1 //===--- IdentifierTable.cpp - Hash table for identifier lookup -----------===// 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 IdentifierInfo, IdentifierVisitor, and 11 // IdentifierTable interfaces. 12 // 13 //===----------------------------------------------------------------------===// 14 15 #include "clang/Basic/IdentifierTable.h" 16 #include "clang/Basic/LangOptions.h" 17 #include "llvm/ADT/FoldingSet.h" 18 #include "llvm/ADT/DenseMap.h" 19 #include "llvm/ADT/SmallString.h" 20 #include "llvm/ADT/StringSwitch.h" 21 #include "llvm/Support/raw_ostream.h" 22 #include "llvm/Support/ErrorHandling.h" 23 #include <cctype> 24 #include <cstdio> 25 26 using namespace clang; 27 28 //===----------------------------------------------------------------------===// 29 // IdentifierInfo Implementation 30 //===----------------------------------------------------------------------===// 31 32 IdentifierInfo::IdentifierInfo() { 33 TokenID = tok::identifier; 34 ObjCOrBuiltinID = 0; 35 HasMacro = false; 36 IsExtension = false; 37 IsCXX11CompatKeyword = false; 38 IsPoisoned = false; 39 IsCPPOperatorKeyword = false; 40 NeedsHandleIdentifier = false; 41 IsFromAST = false; 42 ChangedAfterLoad = false; 43 RevertedTokenID = false; 44 OutOfDate = false; 45 IsModulesImport = false; 46 FETokenInfo = 0; 47 Entry = 0; 48 } 49 50 //===----------------------------------------------------------------------===// 51 // IdentifierTable Implementation 52 //===----------------------------------------------------------------------===// 53 54 IdentifierIterator::~IdentifierIterator() { } 55 56 IdentifierInfoLookup::~IdentifierInfoLookup() {} 57 58 namespace { 59 /// \brief A simple identifier lookup iterator that represents an 60 /// empty sequence of identifiers. 61 class EmptyLookupIterator : public IdentifierIterator 62 { 63 public: 64 virtual StringRef Next() { return StringRef(); } 65 }; 66 } 67 68 IdentifierIterator *IdentifierInfoLookup::getIdentifiers() const { 69 return new EmptyLookupIterator(); 70 } 71 72 ExternalIdentifierLookup::~ExternalIdentifierLookup() {} 73 74 IdentifierTable::IdentifierTable(const LangOptions &LangOpts, 75 IdentifierInfoLookup* externalLookup) 76 : HashTable(8192), // Start with space for 8K identifiers. 77 ExternalLookup(externalLookup) { 78 79 // Populate the identifier table with info about keywords for the current 80 // language. 81 AddKeywords(LangOpts); 82 83 84 // Add the '_experimental_modules_import' contextual keyword. 85 get("__experimental_modules_import").setModulesImport(true); 86 } 87 88 //===----------------------------------------------------------------------===// 89 // Language Keyword Implementation 90 //===----------------------------------------------------------------------===// 91 92 // Constants for TokenKinds.def 93 namespace { 94 enum { 95 KEYC99 = 0x1, 96 KEYCXX = 0x2, 97 KEYCXX0X = 0x4, 98 KEYGNU = 0x8, 99 KEYMS = 0x10, 100 BOOLSUPPORT = 0x20, 101 KEYALTIVEC = 0x40, 102 KEYNOCXX = 0x80, 103 KEYBORLAND = 0x100, 104 KEYOPENCL = 0x200, 105 KEYC11 = 0x400, 106 KEYARC = 0x800, 107 KEYNOMS = 0x01000, 108 KEYALL = (0xffff & ~KEYNOMS) // Because KEYNOMS is used to exclude. 109 }; 110 } 111 112 /// AddKeyword - This method is used to associate a token ID with specific 113 /// identifiers because they are language keywords. This causes the lexer to 114 /// automatically map matching identifiers to specialized token codes. 115 /// 116 /// The C90/C99/CPP/CPP0x flags are set to 3 if the token is a keyword in a 117 /// future language standard, set to 2 if the token should be enabled in the 118 /// specified language, set to 1 if it is an extension in the specified 119 /// language, and set to 0 if disabled in the specified language. 