clang/lib/Basic/IdentifierTable.cpp

   1 //===- IdentifierTable.cpp - Hash table for identifier lookup -------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8 //
   9 // This file implements the IdentifierInfo, IdentifierVisitor, and
  10 // IdentifierTable interfaces.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "clang/Basic/IdentifierTable.h"
  15 #include "clang/Basic/CharInfo.h"
  16 #include "clang/Basic/DiagnosticLex.h"
  17 #include "clang/Basic/LangOptions.h"
  18 #include "clang/Basic/OperatorKinds.h"
  19 #include "clang/Basic/Specifiers.h"
  20 #include "clang/Basic/TargetBuiltins.h"
  21 #include "clang/Basic/TokenKinds.h"
  22 #include "llvm/ADT/DenseMapInfo.h"
  23 #include "llvm/ADT/FoldingSet.h"
  24 #include "llvm/ADT/SmallString.h"
  25 #include "llvm/ADT/StringMap.h"
  26 #include "llvm/ADT/StringRef.h"
  27 #include "llvm/Support/Allocator.h"
  28 #include "llvm/Support/ErrorHandling.h"
  29 #include "llvm/Support/raw_ostream.h"
  30 #include <cassert>
  31 #include <cstdio>
  32 #include <cstring>
  33 #include <string>
  34
  35 using namespace clang;
  36
  37 // A check to make sure the ObjCOrBuiltinID has sufficient room to store the
  38 // largest possible target/aux-target combination. If we exceed this, we likely
  39 // need to just change the ObjCOrBuiltinIDBits value in IdentifierTable.h.
  40 static_assert(2 * LargestBuiltinID < (2 << (ObjCOrBuiltinIDBits - 1)),
  41               "Insufficient ObjCOrBuiltinID Bits");
  42
  43 //===----------------------------------------------------------------------===//
  44 // IdentifierTable Implementation
  45 //===----------------------------------------------------------------------===//
  46
  47 IdentifierIterator::~IdentifierIterator() = default;
  48
  49 IdentifierInfoLookup::~IdentifierInfoLookup() = default;
  50
  51 namespace {
  52
  53 /// A simple identifier lookup iterator that represents an
  54 /// empty sequence of identifiers.
  55 class EmptyLookupIterator : public IdentifierIterator
  56 {
  57 public:
  58   StringRef Next() override { return StringRef(); }
  59 };
  60
  61 } // namespace
  62
  63 IdentifierIterator *IdentifierInfoLookup::getIdentifiers() {
  64   return new EmptyLookupIterator();
  65 }
  66
  67 IdentifierTable::IdentifierTable(IdentifierInfoLookup *ExternalLookup)
  68     : HashTable(8192), // Start with space for 8K identifiers.
  69       ExternalLookup(ExternalLookup) {}
  70
  71 IdentifierTable::IdentifierTable(const LangOptions &LangOpts,
  72                                  IdentifierInfoLookup *ExternalLookup)
  73     : IdentifierTable(ExternalLookup) {
  74   // Populate the identifier table with info about keywords for the current
  75   // language.
  76   AddKeywords(LangOpts);
  77 }
  78
  79 //===----------------------------------------------------------------------===//
  80 // Language Keyword Implementation
  81 //===----------------------------------------------------------------------===//
  82
  83 // Constants for TokenKinds.def
  84 namespace {
  85
  86   enum TokenKey : unsigned {
  87     KEYC99        = 0x1,
  88     KEYCXX        = 0x2,
  89     KEYCXX11      = 0x4,
  90     KEYGNU        = 0x8,
  91     KEYMS         = 0x10,
  92     BOOLSUPPORT   = 0x20,
  93     KEYALTIVEC    = 0x40,
  94     KEYNOCXX      = 0x80,
  95     KEYBORLAND    = 0x100,
  96     KEYOPENCLC    = 0x200,
  97     KEYC2X        = 0x400,
  98     KEYNOMS18     = 0x800,
  99     KEYNOOPENCL   = 0x1000,
 100     WCHARSUPPORT  = 0x2000,
 101     HALFSUPPORT   = 0x4000,
 102     CHAR8SUPPORT  = 0x8000,
 103     KEYOBJC       = 0x10000,
 104     KEYZVECTOR    = 0x20000,
 105     KEYCOROUTINES = 0x40000,
 106     KEYMODULES    = 0x80000,
 107     KEYCXX20      = 0x100000,
 108     KEYOPENCLCXX  = 0x200000,
 109     KEYMSCOMPAT   = 0x400000,
 110     KEYSYCL       = 0x800000,
 111     KEYCUDA       = 0x1000000,
 112     KEYMAX        = KEYCUDA, // The maximum key
 113     KEYALLCXX = KEYCXX | KEYCXX11 | KEYCXX20,
 114     KEYALL = (KEYMAX | (KEYMAX-1)) & ~KEYNOMS18 &
 115              ~KEYNOOPENCL // KEYNOMS18 and KEYNOOPENCL are used to exclude.
