lib/CodeGen/MIRParser/MILexer.cpp

   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/APSInt.h"
  16 #include "llvm/ADT/None.h"
  17 #include "llvm/ADT/STLExtras.h"
  18 #include "llvm/ADT/StringExtras.h"
  19 #include "llvm/ADT/StringSwitch.h"
  20 #include "llvm/ADT/StringRef.h"
  21 #include "llvm/ADT/Twine.h"
  22 #include <algorithm>
  23 #include <cassert>
  24 #include <cctype>
  25 #include <string>
  26
  27 using namespace llvm;
  28
  29 namespace {
  30
  31 using ErrorCallbackType =
  32     function_ref<void(StringRef::iterator Loc, const Twine &)>;
  33
  34 /// This class provides a way to iterate and get characters from the source
  35 /// string.
  36 class Cursor {
  37   const char *Ptr = nullptr;
  38   const char *End = nullptr;
  39
  40 public:
  41   Cursor(NoneType) {}
  42
  43   explicit Cursor(StringRef Str) {
  44     Ptr = Str.data();
  45     End = Ptr + Str.size();
  46   }
  47
  48   bool isEOF() const { return Ptr == End; }
  49
  50   char peek(int I = 0) const { return End - Ptr <= I ? 0 : Ptr[I]; }
  51
  52   void advance(unsigned I = 1) { Ptr += I; }
  53
  54   StringRef remaining() const { return StringRef(Ptr, End - Ptr); }
  55
  56   StringRef upto(Cursor C) const {
  57     assert(C.Ptr >= Ptr && C.Ptr <= End);
  58     return StringRef(Ptr, C.Ptr - Ptr);
  59   }
  60
  61   StringRef::iterator location() const { return Ptr; }
  62
  63   operator bool() const { return Ptr != nullptr; }
  64 };
  65
  66 } // end anonymous namespace
  67
  68 MIToken &MIToken::reset(TokenKind Kind, StringRef Range) {
  69   this->Kind = Kind;
  70   this->Range = Range;
  71   return *this;
  72 }
  73
  74 MIToken &MIToken::setStringValue(StringRef StrVal) {
  75   StringValue = StrVal;
  76   return *this;
  77 }
  78
  79 MIToken &MIToken::setOwnedStringValue(std::string StrVal) {
  80   StringValueStorage = std::move(StrVal);
  81   StringValue = StringValueStorage;
  82   return *this;
  83 }
  84
  85 MIToken &MIToken::setIntegerValue(APSInt IntVal) {
  86   this->IntVal = std::move(IntVal);
  87   return *this;
  88 }
  89
  90 /// Skip the leading whitespace characters and return the updated cursor.
  91 static Cursor skipWhitespace(Cursor C) {
  92   while (isblank(C.peek()))
  93     C.advance();
  94   return C;
  95 }
  96
  97 static bool isNewlineChar(char C) { return C == '\n' || C == '\r'; }
  98
  99 /// Skip a line comment and return the updated cursor.
 100 static Cursor skipComment(Cursor C) {
 101   if (C.peek() != ';')
 102     return C;
 103   while (!isNewlineChar(C.peek()) && !C.isEOF())
 104     C.advance();
 105   return C;
 106 }
 107
 108 /// Return true if the given character satisfies the following regular
 109 /// expression: [-a-zA-Z$._0-9]
 110 static bool isIdentifierChar(char C) {
 111   return isalpha(C) || isdigit(C) || C == '_' || C == '-' || C == '.' ||
 112          C == '$';
 113 }
 114
 115 /// Unescapes the given string value.
 116 ///
 117 /// Expects the string value to be quoted.
