asmtok::AmpAmp;
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':
315 case '<': return ++CurPtr, asmtok::LessLess;
316 case '=': return ++CurPtr, asmtok::LessEqual;
317 case '>': return ++CurPtr, asmtok::LessGreater;
318 default: return asmtok::Less;
322 case '>': return ++CurPtr, asmtok::GreaterGreater;
323 case '=': return ++CurPtr, asmtok::GreaterEqual;
324 default: return asmtok::Greater;
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.