 116   };
 117
 118   /// How a keyword is treated in the selected standard. This enum is ordered
 119   /// intentionally so that the value that 'wins' is the most 'permissive'.
 120   enum KeywordStatus {
 121     KS_Unknown,     // Not yet calculated. Used when figuring out the status.
 122     KS_Disabled,    // Disabled
 123     KS_Future,      // Is a keyword in future standard
 124     KS_Extension,   // Is an extension
 125     KS_Enabled,     // Enabled
 126   };
 127
 128 } // namespace
 129
 130 // This works on a single TokenKey flag and checks the LangOpts to get the
 131 // KeywordStatus based exclusively on this flag, so that it can be merged in
 132 // getKeywordStatus. Most should be enabled/disabled, but some might imply
 133 // 'future' versions, or extensions. Returns 'unknown' unless this is KNOWN to
 134 // be disabled, and the calling function makes it 'disabled' if no other flag
 135 // changes it. This is necessary for the KEYNOCXX and KEYNOOPENCL flags.
 136 static KeywordStatus getKeywordStatusHelper(const LangOptions &LangOpts,
 137                                             TokenKey Flag) {
 138   // Flag is a single bit version of TokenKey (that is, not
 139   // KEYALL/KEYALLCXX/etc), so we can check with == throughout this function.
 140   assert((Flag & ~(Flag - 1)) == Flag && "Multiple bits set?");
 141
 142   switch (Flag) {
 143   case KEYC99:
 144     if (LangOpts.C99)
 145       return KS_Enabled;
 146     return !LangOpts.CPlusPlus ? KS_Future : KS_Unknown;
 147   case KEYC2X:
 148     if (LangOpts.C2x)
 149       return KS_Enabled;
 150     return !LangOpts.CPlusPlus ? KS_Future : KS_Unknown;
 151   case KEYCXX:
 152     return LangOpts.CPlusPlus ? KS_Enabled : KS_Unknown;
 153   case KEYCXX11:
 154     if (LangOpts.CPlusPlus11)
 155       return KS_Enabled;
 156     return LangOpts.CPlusPlus ? KS_Future : KS_Unknown;
 157   case KEYCXX20:
 158     if (LangOpts.CPlusPlus20)
 159       return KS_Enabled;
 160     return LangOpts.CPlusPlus ? KS_Future : KS_Unknown;
 161   case KEYGNU:
 162     return LangOpts.GNUKeywords ? KS_Extension : KS_Unknown;
 163   case KEYMS:
 164     return LangOpts.MicrosoftExt ? KS_Extension : KS_Unknown;
 165   case BOOLSUPPORT:
 166     if (LangOpts.Bool)      return KS_Enabled;
 167     return !LangOpts.CPlusPlus ? KS_Future : KS_Unknown;
 168   case KEYALTIVEC:
 169     return LangOpts.AltiVec ? KS_Enabled : KS_Unknown;
 170   case KEYBORLAND:
 171     return LangOpts.Borland ? KS_Extension : KS_Unknown;
 172   case KEYOPENCLC:
 173     return LangOpts.OpenCL && !LangOpts.OpenCLCPlusPlus ? KS_Enabled
 174                                                         : KS_Unknown;
 175   case WCHARSUPPORT:
 176     return LangOpts.WChar ? KS_Enabled : KS_Unknown;
 177   case HALFSUPPORT:
 178     return LangOpts.Half ? KS_Enabled : KS_Unknown;
 179   case CHAR8SUPPORT:
 180     if (LangOpts.Char8) return KS_Enabled;
 181     if (LangOpts.CPlusPlus20) return KS_Unknown;
 182     if (LangOpts.CPlusPlus) return KS_Future;
 183     return KS_Unknown;
 184   case KEYOBJC:
 185     // We treat bridge casts as objective-C keywords so we can warn on them
 186     // in non-arc mode.
