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 Cursor lexName(
118     Cursor C, MIToken &Token, MIToken::TokenKind Type,
119     MIToken::TokenKind QuotedType, unsigned PrefixLength,
120     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
121   auto Range = C;
122   C.advance(PrefixLength);
123   if (C.peek() == '"') {
124     if (Cursor R = lexStringConstant(C, ErrorCallback)) {
125       Token = MIToken(QuotedType, Range.upto(R), PrefixLength);
126       return R;
127     }
128     Token = MIToken(MIToken::Error, Range.remaining());
129     return Range;
130   }
131   while (isIdentifierChar(C.peek()))
132     C.advance();
133   Token = MIToken(Type, Range.upto(C), PrefixLength);
134   return C;
135 }
136 
137 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
138   return StringSwitch<MIToken::TokenKind>(Identifier)
139       .Case("_", MIToken::underscore)
140       .Case("implicit", MIToken::kw_implicit)
141       .Case("implicit-def", MIToken::kw_implicit_define)
142       .Case("dead", MIToken::kw_dead)
143       .Case("killed", MIToken::kw_killed)
144       .Case("undef", MIToken::kw_undef)
145       .Case("frame-setup", MIToken::kw_frame_setup)
146       .Case("debug-location", MIToken::kw_debug_location)
147       .Case(".cfi_offset", MIToken::kw_cfi_offset)
148       .Case(".cfi_def_cfa_register", MIToken::kw_cfi_def_cfa_register)
149       .Case(".cfi_def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
150       .Case(".cfi_def_cfa", MIToken::kw_cfi_def_cfa)
151       .Case("blockaddress", MIToken::kw_blockaddress)
152       .Case("target-index", MIToken::kw_target_index)
153       .Default(MIToken::Identifier);
154 }
155 
156 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
157   if (!isalpha(C.peek()) && C.peek() != '_' && C.peek() != '.')
158     return None;
159   auto Range = C;
160   while (isIdentifierChar(C.peek()))
161     C.advance();
162   auto Identifier = Range.upto(C);
163   Token = MIToken(getIdentifierKind(Identifier), Identifier);
164   return C;
165 }
166 
167 static Cursor maybeLexMachineBasicBlock(
168     Cursor C, MIToken &Token,
169     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
170   if (!C.remaining().startswith("%bb."))
171     return None;
172   auto Range = C;
173   C.advance(4); // Skip '%bb.'
174   if (!isdigit(C.peek())) {
175     Token = MIToken(MIToken::Error, C.remaining());
176     ErrorCallback(C.location(), "expected a number after '%bb.'");
177     return C;
178   }
179   auto NumberRange = C;
180   while (isdigit(C.peek()))
181     C.advance();
182   StringRef Number = NumberRange.upto(C);
183   unsigned StringOffset = 4 + Number.size(); // Drop '%bb.<id>'
184   if (C.peek() == '.') {
185     C.advance(); // Skip '.'
186     ++StringOffset;
187     while (isIdentifierChar(C.peek()))
188       C.advance();
189   }
190   Token = MIToken(MIToken::MachineBasicBlock, Range.upto(C), APSInt(Number),
191                   StringOffset);
192   return C;
193 }
194 
195 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
196                             MIToken::TokenKind Kind) {
197   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
198     return None;
199   auto Range = C;
200   C.advance(Rule.size());
201   auto NumberRange = C;
202   while (isdigit(C.peek()))
203     C.advance();
204   Token = MIToken(Kind, Range.upto(C), APSInt(NumberRange.upto(C)));
205   return C;
206 }
207 
208 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
209                                    MIToken::TokenKind Kind) {
210   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
211     return None;
212   auto Range = C;
213   C.advance(Rule.size());
214   auto NumberRange = C;
215   while (isdigit(C.peek()))
216     C.advance();
217   StringRef Number = NumberRange.upto(C);
218   unsigned StringOffset = Rule.size() + Number.size();
219   if (C.peek() == '.') {
220     C.advance();
221     ++StringOffset;
222     while (isIdentifierChar(C.peek()))
223       C.advance();
224   }
225   Token = MIToken(Kind, Range.upto(C), APSInt(Number), StringOffset);
226   return C;
227 }
228 
229 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
230   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
231 }
232 
233 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
234   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
235 }
236 
237 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
238   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
239 }
240 
241 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
242   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
243 }
244 
245 static Cursor maybeLexIRBlock(
246     Cursor C, MIToken &Token,
247     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