 118 static std::string unescapeQuotedString(StringRef Value) {
 119   assert(Value.front() == '"' && Value.back() == '"');
 120   Cursor C = Cursor(Value.substr(1, Value.size() - 2));
 121
 122   std::string Str;
 123   Str.reserve(C.remaining().size());
 124   while (!C.isEOF()) {
 125     char Char = C.peek();
 126     if (Char == '\\') {
 127       if (C.peek(1) == '\\') {
 128         // Two '\' become one
 129         Str += '\\';
 130         C.advance(2);
 131         continue;
 132       }
 133       if (isxdigit(C.peek(1)) && isxdigit(C.peek(2))) {
 134         Str += hexDigitValue(C.peek(1)) * 16 + hexDigitValue(C.peek(2));
 135         C.advance(3);
 136         continue;
 137       }
 138     }
 139     Str += Char;
 140     C.advance();
 141   }
 142   return Str;
 143 }
 144
 145 /// Lex a string constant using the following regular expression: \"[^\"]*\"
 146 static Cursor lexStringConstant(Cursor C, ErrorCallbackType ErrorCallback) {
 147   assert(C.peek() == '"');
 148   for (C.advance(); C.peek() != '"'; C.advance()) {
 149     if (C.isEOF() || isNewlineChar(C.peek())) {
 150       ErrorCallback(
 151           C.location(),
 152           "end of machine instruction reached before the closing '\"'");
 153       return None;
 154     }
 155   }
 156   C.advance();
 157   return C;
 158 }
 159
 160 static Cursor lexName(Cursor C, MIToken &Token, MIToken::TokenKind Type,
 161                       unsigned PrefixLength, ErrorCallbackType ErrorCallback) {
 162   auto Range = C;
 163   C.advance(PrefixLength);
 164   if (C.peek() == '"') {
 165     if (Cursor R = lexStringConstant(C, ErrorCallback)) {
 166       StringRef String = Range.upto(R);
 167       Token.reset(Type, String)
 168           .setOwnedStringValue(
 169               unescapeQuotedString(String.drop_front(PrefixLength)));
 170       return R;
 171     }
 172     Token.reset(MIToken::Error, Range.remaining());
 173     return Range;
 174   }
 175   while (isIdentifierChar(C.peek()))
 176     C.advance();
 177   Token.reset(Type, Range.upto(C))
 178       .setStringValue(Range.upto(C).drop_front(PrefixLength));
 179   return C;
 180 }
 181
 182 static MIToken::TokenKind getIdentifierKind(StringRef Identifier) {
 183   return StringSwitch<MIToken::TokenKind>(Identifier)
 184       .Case("_", MIToken::underscore)
 185       .Case("implicit", MIToken::kw_implicit)
 186       .Case("implicit-def", MIToken::kw_implicit_define)
 187       .Case("def", MIToken::kw_def)
 188       .Case("dead", MIToken::kw_dead)
 189       .Case("killed", MIToken::kw_killed)
 190       .Case("undef", MIToken::kw_undef)
 191       .Case("internal", MIToken::kw_internal)
 192       .Case("early-clobber", MIToken::kw_early_clobber)
 193       .Case("debug-use", MIToken::kw_debug_use)
 194       .Case("renamable", MIToken::kw_renamable)
 195       .Case("tied-def", MIToken::kw_tied_def)
 196       .Case("frame-setup", MIToken::kw_frame_setup)
 197       .Case("frame-destroy", MIToken::kw_frame_destroy)
 198       .Case("nnan", MIToken::kw_nnan)
 199       .Case("ninf", MIToken::kw_ninf)
 200       .Case("nsz", MIToken::kw_nsz)
 201       .Case("arcp", MIToken::kw_arcp)
 202       .Case("contract", MIToken::kw_contract)
 203       .Case("afn", MIToken::kw_afn)
 204       .Case("reassoc", MIToken::kw_reassoc)
 205       .Case("nuw" , MIToken::kw_nuw)