 187     return LangOpts.ObjC ? KS_Enabled : KS_Unknown;
 188   case KEYZVECTOR:
 189     return LangOpts.ZVector ? KS_Enabled : KS_Unknown;
 190   case KEYCOROUTINES:
 191     return LangOpts.Coroutines ? KS_Enabled : KS_Unknown;
 192   case KEYMODULES:
 193     return LangOpts.ModulesTS ? KS_Enabled : KS_Unknown;
 194   case KEYOPENCLCXX:
 195     return LangOpts.OpenCLCPlusPlus ? KS_Enabled : KS_Unknown;
 196   case KEYMSCOMPAT:
 197     return LangOpts.MSVCCompat ? KS_Enabled : KS_Unknown;
 198   case KEYSYCL:
 199     return LangOpts.isSYCL() ? KS_Enabled : KS_Unknown;
 200   case KEYCUDA:
 201     return LangOpts.CUDA ? KS_Enabled : KS_Unknown;
 202   case KEYNOCXX:
 203     // This is enabled in all non-C++ modes, but might be enabled for other
 204     // reasons as well.
 205     return LangOpts.CPlusPlus ? KS_Unknown : KS_Enabled;
 206   case KEYNOOPENCL:
 207     // The disable behavior for this is handled in getKeywordStatus.
 208     return KS_Unknown;
 209   case KEYNOMS18:
 210     // The disable behavior for this is handled in getKeywordStatus.
 211     return KS_Unknown;
 212   default:
 213     llvm_unreachable("Unknown KeywordStatus flag");
 214   }
 215 }
 216
 217 /// Translates flags as specified in TokenKinds.def into keyword status
 218 /// in the given language standard.
 219 static KeywordStatus getKeywordStatus(const LangOptions &LangOpts,
 220                                       unsigned Flags) {
 221   // KEYALL means always enabled, so special case this one.
 222   if (Flags == KEYALL) return KS_Enabled;
 223   // These are tests that need to 'always win', as they are special in that they
 224   // disable based on certain conditions.
 225   if (LangOpts.OpenCL && (Flags & KEYNOOPENCL)) return KS_Disabled;
 226   if (LangOpts.MSVCCompat && (Flags & KEYNOMS18) &&
 227       !LangOpts.isCompatibleWithMSVC(LangOptions::MSVC2015))
 228     return KS_Disabled;
 229
 230   KeywordStatus CurStatus = KS_Unknown;
 231
 232   while (Flags != 0) {
 233     unsigned CurFlag = Flags & ~(Flags - 1);
 234     Flags = Flags & ~CurFlag;
 235     CurStatus = std::max(
 236         CurStatus,
 237         getKeywordStatusHelper(LangOpts, static_cast<TokenKey>(CurFlag)));
 238   }
 239
 240   if (CurStatus == KS_Unknown)
 241     return KS_Disabled;
 242   return CurStatus;
 243 }
 244
 245 /// AddKeyword - This method is used to associate a token ID with specific
 246 /// identifiers because they are language keywords.  This causes the lexer to
 247 /// automatically map matching identifiers to specialized token codes.
 248 static void AddKeyword(StringRef Keyword,
 249                        tok::TokenKind TokenCode, unsigned Flags,
 250                        const LangOptions &LangOpts, IdentifierTable &Table) {
 251   KeywordStatus AddResult = getKeywordStatus(LangOpts, Flags);
 252
 253   // Don't add this keyword if disabled in this language.
 254   if (AddResult == KS_Disabled) return;
 255
 256   IdentifierInfo &Info =
 257       Table.get(Keyword, AddResult == KS_Future ? tok::identifier : TokenCode);
 258   Info.setIsExtensionToken(AddResult == KS_Extension);
 259   Info.setIsFutureCompatKeyword(AddResult == KS_Future);
 260 }
 261
 262 /// AddCXXOperatorKeyword - Register a C++ operator keyword alternative
 263 /// representations.
 264 static void AddCXXOperatorKeyword(StringRef Keyword,
 265                                   tok::TokenKind TokenCode,
 266                                   IdentifierTable &Table) {
 267   IdentifierInfo &Info = Table.get(Keyword, TokenCode);
 268   Info.setIsCPlusPlusOperatorKeyword();
 269 }
 270
 271 /// AddObjCKeyword - Register an Objective-C \@keyword like "class" "selector"
 272 /// or "property".
 273 static void AddObjCKeyword(StringRef Name,
 274                            tok::ObjCKeywordKind ObjCID,
 275                            IdentifierTable &Table) {
 276   Table.get(Name).setObjCKeywordID(ObjCID);
 277 }
 278
 279 /// AddKeywords - Add all keywords to the symbol table.
 280 ///
 281 void IdentifierTable::AddKeywords(const LangOptions &LangOpts) {
 282   // Add keywords and tokens for the current language.