120 static void AddKeyword(StringRef Keyword, 121 tok::TokenKind TokenCode, unsigned Flags, 122 const LangOptions &LangOpts, IdentifierTable &Table) { 123 unsigned AddResult = 0; 124 if (Flags == KEYALL) AddResult = 2; 125 else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2; 126 else if (LangOpts.CPlusPlus0x && (Flags & KEYCXX0X)) AddResult = 2; 127 else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2; 128 else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1; 129 else if (LangOpts.MicrosoftExt && (Flags & KEYMS)) AddResult = 1; 130 else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1; 131 else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2; 132 else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2; 133 else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2; 134 else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2; 135 else if (LangOpts.C11 && (Flags & KEYC11)) AddResult = 2; 136 // We treat bridge casts as objective-C keywords so we can warn on them 137 // in non-arc mode. 138 else if (LangOpts.ObjC2 && (Flags & KEYARC)) AddResult = 2; 139 else if (LangOpts.CPlusPlus && (Flags & KEYCXX0X)) AddResult = 3; 140 141 // Don't add this keyword under MicrosoftMode. 142 if (LangOpts.MicrosoftMode && (Flags & KEYNOMS)) 143 return; 144 // Don't add this keyword if disabled in this language. 145 if (AddResult == 0) return; 146 147 IdentifierInfo &Info = 148 Table.get(Keyword, AddResult == 3 ? tok::identifier : TokenCode); 149 Info.setIsExtensionToken(AddResult == 1); 150 Info.setIsCXX11CompatKeyword(AddResult == 3); 151 } 152 153 /// AddCXXOperatorKeyword - Register a C++ operator keyword alternative 154 /// representations. 155 static void AddCXXOperatorKeyword(StringRef Keyword, 156 tok::TokenKind TokenCode, 157 IdentifierTable &Table) { 158 IdentifierInfo &Info = Table.get(Keyword, TokenCode); 159 Info.setIsCPlusPlusOperatorKeyword(); 160 } 161 162 /// AddObjCKeyword - Register an Objective-C \@keyword like "class" "selector" 163 /// or "property". 164 static void AddObjCKeyword(StringRef Name, 165 tok::ObjCKeywordKind ObjCID, 166 IdentifierTable &Table) { 167 Table.get(Name).setObjCKeywordID(ObjCID); 168 } 169 170 /// AddKeywords - Add all keywords to the symbol table. 171 /// 172 void IdentifierTable::AddKeywords(const LangOptions &LangOpts) { 173 // Add keywords and tokens for the current language. 174 #define KEYWORD(NAME, FLAGS) \ 175 AddKeyword(StringRef(#NAME), tok::kw_ ## NAME, \ 176 FLAGS, LangOpts, *this); 177 #define ALIAS(NAME, TOK, FLAGS) \ 178 AddKeyword(StringRef(NAME), tok::kw_ ## TOK, \ 179 FLAGS, LangOpts, *this); 180 #define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \ 181 if (LangOpts.CXXOperatorNames) \ 182 AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this); 183 #define OBJC1_AT_KEYWORD(NAME) \ 184 if (LangOpts.ObjC1) \ 185 AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this); 186 #define OBJC2_AT_KEYWORD(NAME) \ 187 if (LangOpts.ObjC2) \ 188 AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this); 189 #define TESTING_KEYWORD(NAME, FLAGS) 190 #include "clang/Basic/TokenKinds.def" 191 192 if (LangOpts.ParseUnknownAnytype) 193 AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL, 194 LangOpts, *this); 195 } 196 197 tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const { 198 // We use a perfect hash function here involving the length of the keyword, 199 // the first and third character. For preprocessor ID's there are no 200 // collisions (if there were, the switch below would complain about duplicate 201 // case values). Note that this depends on 'if' being null terminated. 