 206       .Case("nsw" , MIToken::kw_nsw)
 207       .Case("exact" , MIToken::kw_exact)
 208       .Case("debug-location", MIToken::kw_debug_location)
 209       .Case("same_value", MIToken::kw_cfi_same_value)
 210       .Case("offset", MIToken::kw_cfi_offset)
 211       .Case("rel_offset", MIToken::kw_cfi_rel_offset)
 212       .Case("def_cfa_register", MIToken::kw_cfi_def_cfa_register)
 213       .Case("def_cfa_offset", MIToken::kw_cfi_def_cfa_offset)
 214       .Case("adjust_cfa_offset", MIToken::kw_cfi_adjust_cfa_offset)
 215       .Case("escape", MIToken::kw_cfi_escape)
 216       .Case("def_cfa", MIToken::kw_cfi_def_cfa)
 217       .Case("remember_state", MIToken::kw_cfi_remember_state)
 218       .Case("restore", MIToken::kw_cfi_restore)
 219       .Case("restore_state", MIToken::kw_cfi_restore_state)
 220       .Case("undefined", MIToken::kw_cfi_undefined)
 221       .Case("register", MIToken::kw_cfi_register)
 222       .Case("window_save", MIToken::kw_cfi_window_save)
 223       .Case("blockaddress", MIToken::kw_blockaddress)
 224       .Case("intrinsic", MIToken::kw_intrinsic)
 225       .Case("target-index", MIToken::kw_target_index)
 226       .Case("half", MIToken::kw_half)
 227       .Case("float", MIToken::kw_float)
 228       .Case("double", MIToken::kw_double)
 229       .Case("x86_fp80", MIToken::kw_x86_fp80)
 230       .Case("fp128", MIToken::kw_fp128)
 231       .Case("ppc_fp128", MIToken::kw_ppc_fp128)
 232       .Case("target-flags", MIToken::kw_target_flags)
 233       .Case("volatile", MIToken::kw_volatile)
 234       .Case("non-temporal", MIToken::kw_non_temporal)
 235       .Case("dereferenceable", MIToken::kw_dereferenceable)
 236       .Case("invariant", MIToken::kw_invariant)
 237       .Case("align", MIToken::kw_align)
 238       .Case("addrspace", MIToken::kw_addrspace)
 239       .Case("stack", MIToken::kw_stack)
 240       .Case("got", MIToken::kw_got)
 241       .Case("jump-table", MIToken::kw_jump_table)
 242       .Case("constant-pool", MIToken::kw_constant_pool)
 243       .Case("call-entry", MIToken::kw_call_entry)
 244       .Case("liveout", MIToken::kw_liveout)
 245       .Case("address-taken", MIToken::kw_address_taken)
 246       .Case("landing-pad", MIToken::kw_landing_pad)
 247       .Case("liveins", MIToken::kw_liveins)
 248       .Case("successors", MIToken::kw_successors)
 249       .Case("floatpred", MIToken::kw_floatpred)
 250       .Case("intpred", MIToken::kw_intpred)
 251       .Case("pre-instr-symbol", MIToken::kw_pre_instr_symbol)
 252       .Case("post-instr-symbol", MIToken::kw_post_instr_symbol)
 253       .Case("unknown-size", MIToken::kw_unknown_size)
 254       .Default(MIToken::Identifier);
 255 }
 256
 257 static Cursor maybeLexIdentifier(Cursor C, MIToken &Token) {
 258   if (!isalpha(C.peek()) && C.peek() != '_')
 259     return None;
 260   auto Range = C;
 261   while (isIdentifierChar(C.peek()))
 262     C.advance();
 263   auto Identifier = Range.upto(C);
 264   Token.reset(getIdentifierKind(Identifier), Identifier)
 265       .setStringValue(Identifier);
 266   return C;
 267 }
 268
 269 static Cursor maybeLexMachineBasicBlock(Cursor C, MIToken &Token,
 270                                         ErrorCallbackType ErrorCallback) {
 271   bool IsReference = C.remaining().startswith("%bb.");
 272   if (!IsReference && !C.remaining().startswith("bb."))