 283 #define KEYWORD(NAME, FLAGS) \
 284   AddKeyword(StringRef(#NAME), tok::kw_ ## NAME,  \
 285              FLAGS, LangOpts, *this);
 286 #define ALIAS(NAME, TOK, FLAGS) \
 287   AddKeyword(StringRef(NAME), tok::kw_ ## TOK,  \
 288              FLAGS, LangOpts, *this);
 289 #define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \
 290   if (LangOpts.CXXOperatorNames)          \
 291     AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this);
 292 #define OBJC_AT_KEYWORD(NAME)  \
 293   if (LangOpts.ObjC)           \
 294     AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
 295 #define TESTING_KEYWORD(NAME, FLAGS)
 296 #include "clang/Basic/TokenKinds.def"
 297
 298   if (LangOpts.ParseUnknownAnytype)
 299     AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL,
 300                LangOpts, *this);
 301
 302   if (LangOpts.DeclSpecKeyword)
 303     AddKeyword("__declspec", tok::kw___declspec, KEYALL, LangOpts, *this);
 304
 305   if (LangOpts.IEEE128)
 306     AddKeyword("__ieee128", tok::kw___float128, KEYALL, LangOpts, *this);
 307
 308   // Add the 'import' contextual keyword.
 309   get("import").setModulesImport(true);
 310 }
 311
 312 /// Checks if the specified token kind represents a keyword in the
 313 /// specified language.
 314 /// \returns Status of the keyword in the language.
 315 static KeywordStatus getTokenKwStatus(const LangOptions &LangOpts,
 316                                       tok::TokenKind K) {
 317   switch (K) {
 318 #define KEYWORD(NAME, FLAGS) \
 319   case tok::kw_##NAME: return getKeywordStatus(LangOpts, FLAGS);
 320 #include "clang/Basic/TokenKinds.def"
 321   default: return KS_Disabled;
 322   }
 323 }
 324
 325 /// Returns true if the identifier represents a keyword in the
 326 /// specified language.
 327 bool IdentifierInfo::isKeyword(const LangOptions &LangOpts) const {
 328   switch (getTokenKwStatus(LangOpts, getTokenID())) {
 329   case KS_Enabled:
 330   case KS_Extension:
 331     return true;
 332   default:
 333     return false;
 334   }
 335 }
 336
 337 /// Returns true if the identifier represents a C++ keyword in the
 338 /// specified language.
 339 bool IdentifierInfo::isCPlusPlusKeyword(const LangOptions &LangOpts) const {
 340   if (!LangOpts.CPlusPlus || !isKeyword(LangOpts))
 341     return false;
 342   // This is a C++ keyword if this identifier is not a keyword when checked
 343   // using LangOptions without C++ support.
 344   LangOptions LangOptsNoCPP = LangOpts;
 345   LangOptsNoCPP.CPlusPlus = false;
 346   LangOptsNoCPP.CPlusPlus11 = false;
 347   LangOptsNoCPP.CPlusPlus20 = false;
 348   return !isKeyword(LangOptsNoCPP);
 349 }
 350
 351 ReservedIdentifierStatus
 352 IdentifierInfo::isReserved(const LangOptions &LangOpts) const {
 353   StringRef Name = getName();
 354
 355   // '_' is a reserved identifier, but its use is so common (e.g. to store
 356   // ignored values) that we don't warn on it.
 357   if (Name.size() <= 1)
 358     return ReservedIdentifierStatus::NotReserved;
 359
 360   // [lex.name] p3
 361   if (Name[0] == '_') {
 362
 363     // Each name that begins with an underscore followed by an uppercase letter
 364     // or another underscore is reserved.
 365     if (Name[1] == '_')
 366       return ReservedIdentifierStatus::StartsWithDoubleUnderscore;
 367
 368     if ('A' <= Name[1] && Name[1] <= 'Z')
 369       return ReservedIdentifierStatus::
 370           StartsWithUnderscoreFollowedByCapitalLetter;
 371
 372     // This is a bit misleading: it actually means it's only reserved if we're
 373     // at global scope because it starts with an underscore.
 374     return ReservedIdentifierStatus::StartsWithUnderscoreAtGlobalScope;
 375   }
 376
 377   // Each name that contains a double underscore (__) is reserved.
 378   if (LangOpts.CPlusPlus && Name.contains("__"))
 379     return ReservedIdentifierStatus::ContainsDoubleUnderscore;
 380
 381   return ReservedIdentifierStatus::NotReserved;
 382 }
 383
 384 StringRef IdentifierInfo::deuglifiedName() const {
 385   StringRef Name = getName();
 386   if (Name.size() >= 2 && Name.front() == '_' &&
 387       (Name[1] == '_' || (Name[1] >= 'A' && Name[1] <= 'Z')))
 388     return Name.ltrim('_');
 389   return Name;
 390 }
 391
 392 tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const {
 393   // We use a perfect hash function here involving the length of the keyword,
 394   // the first and third character.  For preprocessor ID's there are no
 395   // collisions (if there were, the switch below would complain about duplicate
 396   // case values).  Note that this depends on 'if' being null terminated.
