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/StringRef.h" 20 #include "llvm/ADT/StringSwitch.h" 21 #include "llvm/Support/raw_ostream.h" 22 #include <cstdio> 23 24 using namespace clang; 25 26 //===----------------------------------------------------------------------===// 27 // IdentifierInfo Implementation 28 //===----------------------------------------------------------------------===// 29 30 IdentifierInfo::IdentifierInfo() { 31 TokenID = tok::identifier; 32 ObjCOrBuiltinID = 0; 33 HasMacro = false; 34 IsExtension = false; 35 IsCXX11CompatKeyword = false; 36 IsPoisoned = false; 37 IsCPPOperatorKeyword = false; 38 NeedsHandleIdentifier = false; 39 IsFromAST = false; 40 ChangedAfterLoad = false; 41 RevertedTokenID = false; 42 OutOfDate = false; 43 IsImport = false; 44 FETokenInfo = 0; 45 Entry = 0; 46 } 47 48 //===----------------------------------------------------------------------===// 49 // IdentifierTable Implementation 50 //===----------------------------------------------------------------------===// 51 52 IdentifierIterator::~IdentifierIterator() { } 53 54 IdentifierInfoLookup::~IdentifierInfoLookup() {} 55 56 namespace { 57 /// \brief A simple identifier lookup iterator that represents an 58 /// empty sequence of identifiers. 59 class EmptyLookupIterator : public IdentifierIterator 60 { 61 public: 62 virtual StringRef Next() { return StringRef(); } 63 }; 64 } 65 66 IdentifierIterator *IdentifierInfoLookup::getIdentifiers() const { 67 return new EmptyLookupIterator(); 68 } 69 70 ExternalIdentifierLookup::~ExternalIdentifierLookup() {} 71 72 IdentifierTable::IdentifierTable(const LangOptions &LangOpts, 73 IdentifierInfoLookup* externalLookup) 74 : HashTable(8192), // Start with space for 8K identifiers. 75 ExternalLookup(externalLookup) { 76 77 // Populate the identifier table with info about keywords for the current 78 // language. 79 AddKeywords(LangOpts); 80 } 81 82 //===----------------------------------------------------------------------===// 83 // Language Keyword Implementation 84 //===----------------------------------------------------------------------===// 85 86 // Constants for TokenKinds.def 87 namespace { 88 enum { 89 KEYC99 = 0x1, 90 KEYCXX = 0x2, 91 KEYCXX0X = 0x4, 92 KEYGNU = 0x8, 93 KEYMS = 0x10, 94 BOOLSUPPORT = 0x20, 95 KEYALTIVEC = 0x40, 96 KEYNOCXX = 0x80, 97 KEYBORLAND = 0x100, 98 KEYOPENCL = 0x200, 99 KEYC11 = 0x400, 100 KEYARC = 0x800, 101 KEYALL = 0x0fff 102 }; 103 } 104 105 /// AddKeyword - This method is used to associate a token ID with specific 106 /// identifiers because they are language keywords. This causes the lexer to 107 /// automatically map matching identifiers to specialized token codes. 108 /// 109 /// The C90/C99/CPP/CPP0x flags are set to 3 if the token is a keyword in a 110 /// future language standard, set to 2 if the token should be enabled in the 111 /// specified langauge, set to 1 if it is an extension in the specified 112 /// language, and set to 0 if disabled in the specified language. 113 static void AddKeyword(StringRef Keyword, 114 tok::TokenKind TokenCode, unsigned Flags, 115 const LangOptions &LangOpts, IdentifierTable &Table) { 116 unsigned AddResult = 0; 117 if (Flags == KEYALL) AddResult = 2; 118 else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2; 119 else if (LangOpts.CPlusPlus0x && (Flags & KEYCXX0X)) AddResult = 2; 120 else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2; 121 else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1; 122 else if (LangOpts.MicrosoftExt && (Flags & KEYMS)) AddResult = 1; 123 else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1; 124 else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2; 125 else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2; 126 else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2; 127 else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2; 128 else if (LangOpts.C11 && (Flags & KEYC11)) AddResult = 2; 129 // We treat bridge casts as objective-C keywords so we can warn on them 130 // in non-arc mode. 131 else if (LangOpts.ObjC2 && (Flags & KEYARC)) AddResult = 2; 132 else if (LangOpts.CPlusPlus && (Flags & KEYCXX0X)) AddResult = 3; 133 134 // Don't add this keyword if disabled in this language. 