248   const StringRef Rule = "%ir-block.";
249   if (!C.remaining().startswith(Rule))
250     return None;
251   if (isdigit(C.peek(Rule.size())))
252     return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
253   return lexName(C, Token, MIToken::NamedIRBlock, MIToken::QuotedNamedIRBlock,
254                  Rule.size(), ErrorCallback);
255 }
256 
257 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
258   auto Range = C;
259   C.advance(); // Skip '%'
260   auto NumberRange = C;
261   while (isdigit(C.peek()))
262     C.advance();
263   Token = MIToken(MIToken::VirtualRegister, Range.upto(C),
264                   APSInt(NumberRange.upto(C)));
265   return C;
266 }
267 
268 static Cursor maybeLexRegister(Cursor C, MIToken &Token) {
269   if (C.peek() != '%')
270     return None;
271   if (isdigit(C.peek(1)))
272     return lexVirtualRegister(C, Token);
273   auto Range = C;
274   C.advance(); // Skip '%'
275   while (isIdentifierChar(C.peek()))
276     C.advance();
277   Token = MIToken(MIToken::NamedRegister, Range.upto(C),
278                   /*StringOffset=*/1); // Drop the '%'
279   return C;
280 }
281 
282 static Cursor maybeLexGlobalValue(
283     Cursor C, MIToken &Token,
284     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
285   if (C.peek() != '@')
286     return None;
287   if (!isdigit(C.peek(1)))
288     return lexName(C, Token, MIToken::NamedGlobalValue,
289                    MIToken::QuotedNamedGlobalValue, /*PrefixLength=*/1,
290                    ErrorCallback);
291   auto Range = C;
292   C.advance(1); // Skip the '@'
293   auto NumberRange = C;
294   while (isdigit(C.peek()))
295     C.advance();
296   Token =
297       MIToken(MIToken::GlobalValue, Range.upto(C), APSInt(NumberRange.upto(C)));
298   return C;
299 }
300 
301 static Cursor maybeLexExternalSymbol(
302     Cursor C, MIToken &Token,
303     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
304   if (C.peek() != '$')
305     return None;
306   return lexName(C, Token, MIToken::ExternalSymbol,
307                  MIToken::QuotedExternalSymbol,
308                  /*PrefixLength=*/1, ErrorCallback);
309 }
310 
311 static Cursor maybeLexIntegerLiteral(Cursor C, MIToken &Token) {
312   if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
313     return None;
314   auto Range = C;
315   C.advance();
316   while (isdigit(C.peek()))
317     C.advance();
318   StringRef StrVal = Range.upto(C);
319   Token = MIToken(MIToken::IntegerLiteral, StrVal, APSInt(StrVal));
320   return C;
321 }
322 
323 static MIToken::TokenKind symbolToken(char C) {
324   switch (C) {
325   case ',':
326     return MIToken::comma;
327   case '=':
328     return MIToken::equal;
329   case ':':
330     return MIToken::colon;
331   case '!':
332     return MIToken::exclaim;
333   case '(':
334     return MIToken::lparen;
335   case ')':
336     return MIToken::rparen;
337   default:
338     return MIToken::Error;
339   }
340 }
341 
342 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
343   auto Kind = symbolToken(C.peek());
344   if (Kind == MIToken::Error)
345     return None;
346   auto Range = C;
347   C.advance();
348   Token = MIToken(Kind, Range.upto(C));
349   return C;
350 }
351 
352 StringRef llvm::lexMIToken(
353     StringRef Source, MIToken &Token,
354     function_ref<void(StringRef::iterator Loc, const Twine &)> ErrorCallback) {
355   auto C = skipWhitespace(Cursor(Source));
356   if (C.isEOF()) {
357     Token = MIToken(MIToken::Eof, C.remaining());
358     return C.remaining();
359   }
360 
361   if (Cursor R = maybeLexIdentifier(C, Token))
362     return R.remaining();
363   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
364     return R.remaining();
365   if (Cursor R = maybeLexJumpTableIndex(C, Token))
366     return R.remaining();
367   if (Cursor R = maybeLexStackObject(C, Token))
368     return R.remaining();
369   if (Cursor R = maybeLexFixedStackObject(C, Token))
370     return R.remaining();
371   if (Cursor R = maybeLexConstantPoolItem(C, Token))
372     return R.remaining();
373   if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
374     return R.remaining();
375   if (Cursor R = maybeLexRegister(C, Token))
376     return R.remaining();
377   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
378     return R.remaining();
379   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
380     return R.remaining();
381   if (Cursor R = maybeLexIntegerLiteral(C, Token))
382     return R.remaining();
383   if (Cursor R = maybeLexSymbol(C, Token))
384     return R.remaining();
385 
386   Token = MIToken(MIToken::Error, C.remaining());
387   ErrorCallback(C.location(),
388                 Twine("unexpected character '") + Twine(C.peek()) + "'");
389   return C.remaining();
390 }
391