1 //===- MILexer.cpp - Machine instructions lexer implementation ----------===//
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 file implements the lexing of machine instructions.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "MILexer.h"
15 #include "llvm/ADT/StringExtras.h"
16 #include "llvm/ADT/StringSwitch.h"
17 #include "llvm/ADT/Twine.h"
18 #include <cctype>
19 
20 using namespace llvm;
21 
22 namespace {
23 
24 /// This class provides a way to iterate and get characters from the source
25 /// string.
26 class Cursor {
27   const char *Ptr;
28   const char *End;
29 
30 public:
31   Cursor(NoneType) : Ptr(nullptr), End(nullptr) {}
32 
33   explicit Cursor(StringRef Str) {
34     Ptr = Str.data();
35     End = Ptr + Str.size();
36   }
37 
38   bool isEOF() const { return Ptr == End; }
39 
40   char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
41 
42   void advance(unsigned I = 1) { Ptr += I; }
43 
44   StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
45 
46   StringRef upto(Cursor C) const {
47     assert(C.Ptr >= Ptr && C.Ptr <= End);
48     return StringRef(Ptr, C.Ptr - Ptr);
49   }
50 
51   StringRef::iterator location() const { return Ptr; }
52 
53   operator bool() const { return Ptr != nullptr; }
54 };
55 
56 } // end anonymous namespace
57 
58 /// Skip the leading whitespace characters and return the updated cursor.
59 static Cursor skipWhitespace(Cursor C) {
60   while (isspace(C.peek()))
61     C.advance();
62   return C;
63 }
64 
65 /// Return true if the given character satisfies the following regular
66 /// expression: [-a-zA-Z$._0-9]
67 static bool isIdentifierChar(char C) {
68   return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
69          C == '$';
70 }
71 
72 void MIToken::unescapeQuotedStringValue(std::string &Str) const {
73   assert(isStringValueQuoted() && "String value isn't quoted");
74   StringRef Value = Range.drop_front(StringOffset);
75   assert(Value.front() == '"' && Value.back() == '"');
76   Cursor C = Cursor(Value.substr(1, Value.size() - 2));
77 
78   Str.clear();
79   Str.reserve(C.remaining().size());
80   while (!C.isEOF()) {
81     char Char = C.peek();
82     if (Char == '\\') {
83       if (C.peek(1) == '\\') {
84         // Two '\' become one
85         Str += '\\';
86         C.advance(2);
87         continue;
88       }
89       if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
90         Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
91         C.advance(3);
92         continue;
93       }
94     }
95     Str += Char;
96     C.advance();
97   }
98 }
99 
100 /// Lex a string constant using the following regular expression: \"[^\"]*\"
101 static Cursor lexStringConstant(
102     Cursor C,
103     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
104   assert(C.peek() == '"');
105   for (C.advance(); C.peek() != '"'; C.advance()) {
106     if (C.isEOF()) {
107       ErrorCallback(
108           C.location(),
109           "end of machine instruction reached before the closing '\"'");
110       return None;
111     }
112   }
113   C.advance();
114   return C;
115 }
116 
117 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
118   return StringSwitch<MIToken::TokenKind>(Identifier)
119       .Case("_", MIToken::underscore)
120       .Case("implicit", MIToken::kw_implicit)
121       .Case("implicit-def", MIToken::kw_implicit_define)
122       .Case("dead", MIToken::kw_dead)
123       .Case("killed", MIToken::kw_killed)
124       .Case("undef", MIToken::kw_undef)
125       .Case("frame-setup", MIToken::kw_frame_setup)
126       .Case(".cfi_def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
127       .Default(MIToken::Identifier);
128 }
129 
130 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
131   if (!isalpha(C.peek()) && C.peek() != '_' && C.peek() != '.')
132     return None;
133   auto Range = C;
134   while (isIdentifierChar(C.peek()))
135     C.advance();
136   auto Identifier = Range.upto(C);
137   Token = MIToken(getIdentifierKind(Identifier), Identifier);
138   return C;
139 }
140 
141 static Cursor maybeLexMachineBasicBlock(
142     Cursor C, MIToken &Token,
143     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
144   if (!C.remaining().startswith("%bb."))
145     return None;
146   auto Range = C;
147   C.advance(4); // Skip '%bb.'
148   if (!isdigit(C.peek())) {
149     Token = MIToken(MIToken::Error, C.remaining());
150     ErrorCallback(C.location(), "expected a number after '%bb.'");
151     return C;
152   }
153   auto NumberRange = C;
154   while (isdigit(C.peek()))
155     C.advance();
156   StringRef Number = NumberRange.upto(C);
157   unsigned StringOffset = 4 + Number.size(); // Drop '%bb.<id>'