 397
 398 #define HASH(LEN, FIRST, THIRD) \
 399   (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31)
 400 #define CASE(LEN, FIRST, THIRD, NAME) \
 401   case HASH(LEN, FIRST, THIRD): \
 402     return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME
 403
 404   unsigned Len = getLength();
 405   if (Len < 2) return tok::pp_not_keyword;
 406   const char *Name = getNameStart();
 407   switch (HASH(Len, Name[0], Name[2])) {
 408   default: return tok::pp_not_keyword;
 409   CASE( 2, 'i', '\0', if);
 410   CASE( 4, 'e', 'i', elif);
 411   CASE( 4, 'e', 's', else);
 412   CASE( 4, 'l', 'n', line);
 413   CASE( 4, 's', 'c', sccs);
 414   CASE( 5, 'e', 'd', endif);
 415   CASE( 5, 'e', 'r', error);
 416   CASE( 5, 'i', 'e', ident);
 417   CASE( 5, 'i', 'd', ifdef);
 418   CASE( 5, 'u', 'd', undef);
 419
 420   CASE( 6, 'a', 's', assert);
 421   CASE( 6, 'd', 'f', define);
 422   CASE( 6, 'i', 'n', ifndef);
 423   CASE( 6, 'i', 'p', import);
 424   CASE( 6, 'p', 'a', pragma);
 425
 426   CASE( 7, 'd', 'f', defined);
 427   CASE( 7, 'e', 'i', elifdef);
 428   CASE( 7, 'i', 'c', include);
 429   CASE( 7, 'w', 'r', warning);
 430
 431   CASE( 8, 'e', 'i', elifndef);
 432   CASE( 8, 'u', 'a', unassert);
 433   CASE(12, 'i', 'c', include_next);
 434
 435   CASE(14, '_', 'p', __public_macro);
 436
 437   CASE(15, '_', 'p', __private_macro);
 438
 439   CASE(16, '_', 'i', __include_macros);
 440 #undef CASE
 441 #undef HASH
 442   }
 443 }
 444
 445 //===----------------------------------------------------------------------===//
 446 // Stats Implementation
 447 //===----------------------------------------------------------------------===//
 448
 449 /// PrintStats - Print statistics about how well the identifier table is doing
 450 /// at hashing identifiers.
 451 void IdentifierTable::PrintStats() const {
 452   unsigned NumBuckets = HashTable.getNumBuckets();
 453   unsigned NumIdentifiers = HashTable.getNumItems();
 454   unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers;
 455   unsigned AverageIdentifierSize = 0;
 456   unsigned MaxIdentifierLength = 0;
 457
 458   // TODO: Figure out maximum times an identifier had to probe for -stats.
 459   for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator
 460        I = HashTable.begin(), E = HashTable.end(); I != E; ++I) {
 461     unsigned IdLen = I->getKeyLength();
 462     AverageIdentifierSize += IdLen;
 463     if (MaxIdentifierLength < IdLen)
 464       MaxIdentifierLength = IdLen;
 465   }
 466
 467   fprintf(stderr, "\n*** Identifier Table Stats:\n");
 468   fprintf(stderr, "# Identifiers:   %d\n", NumIdentifiers);
 469   fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets);
 470   fprintf(stderr, "Hash density (#identifiers per bucket): %f\n",
 471           NumIdentifiers/(double)NumBuckets);
 472   fprintf(stderr, "Ave identifier length: %f\n",
 473           (AverageIdentifierSize/(double)NumIdentifiers));
 474   fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength);
 475
 476   // Compute statistics about the memory allocated for identifiers.
 477   HashTable.getAllocator().PrintStats();
 478 }
 479
 480 //===----------------------------------------------------------------------===//
 481 // SelectorTable Implementation
 482 //===----------------------------------------------------------------------===//
 483
 484 unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) {
 485   return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr());
 486 }
 487
 488 namespace clang {
 489
 490 /// One of these variable length records is kept for each
 491 /// selector containing more than one keyword. We use a folding set
 492 /// to unique aggregate names (keyword selectors in ObjC parlance). Access to
 493 /// this class is provided strictly through Selector.
 494 class alignas(IdentifierInfoAlignment) MultiKeywordSelector
 495     : public detail::DeclarationNameExtra,
 496       public llvm::FoldingSetNode {
 497   MultiKeywordSelector(unsigned nKeys) : DeclarationNameExtra(nKeys) {}
 498
 499 public:
 500   // Constructor for keyword selectors.
