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.
23 AsmLexer::AsmLexer(SourceMgr &SM) : SrcMgr(SM) {
25 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
26 CurPtr = CurBuf->getBufferStart();
30 SMLoc AsmLexer::getLoc() const {
31 return SMLoc::getFromPointer(TokStart);
34 void AsmLexer::PrintMessage(SMLoc Loc, const std::string &Msg) const {
35 SrcMgr.PrintMessage(Loc, Msg);
38 /// ReturnError - Set the error to the specified string at the specified
39 /// location. This is defined to always return asmtok::Error.
40 asmtok::TokKind AsmLexer::ReturnError(const char *Loc, const std::string &Msg) {
41 SrcMgr.PrintMessage(SMLoc::getFromPointer(Loc), Msg);
45 int AsmLexer::getNextChar() {
46 char CurChar = *CurPtr++;
49 return (unsigned char)CurChar;
51 // A nul character in the stream is either the end of the current buffer or
52 // a random nul in the file. Disambiguate that here.
53 if (CurPtr-1 != CurBuf->getBufferEnd())
54 return 0; // Just whitespace.
56 // If this is the end of an included file, pop the parent file off the
58 SMLoc ParentIncludeLoc = SrcMgr.getParentIncludeLoc(CurBuffer);
59 if (ParentIncludeLoc != SMLoc()) {
60 CurBuffer = SrcMgr.FindBufferContainingLoc(ParentIncludeLoc);
61 CurBuf = SrcMgr.getMemoryBuffer(CurBuffer);
62 CurPtr = ParentIncludeLoc.getPointer();
66 // Otherwise, return end of file.
67 --CurPtr; // Another call to lex will return EOF again.
73 /// LexIdentifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
74 asmtok::TokKind AsmLexer::LexIdentifier() {
75 while (isalnum(*CurPtr) || *CurPtr == '_' || *CurPtr == '$' ||
76 *CurPtr == '.' || *CurPtr == '@')
78 CurStrVal.assign(TokStart, CurPtr); // Include %
79 return asmtok::Identifier;
82 /// LexPercent: Register: %[a-zA-Z0-9]+
83 asmtok::TokKind AsmLexer::LexPercent() {
84 if (!isalnum(*CurPtr))
85 return ReturnError(TokStart, "invalid register name");
86 while (isalnum(*CurPtr))
88 CurStrVal.assign(TokStart, CurPtr); // Skip %
89 return asmtok::Register;
92 /// LexSlash: Slash: /
93 /// C-Style Comment: /* ... */
94 asmtok::TokKind AsmLexer::LexSlash() {
99 ++CurPtr; // skip the star.
101 int CurChar = getNextChar();
104 return ReturnError(TokStart, "unterminated comment");
106 // End of the comment?
107 if (CurPtr[0] != '/') break;
109 ++CurPtr; // End the */.
115 /// LexHash: Comment: #[^\n]*
116 asmtok::TokKind AsmLexer::LexHash() {
117 int CurChar = getNextChar();
118 while (CurChar != '\n' && CurChar != '\n' && CurChar != EOF)
119 CurChar = getNextChar();
123 return asmtok::EndOfStatement;
127 /// LexDigit: First character is [0-9].
128 /// Local Label: [0-9][:]
129 /// Forward/Backward Label: [0-9][fb]
130 /// Binary integer: 0b[01]+
131 /// Octal integer: 0[0-7]+
132 /// Hex integer: 0x[0-9a-fA-F]+
133 /// Decimal integer: [1-9][0-9]*
134 /// TODO: FP literal.
135 asmtok::TokKind AsmLexer::LexDigit() {
137 return ReturnError(TokStart, "FIXME: local label not implemented");
138 if (*CurPtr == 'f' || *CurPtr == 'b')
139 return ReturnError(TokStart, "FIXME: directional label not implemented");
141 // Decimal integer: [1-9][0-9]*
142 if (CurPtr[-1] != '0') {
143 while (isdigit(*CurPtr))
145 CurIntVal = strtoll(TokStart, 0, 10);
146 return asmtok::IntVal;
149 if (*CurPtr == 'b') {
151 const char *NumStart = CurPtr;
152 while (CurPtr[0] == '0' || CurPtr[0] == '1')
155 // Requires at least one binary digit.
156 if (CurPtr == NumStart)
157 return ReturnError(CurPtr-2, "Invalid binary number");
158 CurIntVal = strtoll(NumStart, 0, 2);
159 return asmtok::IntVal;
162 if (*CurPtr == 'x') {
164 const char *NumStart = CurPtr;
165 while (isxdigit(CurPtr[0]))
168 // Requires at least one hex digit.
169 if (CurPtr == NumStart)
170 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
173 CurIntVal = strtoll(NumStart, 0, 16);
175 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
176 if (errno == ERANGE) {
178 CurIntVal = (int64_t)strtoull(NumStart, 0, 16);
180 return ReturnError(CurPtr-2, "Invalid hexadecimal number");
182 return ReturnError(CurPtr-2, "Hexadecimal number out of range");
184 return asmtok::IntVal;
187 // Must be an octal number, it starts with 0.
188 while (*CurPtr >= '0' && *CurPtr <= '7')
190 CurIntVal = strtoll(TokStart, 0, 8);
191 return asmtok::IntVal;
194 /// LexQuote: String: "..."
195 asmtok::TokKind AsmLexer::LexQuote() {
196 int CurChar = getNextChar();
197 // TODO: does gas allow multiline string constants?
198 while (CurChar != '"') {
199 if (CurChar == '\\') {
201 CurChar = getNextChar();
205 return ReturnError(TokStart, "unterminated string constant");
207 CurChar = getNextChar();
210 CurStrVal.assign(TokStart, CurPtr); // include quotes.
211 return asmtok::String;
215 asmtok::TokKind AsmLexer::LexToken() {
217 // This always consumes at least one character.
218 int CurChar = getNextChar();
222 // Handle identifier: [a-zA-Z_.][a-zA-Z0-9_$.@]*
223 if (isalpha(CurChar) || CurChar == '_' || CurChar == '.')
224 return LexIdentifier();
226 // Unknown character, emit an error.
227 return ReturnError(TokStart, "invalid character in input");
228 case EOF: return asmtok::Eof;
232 // Ignore whitespace.
234 case '\n': // FALL THROUGH.
235 case '\r': // FALL THROUGH.
236 case ';': return asmtok::EndOfStatement;
237 case ':': return asmtok::Colon;
238 case '+': return asmtok::Plus;
239 case '-': return asmtok::Minus;
240 case '~': return asmtok::Tilde;
241 case '(': return asmtok::LParen;
242 case ')': return asmtok::RParen;
243 case '*': return asmtok::Star;
244 case ',': return asmtok::Comma;
245 case '$': return asmtok::Dollar;
246 case '%': return LexPercent();
247 case '/': return LexSlash();
248 case '#': return LexHash();
249 case '"': return LexQuote();
250 case '0': case '1': case '2': case '3': case '4':
251 case '5': case '6': case '7': case '8': case '9':
254 // TODO: Quoted identifiers (objc methods etc)
255 // local labels: [0-9][:]
256 // Forward/backward labels: [0-9][fb]
257 // Integers, fp constants, character constants.