 273     return None;
 274   auto Range = C;
 275   unsigned PrefixLength = IsReference ? 4 : 3;
 276   C.advance(PrefixLength); // Skip '%bb.' or 'bb.'
 277   if (!isdigit(C.peek())) {
 278     Token.reset(MIToken::Error, C.remaining());
 279     ErrorCallback(C.location(), "expected a number after '%bb.'");
 280     return C;
 281   }
 282   auto NumberRange = C;
 283   while (isdigit(C.peek()))
 284     C.advance();
 285   StringRef Number = NumberRange.upto(C);
 286   unsigned StringOffset = PrefixLength + Number.size(); // Drop '%bb.<id>'
 287   // TODO: The format bb.<id>.<irname> is supported only when it's not a
 288   // reference. Once we deprecate the format where the irname shows up, we
 289   // should only lex forward if it is a reference.
 290   if (C.peek() == '.') {
 291     C.advance(); // Skip '.'
 292     ++StringOffset;
 293     while (isIdentifierChar(C.peek()))
 294       C.advance();
 295   }
 296   Token.reset(IsReference ? MIToken::MachineBasicBlock
 297                           : MIToken::MachineBasicBlockLabel,
 298               Range.upto(C))
 299       .setIntegerValue(APSInt(Number))
 300       .setStringValue(Range.upto(C).drop_front(StringOffset));
 301   return C;
 302 }
 303
 304 static Cursor maybeLexIndex(Cursor C, MIToken &Token, StringRef Rule,
 305                             MIToken::TokenKind Kind) {
 306   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
 307     return None;
 308   auto Range = C;
 309   C.advance(Rule.size());
 310   auto NumberRange = C;
 311   while (isdigit(C.peek()))
 312     C.advance();
 313   Token.reset(Kind, Range.upto(C)).setIntegerValue(APSInt(NumberRange.upto(C)));
 314   return C;
 315 }
 316
 317 static Cursor maybeLexIndexAndName(Cursor C, MIToken &Token, StringRef Rule,
 318                                    MIToken::TokenKind Kind) {
 319   if (!C.remaining().startswith(Rule) || !isdigit(C.peek(Rule.size())))
 320     return None;
 321   auto Range = C;
 322   C.advance(Rule.size());
 323   auto NumberRange = C;
 324   while (isdigit(C.peek()))
 325     C.advance();
 326   StringRef Number = NumberRange.upto(C);
 327   unsigned StringOffset = Rule.size() + Number.size();
 328   if (C.peek() == '.') {
 329     C.advance();
 330     ++StringOffset;
 331     while (isIdentifierChar(C.peek()))
 332       C.advance();
 333   }
 334   Token.reset(Kind, Range.upto(C))
 335       .setIntegerValue(APSInt(Number))
 336       .setStringValue(Range.upto(C).drop_front(StringOffset));
 337   return C;
 338 }
 339
 340 static Cursor maybeLexJumpTableIndex(Cursor C, MIToken &Token) {
 341   return maybeLexIndex(C, Token, "%jump-table.", MIToken::JumpTableIndex);
 342 }
 343
 344 static Cursor maybeLexStackObject(Cursor C, MIToken &Token) {
 345   return maybeLexIndexAndName(C, Token, "%stack.", MIToken::StackObject);
 346 }
 347
 348 static Cursor maybeLexFixedStackObject(Cursor C, MIToken &Token) {
 349   return maybeLexIndex(C, Token, "%fixed-stack.", MIToken::FixedStackObject);
 350 }
 351
 352 static Cursor maybeLexConstantPoolItem(Cursor C, MIToken &Token) {
 353   return maybeLexIndex(C, Token, "%const.", MIToken::ConstantPoolItem);
 354 }
 355
 356 static Cursor maybeLexSubRegisterIndex(Cursor C, MIToken &Token,
 357                                        ErrorCallbackType ErrorCallback) {
 358   const StringRef Rule = "%subreg.";
 359   if (!C.remaining().startswith(Rule))
 360     return None;
 361   return lexName(C, Token, MIToken::SubRegisterIndex, Rule.size(),
 362                  ErrorCallback);
 363 }
 364
 365 static Cursor maybeLexIRBlock(Cursor C, MIToken &Token,
 366                               ErrorCallbackType ErrorCallback) {
 367   const StringRef Rule = "%ir-block.";
 368   if (!C.remaining().startswith(Rule))
 369     return None;
 370   if (isdigit(C.peek(Rule.size())))