 501   MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV)
 502       : DeclarationNameExtra(nKeys) {
 503     assert((nKeys > 1) && "not a multi-keyword selector");
 504
 505     // Fill in the trailing keyword array.
 506     IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this + 1);
 507     for (unsigned i = 0; i != nKeys; ++i)
 508       KeyInfo[i] = IIV[i];
 509   }
 510
 511   // getName - Derive the full selector name and return it.
 512   std::string getName() const;
 513
 514   using DeclarationNameExtra::getNumArgs;
 515
 516   using keyword_iterator = IdentifierInfo *const *;
 517
 518   keyword_iterator keyword_begin() const {
 519     return reinterpret_cast<keyword_iterator>(this + 1);
 520   }
 521
 522   keyword_iterator keyword_end() const {
 523     return keyword_begin() + getNumArgs();
 524   }
 525
 526   IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const {
 527     assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index");
 528     return keyword_begin()[i];
 529   }
 530
 531   static void Profile(llvm::FoldingSetNodeID &ID, keyword_iterator ArgTys,
 532                       unsigned NumArgs) {
 533     ID.AddInteger(NumArgs);
 534     for (unsigned i = 0; i != NumArgs; ++i)
 535       ID.AddPointer(ArgTys[i]);
 536   }
 537
 538   void Profile(llvm::FoldingSetNodeID &ID) {
 539     Profile(ID, keyword_begin(), getNumArgs());
 540   }
 541 };
 542
 543 } // namespace clang.
 544
 545 bool Selector::isKeywordSelector(ArrayRef<StringRef> Names) const {
 546   assert(!Names.empty() && "must have >= 1 selector slots");
 547   if (getNumArgs() != Names.size())
 548     return false;
 549   for (unsigned I = 0, E = Names.size(); I != E; ++I) {
 550     if (getNameForSlot(I) != Names[I])
 551       return false;
 552   }
 553   return true;
 554 }
 555
 556 bool Selector::isUnarySelector(StringRef Name) const {
 557   return isUnarySelector() && getNameForSlot(0) == Name;
 558 }
 559
 560 unsigned Selector::getNumArgs() const {
 561   unsigned IIF = getIdentifierInfoFlag();
 562   if (IIF <= ZeroArg)
 563     return 0;
 564   if (IIF == OneArg)
 565     return 1;
 566   // We point to a MultiKeywordSelector.
 567   MultiKeywordSelector *SI = getMultiKeywordSelector();
 568   return SI->getNumArgs();
 569 }
 570
 571 IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const {
 572   if (getIdentifierInfoFlag() < MultiArg) {
 573     assert(argIndex == 0 && "illegal keyword index");
 574     return getAsIdentifierInfo();
 575   }
 576
 577   // We point to a MultiKeywordSelector.
 578   MultiKeywordSelector *SI = getMultiKeywordSelector();
 579   return SI->getIdentifierInfoForSlot(argIndex);
 580 }
 581
 582 StringRef Selector::getNameForSlot(unsigned int argIndex) const {
 583   IdentifierInfo *II = getIdentifierInfoForSlot(argIndex);
 584   return II ? II->getName() : StringRef();
 585 }
 586
 587 std::string MultiKeywordSelector::getName() const {
 588   SmallString<256> Str;
 589   llvm::raw_svector_ostream OS(Str);
 590   for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) {
 591     if (*I)
 592       OS << (*I)->getName();
 593     OS << ':';
 594   }
 595
 596   return std::string(OS.str());
 597 }
 598
 599 std::string Selector::getAsString() const {
 600   if (InfoPtr == 0)
 601     return "<null selector>";
 602
 603   if (getIdentifierInfoFlag() < MultiArg) {
 604     IdentifierInfo *II = getAsIdentifierInfo();
 605
 606     if (getNumArgs() == 0) {
 607       assert(II && "If the number of arguments is 0 then II is guaranteed to "
 608                    "not be null.");
 609       return std::string(II->getName());
 610     }
 611
 612     if (!II)
 613       return ":";
 614
 615     return II->getName().str() + ":";
 616   }
 617
 618   // We have a multiple keyword selector.
 619   return getMultiKeywordSelector()->getName();
 620 }
 621
 622 void Selector::print(llvm::raw_ostream &OS) const {
 623   OS << getAsString();
 624 }
 625
 626 LLVM_DUMP_METHOD void Selector::dump() const { print(llvm::errs()); }
 627
 628 /// Interpreting the given string using the normal CamelCase
 629 /// conventions, determine whether the given string starts with the
 630 /// given "word", which is assumed to end in a lowercase letter.