158   if (C.peek() == '.') {
159     C.advance(); // Skip '.'
160     ++StringOffset;
161     while (isIdentifierChar(C.peek()))
162       C.advance();
163   }
164   Token = MIToken(MIToken::MachineBasicBlock, Range.upto(C), APSInt(Number),
165                   StringOffset);
166   return C;
167 }
168 
169 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
170                             MIToken::TokenKind Kind) {
171   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
172     return None;
173   auto Range = C;
174   C.advance(Rule.size());
175   auto NumberRange = C;
176   while (isdigit(C.peek()))
177     C.advance();
178   Token = MIToken(Kind, Range.upto(C), APSInt(NumberRange.upto(C)));
179   return C;
180 }
181 
182 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
183                                    MIToken::TokenKind Kind) {
184   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
185     return None;
186   auto Range = C;
187   C.advance(Rule.size());
188   auto NumberRange = C;
189   while (isdigit(C.peek()))
190     C.advance();
191   StringRef Number = NumberRange.upto(C);
192   unsigned StringOffset = Rule.size() + Number.size();
193   if (C.peek() == '.') {
194     C.advance();
195     ++StringOffset;
196     while (isIdentifierChar(C.peek()))
197       C.advance();
198   }
199   Token = MIToken(Kind, Range.upto(C), APSInt(Number), StringOffset);
200   return C;
201 }
202 
203 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
204   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
205 }
206 
207 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
208   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
209 }
210 
211 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
212   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
213 }
214 
215 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
216   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
217 }
218 
219 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
220   auto Range = C;
221   C.advance(); // Skip '%'
222   auto NumberRange = C;
223   while (isdigit(C.peek()))
224     C.advance();
225   Token = MIToken(MIToken::VirtualRegister, Range.upto(C),
226                   APSInt(NumberRange.upto(C)));
227   return C;
228 }
229 
230 static Cursor maybeLexRegister(Cursor C, MIToken &Token) {
231   if (C.peek() != '%')
232     return None;
233   if (isdigit(C.peek(1)))
234     return lexVirtualRegister(C, Token);
235   auto Range = C;
236   C.advance(); // Skip '%'
237   while (isIdentifierChar(C.peek()))
238     C.advance();
239   Token = MIToken(MIToken::NamedRegister, Range.upto(C),
240                   /*StringOffset=*/1); // Drop the '%'
241   return C;
242 }
243 
244 static Cursor lexName(
245     Cursor C, MIToken &Token, MIToken::TokenKind Type,
246     MIToken::TokenKind QuotedType, unsigned PrefixLength,
247     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
248   auto Range = C;
249   C.advance(PrefixLength);
250   if (C.peek() == '"') {
251     if (Cursor R = lexStringConstant(C, ErrorCallback)) {
252       Token = MIToken(QuotedType, Range.upto(R), PrefixLength);
253       return R;
254     }
255     Token = MIToken(MIToken::Error, Range.remaining());
256     return Range;
257   }
258   while (isIdentifierChar(C.peek()))
259     C.advance();
260   Token = MIToken(Type, Range.upto(C), PrefixLength);
261   return C;
262 }
263 
264 static Cursor maybeLexGlobalValue(
265     Cursor C, MIToken &Token,
266     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
267   if (C.peek() != '@')
268     return None;
269   if (!isdigit(C.peek(1)))
270     return lexName(C, Token, MIToken::NamedGlobalValue,
271                    MIToken::QuotedNamedGlobalValue, /*PrefixLength=*/1,
272                    ErrorCallback);
273   auto Range = C;
274   C.advance(1); // Skip the '@'
275   auto NumberRange = C;
276   while (isdigit(C.peek()))
277     C.advance();
278   Token =
279       MIToken(MIToken::GlobalValue, Range.upto(C), APSInt(NumberRange.upto(C)));
280   return C;
281 }
282 
283 static Cursor maybeLexExternalSymbol(
284     Cursor C, MIToken &Token,
285     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
286   if (C.peek() != '$')
287     return None;
288   return lexName(C, Token, MIToken::ExternalSymbol,
289                  MIToken::QuotedExternalSymbol,
290                  /*PrefixLength=*/1, ErrorCallback);
291 }
292 
293 static Cursor maybeLexIntegerLiteral(Cursor C, MIToken &Token) {
294   if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
295     return None;
296   auto Range = C;
297   C.advance();
298   while (isdigit(C.peek()))
299     C.advance();
300   StringRef StrVal = Range.upto(C);
301   Token = MIToken(MIToken::IntegerLiteral, StrVal, APSInt(StrVal));
302   return C;
303 }
304 
305 static MIToken::TokenKind symbolToken(char C) {
306   switch (C) {
307   case ',':
308     return MIToken::comma;
309   case '=':
310     return MIToken::equal;
311   case ':':
312     return MIToken::colon;
313   default:
314     return MIToken::Error;
315   }
316 }
317 
318 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
319   auto Kind = symbolToken(C.peek());
320   if (Kind == MIToken::Error)
321     return None;
322   auto Range = C;
323   C.advance();
324   Token = MIToken(Kind, Range.upto(C));
325   return C;
326 }
327 
328 StringRef llvm::lexMIToken(
329     StringRef Source, MIToken &Token,
330     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
331   auto C = skipWhitespace(Cursor(Source));
332   if (C.isEOF()) {
333     Token = MIToken(MIToken::Eof, C.remaining());
334     return C.remaining();
335   }
336 
337   if (Cursor R = maybeLexIdentifier(C, Token))
338     return R.remaining();
339   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
340     return R.remaining();
341   if (Cursor R = maybeLexJumpTableIndex(C, Token))
342     return R.remaining();
343   if (Cursor R = maybeLexStackObject(C, Token))
344     return R.remaining();
345   if (Cursor R = maybeLexFixedStackObject(C, Token))
346     return R.remaining();
347   if (Cursor R = maybeLexConstantPoolItem(C, Token))
348     return R.remaining();
349   if (Cursor R = maybeLexRegister(C, Token))
350     return R.remaining();
351   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
352     return R.remaining();
353   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
354     return R.remaining();
355   if (Cursor R = maybeLexIntegerLiteral(C, Token))
356     return R.remaining();
357   if (Cursor R = maybeLexSymbol(C, Token))
358     return R.remaining();
359 
360   Token = MIToken(MIToken::Error, C.remaining());
361   ErrorCallback(C.location(),
362                 Twine("unexpected character '") + Twine(C.peek()) + "'");
363   return C.remaining();
364 }
365