135 if (AddResult == 0) return; 136 137 IdentifierInfo &Info = 138 Table.get(Keyword, AddResult == 3 ? tok::identifier : TokenCode); 139 Info.setIsExtensionToken(AddResult == 1); 140 Info.setIsCXX11CompatKeyword(AddResult == 3); 141 } 142 143 /// AddCXXOperatorKeyword - Register a C++ operator keyword alternative 144 /// representations. 145 static void AddCXXOperatorKeyword(StringRef Keyword, 146 tok::TokenKind TokenCode, 147 IdentifierTable &Table) { 148 IdentifierInfo &Info = Table.get(Keyword, TokenCode); 149 Info.setIsCPlusPlusOperatorKeyword(); 150 } 151 152 /// AddObjCKeyword - Register an Objective-C @keyword like "class" "selector" or 153 /// "property". 154 static void AddObjCKeyword(StringRef Name, 155 tok::ObjCKeywordKind ObjCID, 156 IdentifierTable &Table) { 157 Table.get(Name).setObjCKeywordID(ObjCID); 158 } 159 160 /// AddKeywords - Add all keywords to the symbol table. 161 /// 162 void IdentifierTable::AddKeywords(const LangOptions &LangOpts) { 163 // Add keywords and tokens for the current language. 164 #define KEYWORD(NAME, FLAGS) \ 165 AddKeyword(StringRef(#NAME), tok::kw_ ## NAME, \ 166 FLAGS, LangOpts, *this); 167 #define ALIAS(NAME, TOK, FLAGS) \ 168 AddKeyword(StringRef(NAME), tok::kw_ ## TOK, \ 169 FLAGS, LangOpts, *this); 170 #define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \ 171 if (LangOpts.CXXOperatorNames) \ 172 AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this); 173 #define OBJC1_AT_KEYWORD(NAME) \ 174 if (LangOpts.ObjC1) \ 175 AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this); 176 #define OBJC2_AT_KEYWORD(NAME) \ 177 if (LangOpts.ObjC2) \ 178 AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this); 179 #define TESTING_KEYWORD(NAME, FLAGS) 180 #include "clang/Basic/TokenKinds.def" 181 182 if (LangOpts.ParseUnknownAnytype) 183 AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL, 184 LangOpts, *this); 185 } 186 187 tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const { 188 // We use a perfect hash function here involving the length of the keyword, 189 // the first and third character. For preprocessor ID's there are no 190 // collisions (if there were, the switch below would complain about duplicate 191 // case values). Note that this depends on 'if' being null terminated. 192 193 #define HASH(LEN, FIRST, THIRD) \ 194 (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31) 195 #define CASE(LEN, FIRST, THIRD, NAME) \ 196 case HASH(LEN, FIRST, THIRD): \ 197 return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME 198 199 unsigned Len = getLength(); 200 if (Len < 2) return tok::pp_not_keyword; 201 const char *Name = getNameStart(); 202 switch (HASH(Len, Name[0], Name[2])) { 203 default: return tok::pp_not_keyword; 204 CASE( 2, 'i', '\0', if); 205 CASE( 4, 'e', 'i', elif); 206 CASE( 4, 'e', 's', else); 207 CASE( 4, 'l', 'n', line); 208 CASE( 4, 's', 'c', sccs); 209 CASE( 5, 'e', 'd', endif); 210 CASE( 5, 'e', 'r', error); 211 CASE( 5, 'i', 'e', ident); 212 CASE( 5, 'i', 'd', ifdef); 213 CASE( 5, 'u', 'd', undef); 214 215 CASE( 6, 'a', 's', assert); 216 CASE( 6, 'd', 'f', define); 217 CASE( 6, 'i', 'n', ifndef); 218 CASE( 6, 'i', 'p', import); 219 CASE( 6, 'p', 'a', pragma); 220 221 CASE( 7, 'd', 'f', defined); 222 CASE( 7, 'i', 'c', include); 223 CASE( 7, 'w', 'r', warning); 224 225 CASE( 8, 'u', 'a', unassert); 226 CASE(12, 'i', 'c', include_next); 227 228 CASE(14, '_', 'p', __public_macro); 229 230 CASE(15, '_', 'p', __private_macro); 231 232 CASE(16, '_', 'i', __include_macros); 233 #undef CASE 234 #undef HASH 235 } 236 } 237 238 //===----------------------------------------------------------------------===// 239 // Stats Implementation 240 //===----------------------------------------------------------------------===// 241 242 /// PrintStats - Print statistics about how well the identifier table is doing 243 /// at hashing identifiers. 244 void IdentifierTable::PrintStats() const { 245 unsigned NumBuckets = HashTable.getNumBuckets(); 246 unsigned NumIdentifiers = HashTable.getNumItems(); 247 unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers; 248 unsigned AverageIdentifierSize = 0; 249 unsigned MaxIdentifierLength = 0; 250 251 // TODO: Figure out maximum times an identifier had to probe for -stats. 