 371     return maybeLexIndex(C, Token, Rule, MIToken::IRBlock);
 372   return lexName(C, Token, MIToken::NamedIRBlock, Rule.size(), ErrorCallback);
 373 }
 374
 375 static Cursor maybeLexIRValue(Cursor C, MIToken &Token,
 376                               ErrorCallbackType ErrorCallback) {
 377   const StringRef Rule = "%ir.";
 378   if (!C.remaining().startswith(Rule))
 379     return None;
 380   if (isdigit(C.peek(Rule.size())))
 381     return maybeLexIndex(C, Token, Rule, MIToken::IRValue);
 382   return lexName(C, Token, MIToken::NamedIRValue, Rule.size(), ErrorCallback);
 383 }
 384
 385 static Cursor maybeLexStringConstant(Cursor C, MIToken &Token,
 386                                      ErrorCallbackType ErrorCallback) {
 387   if (C.peek() != '"')
 388     return None;
 389   return lexName(C, Token, MIToken::StringConstant, /*PrefixLength=*/0,
 390                  ErrorCallback);
 391 }
 392
 393 static Cursor lexVirtualRegister(Cursor C, MIToken &Token) {
 394   auto Range = C;
 395   C.advance(); // Skip '%'
 396   auto NumberRange = C;
 397   while (isdigit(C.peek()))
 398     C.advance();
 399   Token.reset(MIToken::VirtualRegister, Range.upto(C))
 400       .setIntegerValue(APSInt(NumberRange.upto(C)));
 401   return C;
 402 }
 403
 404 /// Returns true for a character allowed in a register name.
 405 static bool isRegisterChar(char C) {
 406   return isIdentifierChar(C) && C != '.';
 407 }
 408
 409 static Cursor lexNamedVirtualRegister(Cursor C, MIToken &Token) {
 410   Cursor Range = C;
 411   C.advance(); // Skip '%'
 412   while (isRegisterChar(C.peek()))
 413     C.advance();
 414   Token.reset(MIToken::NamedVirtualRegister, Range.upto(C))
 415       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '%'
 416   return C;
 417 }
 418
 419 static Cursor maybeLexRegister(Cursor C, MIToken &Token,
 420                                ErrorCallbackType ErrorCallback) {
 421   if (C.peek() != '%' && C.peek() != '$')
 422     return None;
 423
 424   if (C.peek() == '%') {
 425     if (isdigit(C.peek(1)))
 426       return lexVirtualRegister(C, Token);
 427
 428     if (isRegisterChar(C.peek(1)))
 429       return lexNamedVirtualRegister(C, Token);
 430
 431     return None;
 432   }
 433
 434   assert(C.peek() == '$');
 435   auto Range = C;
 436   C.advance(); // Skip '$'
 437   while (isRegisterChar(C.peek()))
 438     C.advance();
 439   Token.reset(MIToken::NamedRegister, Range.upto(C))
 440       .setStringValue(Range.upto(C).drop_front(1)); // Drop the '$'
 441   return C;
 442 }
 443
 444 static Cursor maybeLexGlobalValue(Cursor C, MIToken &Token,
 445                                   ErrorCallbackType ErrorCallback) {
 446   if (C.peek() != '@')
 447     return None;
 448   if (!isdigit(C.peek(1)))
 449     return lexName(C, Token, MIToken::NamedGlobalValue, /*PrefixLength=*/1,
 450                    ErrorCallback);
 451   auto Range = C;
 452   C.advance(1); // Skip the '@'
 453   auto NumberRange = C;
 454   while (isdigit(C.peek()))
 455     C.advance();
 456   Token.reset(MIToken::GlobalValue, Range.upto(C))
 457       .setIntegerValue(APSInt(NumberRange.upto(C)));
 458   return C;
 459 }
 460
 461 static Cursor maybeLexExternalSymbol(Cursor C, MIToken &Token,
 462                                      ErrorCallbackType ErrorCallback) {
 463   if (C.peek() != '&')
 464     return None;
 465   return lexName(C, Token, MIToken::ExternalSymbol, /*PrefixLength=*/1,
 466                  ErrorCallback);
 467 }
 468
 469 static Cursor maybeLexMCSymbol(Cursor C, MIToken &Token,
 470                                ErrorCallbackType ErrorCallback) {
 471   const StringRef Rule = "<mcsymbol ";
 472   if (!C.remaining().startswith(Rule))
 473     return None;
 474   auto Start = C;
 475   C.advance(Rule.size());
 476
 477   // Try a simple unquoted name.