202 203 #define HASH(LEN, FIRST, THIRD) \ 204 (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31) 205 #define CASE(LEN, FIRST, THIRD, NAME) \ 206 case HASH(LEN, FIRST, THIRD): \ 207 return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME 208 209 unsigned Len = getLength(); 210 if (Len < 2) return tok::pp_not_keyword; 211 const char *Name = getNameStart(); 212 switch (HASH(Len, Name[0], Name[2])) { 213 default: return tok::pp_not_keyword; 214 CASE( 2, 'i', '\0', if); 215 CASE( 4, 'e', 'i', elif); 216 CASE( 4, 'e', 's', else); 217 CASE( 4, 'l', 'n', line); 218 CASE( 4, 's', 'c', sccs); 219 CASE( 5, 'e', 'd', endif); 220 CASE( 5, 'e', 'r', error); 221 CASE( 5, 'i', 'e', ident); 222 CASE( 5, 'i', 'd', ifdef); 223 CASE( 5, 'u', 'd', undef); 224 225 CASE( 6, 'a', 's', assert); 226 CASE( 6, 'd', 'f', define); 227 CASE( 6, 'i', 'n', ifndef); 228 CASE( 6, 'i', 'p', import); 229 CASE( 6, 'p', 'a', pragma); 230 231 CASE( 7, 'd', 'f', defined); 232 CASE( 7, 'i', 'c', include); 233 CASE( 7, 'w', 'r', warning); 234 235 CASE( 8, 'u', 'a', unassert); 236 CASE(12, 'i', 'c', include_next); 237 238 CASE(14, '_', 'p', __public_macro); 239 240 CASE(15, '_', 'p', __private_macro); 241 242 CASE(16, '_', 'i', __include_macros); 243 #undef CASE 244 #undef HASH 245 } 246 } 247 248 //===----------------------------------------------------------------------===// 249 // Stats Implementation 250 //===----------------------------------------------------------------------===// 251 252 /// PrintStats - Print statistics about how well the identifier table is doing 253 /// at hashing identifiers. 254 void IdentifierTable::PrintStats() const { 255 unsigned NumBuckets = HashTable.getNumBuckets(); 256 unsigned NumIdentifiers = HashTable.getNumItems(); 257 unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers; 258 unsigned AverageIdentifierSize = 0; 259 unsigned MaxIdentifierLength = 0; 260 261 // TODO: Figure out maximum times an identifier had to probe for -stats. 262 for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator 263 I = HashTable.begin(), E = HashTable.end(); I != E; ++I) { 264 unsigned IdLen = I->getKeyLength(); 265 AverageIdentifierSize += IdLen; 266 if (MaxIdentifierLength < IdLen) 267 MaxIdentifierLength = IdLen; 268 } 269 270 fprintf(stderr, "\n*** Identifier Table Stats:\n"); 271 fprintf(stderr, "# Identifiers: %d\n", NumIdentifiers); 272 fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets); 273 fprintf(stderr, "Hash density (#identifiers per bucket): %f\n", 274 NumIdentifiers/(double)NumBuckets); 275 fprintf(stderr, "Ave identifier length: %f\n", 276 (AverageIdentifierSize/(double)NumIdentifiers)); 277 fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength); 278 279 // Compute statistics about the memory allocated for identifiers. 280 HashTable.getAllocator().PrintStats(); 281 } 282 283 //===----------------------------------------------------------------------===// 284 // SelectorTable Implementation 285 //===----------------------------------------------------------------------===// 286 287 unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) { 288 return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr()); 289 } 290 291 namespace clang { 292 /// MultiKeywordSelector - One of these variable length records is kept for each 293 /// selector containing more than one keyword. We use a folding set 294 /// to unique aggregate names (keyword selectors in ObjC parlance). Access to 295 /// this class is provided strictly through Selector. 