252 for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator 253 I = HashTable.begin(), E = HashTable.end(); I != E; ++I) { 254 unsigned IdLen = I->getKeyLength(); 255 AverageIdentifierSize += IdLen; 256 if (MaxIdentifierLength < IdLen) 257 MaxIdentifierLength = IdLen; 258 } 259 260 fprintf(stderr, "\n*** Identifier Table Stats:\n"); 261 fprintf(stderr, "# Identifiers: %d\n", NumIdentifiers); 262 fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets); 263 fprintf(stderr, "Hash density (#identifiers per bucket): %f\n", 264 NumIdentifiers/(double)NumBuckets); 265 fprintf(stderr, "Ave identifier length: %f\n", 266 (AverageIdentifierSize/(double)NumIdentifiers)); 267 fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength); 268 269 // Compute statistics about the memory allocated for identifiers. 270 HashTable.getAllocator().PrintStats(); 271 } 272 273 //===----------------------------------------------------------------------===// 274 // SelectorTable Implementation 275 //===----------------------------------------------------------------------===// 276 277 unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) { 278 return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr()); 279 } 280 281 namespace clang { 282 /// MultiKeywordSelector - One of these variable length records is kept for each 283 /// selector containing more than one keyword. We use a folding set 284 /// to unique aggregate names (keyword selectors in ObjC parlance). Access to 285 /// this class is provided strictly through Selector. 286 class MultiKeywordSelector 287 : public DeclarationNameExtra, public llvm::FoldingSetNode { 288 MultiKeywordSelector(unsigned nKeys) { 289 ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys; 290 } 291 public: 292 // Constructor for keyword selectors. 293 MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) { 294 assert((nKeys > 1) && "not a multi-keyword selector"); 295 ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys; 296 297 // Fill in the trailing keyword array. 298 IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1); 299 for (unsigned i = 0; i != nKeys; ++i) 300 KeyInfo[i] = IIV[i]; 301 } 302 303 // getName - Derive the full selector name and return it. 304 std::string getName() const; 305 306 unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; } 307 308 typedef IdentifierInfo *const *keyword_iterator; 309 keyword_iterator keyword_begin() const { 310 return reinterpret_cast<keyword_iterator>(this+1); 311 } 312 keyword_iterator keyword_end() const { 313 return keyword_begin()+getNumArgs(); 314 } 315 IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const { 316 assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index"); 317 return keyword_begin()[i]; 318 } 319 static void Profile(llvm::FoldingSetNodeID &ID, 320 keyword_iterator ArgTys, unsigned NumArgs) { 321 ID.AddInteger(NumArgs); 322 for (unsigned i = 0; i != NumArgs; ++i) 323 ID.AddPointer(ArgTys[i]); 324 } 325 void Profile(llvm::FoldingSetNodeID &ID) { 326 Profile(ID, keyword_begin(), getNumArgs()); 327 } 328 }; 329 } // end namespace clang. 330 331 unsigned Selector::getNumArgs() const { 332 unsigned IIF = getIdentifierInfoFlag(); 333 if (IIF == ZeroArg) 334 return 0; 335 if (IIF == OneArg) 336 return 1; 337 // We point to a MultiKeywordSelector (pointer doesn't contain any flags). 338 MultiKeywordSelector *SI = reinterpret_cast<MultiKeywordSelector *>(InfoPtr); 339 return SI->getNumArgs(); 340 } 341 342 IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const { 343 if (getIdentifierInfoFlag()) { 344 assert(argIndex == 0 && "illegal keyword index"); 345 return getAsIdentifierInfo(); 346 } 347 // We point to a MultiKeywordSelector (pointer doesn't contain any flags). 348 MultiKeywordSelector *SI = reinterpret_cast<MultiKeywordSelector *>(InfoPtr); 349 return SI->getIdentifierInfoForSlot(argIndex); 350 } 351 352 StringRef Selector::getNameForSlot(unsigned int argIndex) const { 353 IdentifierInfo *II = getIdentifierInfoForSlot(argIndex); 354 return II? II->getName() : StringRef(); 355 } 356 357 std::string MultiKeywordSelector::getName() const { 358 llvm::SmallString<256> Str; 359 llvm::raw_svector_ostream OS(Str); 360 for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) { 361 if (*I) 362 OS << (*I)->getName(); 363 OS << ':'; 364 } 365 366 return OS.str(); 367 } 368 369 std::string Selector::getAsString() const { 370 if (InfoPtr == 0) 371 return "<null selector>"; 372 373 if (InfoPtr & ArgFlags) { 374 IdentifierInfo *II = getAsIdentifierInfo(); 375 376 // If the number of arguments is 0 then II is guaranteed to not be null. 377 if (getNumArgs() == 0) 378 return II->getName(); 379 380 if (!II) 381 return ":"; 382 383 return II->getName().str() + ":"; 384 } 385 386 // We have a multiple keyword selector (no embedded flags). 