 631 static bool startsWithWord(StringRef name, StringRef word) {
 632   if (name.size() < word.size()) return false;
 633   return ((name.size() == word.size() || !isLowercase(name[word.size()])) &&
 634           name.startswith(word));
 635 }
 636
 637 ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) {
 638   IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
 639   if (!first) return OMF_None;
 640
 641   StringRef name = first->getName();
 642   if (sel.isUnarySelector()) {
 643     if (name == "autorelease") return OMF_autorelease;
 644     if (name == "dealloc") return OMF_dealloc;
 645     if (name == "finalize") return OMF_finalize;
 646     if (name == "release") return OMF_release;
 647     if (name == "retain") return OMF_retain;
 648     if (name == "retainCount") return OMF_retainCount;
 649     if (name == "self") return OMF_self;
 650     if (name == "initialize") return OMF_initialize;
 651   }
 652
 653   if (name == "performSelector" || name == "performSelectorInBackground" ||
 654       name == "performSelectorOnMainThread")
 655     return OMF_performSelector;
 656
 657   // The other method families may begin with a prefix of underscores.
 658   while (!name.empty() && name.front() == '_')
 659     name = name.substr(1);
 660
 661   if (name.empty()) return OMF_None;
 662   switch (name.front()) {
 663   case 'a':
 664     if (startsWithWord(name, "alloc")) return OMF_alloc;
 665     break;
 666   case 'c':
 667     if (startsWithWord(name, "copy")) return OMF_copy;
 668     break;
 669   case 'i':
 670     if (startsWithWord(name, "init")) return OMF_init;
 671     break;
 672   case 'm':
 673     if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy;
 674     break;
 675   case 'n':
 676     if (startsWithWord(name, "new")) return OMF_new;
 677     break;
 678   default:
 679     break;
 680   }
 681
 682   return OMF_None;
 683 }
 684
 685 ObjCInstanceTypeFamily Selector::getInstTypeMethodFamily(Selector sel) {
 686   IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
 687   if (!first) return OIT_None;
 688
 689   StringRef name = first->getName();
 690
 691   if (name.empty()) return OIT_None;
 692   switch (name.front()) {
 693     case 'a':
 694       if (startsWithWord(name, "array")) return OIT_Array;
 695       break;
 696     case 'd':
 697       if (startsWithWord(name, "default")) return OIT_ReturnsSelf;
 698       if (startsWithWord(name, "dictionary")) return OIT_Dictionary;
 699       break;
 700     case 's':
 701       if (startsWithWord(name, "shared")) return OIT_ReturnsSelf;
 702       if (startsWithWord(name, "standard")) return OIT_Singleton;
 703       break;
 704     case 'i':
 705       if (startsWithWord(name, "init")) return OIT_Init;
 706       break;
 707     default:
 708       break;
 709   }
 710   return OIT_None;
 711 }
 712
 713 ObjCStringFormatFamily Selector::getStringFormatFamilyImpl(Selector sel) {
 714   IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
 715   if (!first) return SFF_None;
 716
 717   StringRef name = first->getName();
 718
 719   switch (name.front()) {
 720     case 'a':
 721       if (name == "appendFormat") return SFF_NSString;
 722       break;
 723
 724     case 'i':
 725       if (name == "initWithFormat") return SFF_NSString;
 726       break;
 727
 728     case 'l':
 729       if (name == "localizedStringWithFormat") return SFF_NSString;
 730       break;
 731
 732     case 's':
 733       if (name == "stringByAppendingFormat" ||
 734           name == "stringWithFormat") return SFF_NSString;
 735       break;
 736   }
 737   return SFF_None;
 738 }
 739
 740 namespace {
 741
 742 struct SelectorTableImpl {
 743   llvm::FoldingSet<MultiKeywordSelector> Table;
 744   llvm::BumpPtrAllocator Allocator;
 745 };
 746
 747 } // namespace
 748
 749 static SelectorTableImpl &getSelectorTableImpl(void *P) {
 750   return *static_cast<SelectorTableImpl*>(P);
 751 }
 752
 753 SmallString<64>
 754 SelectorTable::constructSetterName(StringRef Name) {
 755   SmallString<64> SetterName("set");
 756   SetterName += Name;
 757   SetterName[3] = toUppercase(SetterName[3]);
 758   return SetterName;
 759 }
 760
 761 Selector
 762 SelectorTable::constructSetterSelector(IdentifierTable &Idents,
 763                                        SelectorTable &SelTable,
 764                                        const IdentifierInfo *Name) {
 765   IdentifierInfo *SetterName =
 766     &Idents.get(constructSetterName(Name->getName()));
 767   return SelTable.getUnarySelector(SetterName);
 768 }
 769
 770 std::string SelectorTable::getPropertyNameFromSetterSelector(Selector Sel) {
 771   StringRef Name = Sel.getNameForSlot(0);
 772   assert(Name.startswith("set") && "invalid setter name");
 773   return (Twine(toLowercase(Name[3])) + Name.drop_front(4)).str();
 774 }
 775
 776 size_t SelectorTable::getTotalMemory() const {
 777   SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
 778   return SelTabImpl.Allocator.getTotalMemory();
 779 }
 780
 781 Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) {
 782   if (nKeys < 2)
 783     return Selector(IIV[0], nKeys);
 784
 785   SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
 786
 787   // Unique selector, to guarantee there is one per name.