296 class MultiKeywordSelector 297 : public DeclarationNameExtra, public llvm::FoldingSetNode { 298 MultiKeywordSelector(unsigned nKeys) { 299 ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys; 300 } 301 public: 302 // Constructor for keyword selectors. 303 MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) { 304 assert((nKeys > 1) && "not a multi-keyword selector"); 305 ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys; 306 307 // Fill in the trailing keyword array. 308 IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1); 309 for (unsigned i = 0; i != nKeys; ++i) 310 KeyInfo[i] = IIV[i]; 311 } 312 313 // getName - Derive the full selector name and return it. 314 std::string getName() const; 315 316 unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; } 317 318 typedef IdentifierInfo *const *keyword_iterator; 319 keyword_iterator keyword_begin() const { 320 return reinterpret_cast<keyword_iterator>(this+1); 321 } 322 keyword_iterator keyword_end() const { 323 return keyword_begin()+getNumArgs(); 324 } 325 IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const { 326 assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index"); 327 return keyword_begin()[i]; 328 } 329 static void Profile(llvm::FoldingSetNodeID &ID, 330 keyword_iterator ArgTys, unsigned NumArgs) { 331 ID.AddInteger(NumArgs); 332 for (unsigned i = 0; i != NumArgs; ++i) 333 ID.AddPointer(ArgTys[i]); 334 } 335 void Profile(llvm::FoldingSetNodeID &ID) { 336 Profile(ID, keyword_begin(), getNumArgs()); 337 } 338 }; 339 } // end namespace clang. 340 341 unsigned Selector::getNumArgs() const { 342 unsigned IIF = getIdentifierInfoFlag(); 343 if (IIF <= ZeroArg) 344 return 0; 345 if (IIF == OneArg) 346 return 1; 347 // We point to a MultiKeywordSelector. 348 MultiKeywordSelector *SI = getMultiKeywordSelector(); 349 return SI->getNumArgs(); 350 } 351 352 IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const { 353 if (getIdentifierInfoFlag() < MultiArg) { 354 assert(argIndex == 0 && "illegal keyword index"); 355 return getAsIdentifierInfo(); 356 } 357 // We point to a MultiKeywordSelector. 358 MultiKeywordSelector *SI = getMultiKeywordSelector(); 359 return SI->getIdentifierInfoForSlot(argIndex); 360 } 361 362 StringRef Selector::getNameForSlot(unsigned int argIndex) const { 363 IdentifierInfo *II = getIdentifierInfoForSlot(argIndex); 364 return II? II->getName() : StringRef(); 365 } 366 367 std::string MultiKeywordSelector::getName() const { 368 SmallString<256> Str; 369 llvm::raw_svector_ostream OS(Str); 370 for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) { 371 if (*I) 372 OS << (*I)->getName(); 373 OS << ':'; 374 } 375 376 return OS.str(); 377 } 378 379 std::string Selector::getAsString() const { 380 if (InfoPtr == 0) 381 return "<null selector>"; 382 383 if (getIdentifierInfoFlag() < MultiArg) { 384 IdentifierInfo *II = getAsIdentifierInfo(); 385 386 // If the number of arguments is 0 then II is guaranteed to not be null. 387 if (getNumArgs() == 0) 388 return II->getName(); 389 390 if (!II) 391 return ":"; 392 393 return II->getName().str() + ":"; 394 } 395 396 // We have a multiple keyword selector. 397 return getMultiKeywordSelector()->getName(); 398 } 399 400 /// Interpreting the given string using the normal CamelCase 401 /// conventions, determine whether the given string starts with the 402 /// given "word", which is assumed to end in a lowercase letter. 