387 return reinterpret_cast<MultiKeywordSelector *>(InfoPtr)->getName(); 388 } 389 390 /// Interpreting the given string using the normal CamelCase 391 /// conventions, determine whether the given string starts with the 392 /// given "word", which is assumed to end in a lowercase letter. 393 static bool startsWithWord(StringRef name, StringRef word) { 394 if (name.size() < word.size()) return false; 395 return ((name.size() == word.size() || 396 !islower(name[word.size()])) 397 && name.startswith(word)); 398 } 399 400 ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) { 401 IdentifierInfo *first = sel.getIdentifierInfoForSlot(0); 402 if (!first) return OMF_None; 403 404 StringRef name = first->getName(); 405 if (sel.isUnarySelector()) { 406 if (name == "autorelease") return OMF_autorelease; 407 if (name == "dealloc") return OMF_dealloc; 408 if (name == "finalize") return OMF_finalize; 409 if (name == "release") return OMF_release; 410 if (name == "retain") return OMF_retain; 411 if (name == "retainCount") return OMF_retainCount; 412 if (name == "self") return OMF_self; 413 } 414 415 if (name == "performSelector") return OMF_performSelector; 416 417 // The other method families may begin with a prefix of underscores. 418 while (!name.empty() && name.front() == '_') 419 name = name.substr(1); 420 421 if (name.empty()) return OMF_None; 422 switch (name.front()) { 423 case 'a': 424 if (startsWithWord(name, "alloc")) return OMF_alloc; 425 break; 426 case 'c': 427 if (startsWithWord(name, "copy")) return OMF_copy; 428 break; 429 case 'i': 430 if (startsWithWord(name, "init")) return OMF_init; 431 break; 432 case 'm': 433 if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy; 434 break; 435 case 'n': 436 if (startsWithWord(name, "new")) return OMF_new; 437 break; 438 default: 439 break; 440 } 441 442 return OMF_None; 443 } 444 445 namespace { 446 struct SelectorTableImpl { 447 llvm::FoldingSet<MultiKeywordSelector> Table; 448 llvm::BumpPtrAllocator Allocator; 449 }; 450 } // end anonymous namespace. 451 452 static SelectorTableImpl &getSelectorTableImpl(void *P) { 453 return *static_cast<SelectorTableImpl*>(P); 454 } 455 456 size_t SelectorTable::getTotalMemory() const { 457 SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl); 458 return SelTabImpl.Allocator.getTotalMemory(); 459 } 460 461 Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) { 462 if (nKeys < 2) 463 return Selector(IIV[0], nKeys); 464 465 SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl); 466 467 // Unique selector, to guarantee there is one per name. 468 llvm::FoldingSetNodeID ID; 469 MultiKeywordSelector::Profile(ID, IIV, nKeys); 470 471 void *InsertPos = 0; 472 if (MultiKeywordSelector *SI = 473 SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos)) 474 return Selector(SI); 475 476 // MultiKeywordSelector objects are not allocated with new because they have a 477 // variable size array (for parameter types) at the end of them. 478 unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *); 479 MultiKeywordSelector *SI = 480 (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size, 481 llvm::alignOf<MultiKeywordSelector>()); 482 new (SI) MultiKeywordSelector(nKeys, IIV); 483 SelTabImpl.Table.InsertNode(SI, InsertPos); 484 return Selector(SI); 485 } 486 487 SelectorTable::SelectorTable() { 488 Impl = new SelectorTableImpl(); 489 } 490 491 SelectorTable::~SelectorTable() { 492 delete &getSelectorTableImpl(Impl); 493 } 494 495 const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) { 496 switch (Operator) { 497 case OO_None: 498 case NUM_OVERLOADED_OPERATORS: 499 return 0; 500 501 #define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \ 502 case OO_##Name: return Spelling; 503 #include "clang/Basic/OperatorKinds.def" 504 } 505 506 return 0; 507 } 508