 478   if (C.peek() != '"') {
 479     while (isIdentifierChar(C.peek()))
 480       C.advance();
 481     StringRef String = Start.upto(C).drop_front(Rule.size());
 482     if (C.peek() != '>') {
 483       ErrorCallback(C.location(),
 484                     "expected the '<mcsymbol ...' to be closed by a '>'");
 485       Token.reset(MIToken::Error, Start.remaining());
 486       return Start;
 487     }
 488     C.advance();
 489
 490     Token.reset(MIToken::MCSymbol, Start.upto(C)).setStringValue(String);
 491     return C;
 492   }
 493
 494   // Otherwise lex out a quoted name.
 495   Cursor R = lexStringConstant(C, ErrorCallback);
 496   if (!R) {
 497     ErrorCallback(C.location(),
 498                   "unable to parse quoted string from opening quote");
 499     Token.reset(MIToken::Error, Start.remaining());
 500     return Start;
 501   }
 502   StringRef String = Start.upto(R).drop_front(Rule.size());
 503   if (R.peek() != '>') {
 504     ErrorCallback(R.location(),
 505                   "expected the '<mcsymbol ...' to be closed by a '>'");
 506     Token.reset(MIToken::Error, Start.remaining());
 507     return Start;
 508   }
 509   R.advance();
 510
 511   Token.reset(MIToken::MCSymbol, Start.upto(R))
 512       .setOwnedStringValue(unescapeQuotedString(String));
 513   return R;
 514 }
 515
 516 static bool isValidHexFloatingPointPrefix(char C) {
 517   return C == 'H' || C == 'K' || C == 'L' || C == 'M';
 518 }
 519
 520 static Cursor lexFloatingPointLiteral(Cursor Range, Cursor C, MIToken &Token) {
 521   C.advance();
 522   // Skip over [0-9]*([eE][-+]?[0-9]+)?
 523   while (isdigit(C.peek()))
 524     C.advance();
 525   if ((C.peek() == 'e' || C.peek() == 'E') &&
 526       (isdigit(C.peek(1)) ||
 527        ((C.peek(1) == '-' || C.peek(1) == '+') && isdigit(C.peek(2))))) {
 528     C.advance(2);
 529     while (isdigit(C.peek()))
 530       C.advance();
 531   }
 532   Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
 533   return C;
 534 }
 535
 536 static Cursor maybeLexHexadecimalLiteral(Cursor C, MIToken &Token) {
 537   if (C.peek() != '0' || (C.peek(1) != 'x' && C.peek(1) != 'X'))
 538     return None;
 539   Cursor Range = C;
 540   C.advance(2);
 541   unsigned PrefLen = 2;
 542   if (isValidHexFloatingPointPrefix(C.peek())) {
 543     C.advance();
 544     PrefLen++;
 545   }
 546   while (isxdigit(C.peek()))
 547     C.advance();
 548   StringRef StrVal = Range.upto(C);
 549   if (StrVal.size() <= PrefLen)
 550     return None;
 551   if (PrefLen == 2)
 552     Token.reset(MIToken::HexLiteral, Range.upto(C));
 553   else // It must be 3, which means that there was a floating-point prefix.