403 static bool startsWithWord(StringRef name, StringRef word) { 404 if (name.size() < word.size()) return false; 405 return ((name.size() == word.size() || 406 !islower(name[word.size()])) 407 && name.startswith(word)); 408 } 409 410 ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) { 411 IdentifierInfo *first = sel.getIdentifierInfoForSlot(0); 412 if (!first) return OMF_None; 413 414 StringRef name = first->getName(); 415 if (sel.isUnarySelector()) { 416 if (name == "autorelease") return OMF_autorelease; 417 if (name == "dealloc") return OMF_dealloc; 418 if (name == "finalize") return OMF_finalize; 419 if (name == "release") return OMF_release; 420 if (name == "retain") return OMF_retain; 421 if (name == "retainCount") return OMF_retainCount; 422 if (name == "self") return OMF_self; 423 } 424 425 if (name == "performSelector") return OMF_performSelector; 426 427 // The other method families may begin with a prefix of underscores. 428 while (!name.empty() && name.front() == '_') 429 name = name.substr(1); 430 431 if (name.empty()) return OMF_None; 432 switch (name.front()) { 433 case 'a': 434 if (startsWithWord(name, "alloc")) return OMF_alloc; 435 break; 436 case 'c': 437 if (startsWithWord(name, "copy")) return OMF_copy; 438 break; 439 case 'i': 440 if (startsWithWord(name, "init")) return OMF_init; 441 break; 442 case 'm': 443 if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy; 444 break; 445 case 'n': 446 if (startsWithWord(name, "new")) return OMF_new; 447 break; 448 default: 449 break; 450 } 451 452 return OMF_None; 453 } 454 455 namespace { 456 struct SelectorTableImpl { 457 llvm::FoldingSet<MultiKeywordSelector> Table; 458 llvm::BumpPtrAllocator Allocator; 459 }; 460 } // end anonymous namespace. 461 462 static SelectorTableImpl &getSelectorTableImpl(void *P) { 463 return *static_cast<SelectorTableImpl*>(P); 464 } 465 466 /*static*/ Selector 467 SelectorTable::constructSetterName(IdentifierTable &Idents, 468 SelectorTable &SelTable, 469 const IdentifierInfo *Name) { 470 SmallString<100> SelectorName; 471 SelectorName = "set"; 472 SelectorName += Name->getName(); 473 SelectorName[3] = toupper(SelectorName[3]); 474 IdentifierInfo *SetterName = &Idents.get(SelectorName); 475 return SelTable.getUnarySelector(SetterName); 476 } 477 478 size_t SelectorTable::getTotalMemory() const { 479 SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl); 480 return SelTabImpl.Allocator.getTotalMemory(); 481 } 482 483 Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) { 484 if (nKeys < 2) 485 return Selector(IIV[0], nKeys); 486 487 SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl); 488 489 // Unique selector, to guarantee there is one per name. 490 llvm::FoldingSetNodeID ID; 491 MultiKeywordSelector::Profile(ID, IIV, nKeys); 492 493 void *InsertPos = 0; 494 if (MultiKeywordSelector *SI = 495 SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos)) 496 return Selector(SI); 497 498 // MultiKeywordSelector objects are not allocated with new because they have a 499 // variable size array (for parameter types) at the end of them. 500 unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *); 501 MultiKeywordSelector *SI = 502 (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size, 503 llvm::alignOf<MultiKeywordSelector>()); 504 new (SI) MultiKeywordSelector(nKeys, IIV); 505 SelTabImpl.Table.InsertNode(SI, InsertPos); 506 return Selector(SI); 507 } 508 509 SelectorTable::SelectorTable() { 510 Impl = new SelectorTableImpl(); 511 } 512 513 SelectorTable::~SelectorTable() { 514 delete &getSelectorTableImpl(Impl); 515 } 516 517 const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) { 518 switch (Operator) { 519 case OO_None: 520 case NUM_OVERLOADED_OPERATORS: 521 return 0; 522 523 #define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \ 524 case OO_##Name: return Spelling; 525 #include "clang/Basic/OperatorKinds.def" 526 } 527 528 llvm_unreachable("Invalid OverloadedOperatorKind!"); 529 } 530