 788   llvm::FoldingSetNodeID ID;
 789   MultiKeywordSelector::Profile(ID, IIV, nKeys);
 790
 791   void *InsertPos = nullptr;
 792   if (MultiKeywordSelector *SI =
 793         SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos))
 794     return Selector(SI);
 795
 796   // MultiKeywordSelector objects are not allocated with new because they have a
 797   // variable size array (for parameter types) at the end of them.
 798   unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *);
 799   MultiKeywordSelector *SI =
 800       (MultiKeywordSelector *)SelTabImpl.Allocator.Allocate(
 801           Size, alignof(MultiKeywordSelector));
 802   new (SI) MultiKeywordSelector(nKeys, IIV);
 803   SelTabImpl.Table.InsertNode(SI, InsertPos);
 804   return Selector(SI);
 805 }
 806
 807 SelectorTable::SelectorTable() {
 808   Impl = new SelectorTableImpl();
 809 }
 810
 811 SelectorTable::~SelectorTable() {
 812   delete &getSelectorTableImpl(Impl);
 813 }
 814
 815 const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) {
 816   switch (Operator) {
 817   case OO_None:
 818   case NUM_OVERLOADED_OPERATORS:
 819     return nullptr;
 820
 821 #define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \
 822   case OO_##Name: return Spelling;
 823 #include "clang/Basic/OperatorKinds.def"
 824   }
 825
 826   llvm_unreachable("Invalid OverloadedOperatorKind!");
 827 }
 828
 829 StringRef clang::getNullabilitySpelling(NullabilityKind kind,
 830                                         bool isContextSensitive) {
 831   switch (kind) {
 832   case NullabilityKind::NonNull:
 833     return isContextSensitive ? "nonnull" : "_Nonnull";
 834
 835   case NullabilityKind::Nullable:
 836     return isContextSensitive ? "nullable" : "_Nullable";
 837
 838   case NullabilityKind::NullableResult:
 839     assert(!isContextSensitive &&
 840            "_Nullable_result isn't supported as context-sensitive keyword");
 841     return "_Nullable_result";
 842
 843   case NullabilityKind::Unspecified:
 844     return isContextSensitive ? "null_unspecified" : "_Null_unspecified";
 845   }
 846   llvm_unreachable("Unknown nullability kind.");
 847 }
 848
 849 diag::kind
 850 IdentifierTable::getFutureCompatDiagKind(const IdentifierInfo &II,
 851                                          const LangOptions &LangOpts) {
 852   assert(II.isFutureCompatKeyword() && "diagnostic should not be needed");
 853
 854   unsigned Flags = llvm::StringSwitch<unsigned>(II.getName())
 855 #define KEYWORD(NAME, FLAGS) .Case(#NAME, FLAGS)
 856 #include "clang/Basic/TokenKinds.def"
 857 #undef KEYWORD
 858       ;
 859
 860   if (LangOpts.CPlusPlus) {
 861     if ((Flags & KEYCXX11) == KEYCXX11)
 862       return diag::warn_cxx11_keyword;
 863
 864     // char8_t is not modeled as a CXX20_KEYWORD because it's not
 865     // unconditionally enabled in C++20 mode. (It can be disabled
 866     // by -fno-char8_t.)
 867     if (((Flags & KEYCXX20) == KEYCXX20) ||
 868         ((Flags & CHAR8SUPPORT) == CHAR8SUPPORT))
 869       return diag::warn_cxx20_keyword;
 870   } else {
 871     if ((Flags & KEYC99) == KEYC99)
 872       return diag::warn_c99_keyword;
 873     if ((Flags & KEYC2X) == KEYC2X)
 874       return diag::warn_c2x_keyword;
 875   }
 876
 877   llvm_unreachable(
 878       "Keyword not known to come from a newer Standard or proposed Standard");
 879 }