 554     Token.reset(MIToken::FloatingPointLiteral, Range.upto(C));
 555   return C;
 556 }
 557
 558 static Cursor maybeLexNumericalLiteral(Cursor C, MIToken &Token) {
 559   if (!isdigit(C.peek()) && (C.peek() != '-' || !isdigit(C.peek(1))))
 560     return None;
 561   auto Range = C;
 562   C.advance();
 563   while (isdigit(C.peek()))
 564     C.advance();
 565   if (C.peek() == '.')
 566     return lexFloatingPointLiteral(Range, C, Token);
 567   StringRef StrVal = Range.upto(C);
 568   Token.reset(MIToken::IntegerLiteral, StrVal).setIntegerValue(APSInt(StrVal));
 569   return C;
 570 }
 571
 572 static MIToken::TokenKind getMetadataKeywordKind(StringRef Identifier) {
 573   return StringSwitch<MIToken::TokenKind>(Identifier)
 574       .Case("!tbaa", MIToken::md_tbaa)
 575       .Case("!alias.scope", MIToken::md_alias_scope)
 576       .Case("!noalias", MIToken::md_noalias)
 577       .Case("!range", MIToken::md_range)
 578       .Case("!DIExpression", MIToken::md_diexpr)
 579       .Default(MIToken::Error);
 580 }
 581
 582 static Cursor maybeLexExlaim(Cursor C, MIToken &Token,
 583                              ErrorCallbackType ErrorCallback) {
 584   if (C.peek() != '!')
 585     return None;
 586   auto Range = C;
 587   C.advance(1);
 588   if (isdigit(C.peek()) || !isIdentifierChar(C.peek())) {
 589     Token.reset(MIToken::exclaim, Range.upto(C));
 590     return C;
 591   }
 592   while (isIdentifierChar(C.peek()))
 593     C.advance();
 594   StringRef StrVal = Range.upto(C);
 595   Token.reset(getMetadataKeywordKind(StrVal), StrVal);
 596   if (Token.isError())
 597     ErrorCallback(Token.location(),
 598                   "use of unknown metadata keyword '" + StrVal + "'");
 599   return C;
 600 }
 601
 602 static MIToken::TokenKind symbolToken(char C) {
 603   switch (C) {
 604   case ',':
 605     return MIToken::comma;
 606   case '.':
 607     return MIToken::dot;
 608   case '=':
 609     return MIToken::equal;
 610   case ':':
 611     return MIToken::colon;
 612   case '(':
 613     return MIToken::lparen;
 614   case ')':
 615     return MIToken::rparen;
 616   case '{':
 617     return MIToken::lbrace;
 618   case '}':
 619     return MIToken::rbrace;
 620   case '+':
 621     return MIToken::plus;
 622   case '-':
 623     return MIToken::minus;
 624   case '<':
 625     return MIToken::less;
 626   case '>':
 627     return MIToken::greater;
 628   default:
 629     return MIToken::Error;
 630   }
 631 }
 632
 633 static Cursor maybeLexSymbol(Cursor C, MIToken &Token) {
 634   MIToken::TokenKind Kind;
 635   unsigned Length = 1;
 636   if (C.peek() == ':' && C.peek(1) == ':') {
 637     Kind = MIToken::coloncolon;
 638     Length = 2;
 639   } else
 640     Kind = symbolToken(C.peek());
 641   if (Kind == MIToken::Error)
 642     return None;
 643   auto Range = C;
 644   C.advance(Length);
 645   Token.reset(Kind, Range.upto(C));
 646   return C;
 647 }
 648
 649 static Cursor maybeLexNewline(Cursor C, MIToken &Token) {
 650   if (!isNewlineChar(C.peek()))
 651     return None;
 652   auto Range = C;
 653   C.advance();
 654   Token.reset(MIToken::Newline, Range.upto(C));
 655   return C;
 656 }
 657
 658 static Cursor maybeLexEscapedIRValue(Cursor C, MIToken &Token,
 659                                      ErrorCallbackType ErrorCallback) {
 660   if (C.peek() != '`')
 661     return None;
 662   auto Range = C;
 663   C.advance();
 664   auto StrRange = C;
 665   while (C.peek() != '`') {
 666     if (C.isEOF() || isNewlineChar(C.peek())) {
 667       ErrorCallback(
 668           C.location(),
 669           "end of machine instruction reached before the closing '`'");
 670       Token.reset(MIToken::Error, Range.remaining());
 671       return C;
 672     }
 673     C.advance();
 674   }
 675   StringRef Value = StrRange.upto(C);
 676   C.advance();
 677   Token.reset(MIToken::QuotedIRValue, Range.upto(C)).setStringValue(Value);
 678   return C;
 679 }
 680
 681 StringRef llvm::lexMIToken(StringRef Source, MIToken &Token,
 682                            ErrorCallbackType ErrorCallback) {
 683   auto C = skipComment(skipWhitespace(Cursor(Source)));
 684   if (C.isEOF()) {
 685     Token.reset(MIToken::Eof, C.remaining());
 686     return C.remaining();
 687   }
 688
 689   if (Cursor R = maybeLexMachineBasicBlock(C, Token, ErrorCallback))
 690     return R.remaining();
 691   if (Cursor R = maybeLexIdentifier(C, Token))
 692     return R.remaining();
 693   if (Cursor R = maybeLexJumpTableIndex(C, Token))
 694     return R.remaining();
 695   if (Cursor R = maybeLexStackObject(C, Token))
 696     return R.remaining();
 697   if (Cursor R = maybeLexFixedStackObject(C, Token))
 698     return R.remaining();
 699   if (Cursor R = maybeLexConstantPoolItem(C, Token))
 700     return R.remaining();
 701   if (Cursor R = maybeLexSubRegisterIndex(C, Token, ErrorCallback))
 702     return R.remaining();
 703   if (Cursor R = maybeLexIRBlock(C, Token, ErrorCallback))
 704     return R.remaining();
 705   if (Cursor R = maybeLexIRValue(C, Token, ErrorCallback))
 706     return R.remaining();
 707   if (Cursor R = maybeLexRegister(C, Token, ErrorCallback))
 708     return R.remaining();
 709   if (Cursor R = maybeLexGlobalValue(C, Token, ErrorCallback))
 710     return R.remaining();
 711   if (Cursor R = maybeLexExternalSymbol(C, Token, ErrorCallback))
 712     return R.remaining();
 713   if (Cursor R = maybeLexMCSymbol(C, Token, ErrorCallback))
 714     return R.remaining();
 715   if (Cursor R = maybeLexHexadecimalLiteral(C, Token))
 716     return R.remaining();
 717   if (Cursor R = maybeLexNumericalLiteral(C, Token))
 718     return R.remaining();
 719   if (Cursor R = maybeLexExlaim(C, Token, ErrorCallback))
 720     return R.remaining();
 721   if (Cursor R = maybeLexSymbol(C, Token))
 722     return R.remaining();
 723   if (Cursor R = maybeLexNewline(C, Token))
 724     return R.remaining();
 725   if (Cursor R = maybeLexEscapedIRValue(C, Token, ErrorCallback))
 726     return R.remaining();
 727   if (Cursor R = maybeLexStringConstant(C, Token, ErrorCallback))
 728     return R.remaining();
 729
 730   Token.reset(MIToken::Error, C.remaining());
 731   ErrorCallback(C.location(),
 732                 Twine("unexpected character '") + Twine(C.peek()) + "'");
 733   return C.remaining();
 734 }