Home | History | Annotate | Download | only in Basic 
      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/CharInfo.h"
     17 #include "clang/Basic/LangOptions.h"
     18 #include "llvm/ADT/DenseMap.h"
     19 #include "llvm/ADT/FoldingSet.h"
     20 #include "llvm/ADT/SmallString.h"
     21 #include "llvm/Support/ErrorHandling.h"
     22 #include "llvm/Support/raw_ostream.h"
     23 #include <cstdio>
     24 
     25 using namespace clang;
     26 
     27 //===----------------------------------------------------------------------===//
     28 // IdentifierInfo Implementation
     29 //===----------------------------------------------------------------------===//
     30 
     31 IdentifierInfo::IdentifierInfo() {
     32   TokenID = tok::identifier;
     33   ObjCOrBuiltinID = 0;
     34   HasMacro = false;
     35   HadMacro = 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("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     KEYCXX11 = 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     WCHARSUPPORT = 0x02000,
    109     KEYALL = (0xffff & ~KEYNOMS) // Because KEYNOMS is used to exclude.
    110   };
    111 }
    112 
    113 /// AddKeyword - This method is used to associate a token ID with specific
    114 /// identifiers because they are language keywords.  This causes the lexer to
    115 /// automatically map matching identifiers to specialized token codes.
    116 ///
    117 /// The C90/C99/CPP/CPP0x flags are set to 3 if the token is a keyword in a
    118 /// future language standard, set to 2 if the token should be enabled in the
    119 /// specified language, set to 1 if it is an extension in the specified
    120 /// language, and set to 0 if disabled in the specified language.
    121 static void AddKeyword(StringRef Keyword,
    122                        tok::TokenKind TokenCode, unsigned Flags,
    123                        const LangOptions &LangOpts, IdentifierTable &Table) {
    124   unsigned AddResult = 0;
    125   if (Flags == KEYALL) AddResult = 2;
    126   else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2;
    127   else if (LangOpts.CPlusPlus11 && (Flags & KEYCXX11)) AddResult = 2;
    128   else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2;
    129   else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1;
    130   else if (LangOpts.MicrosoftExt && (Flags & KEYMS)) AddResult = 1;
    131   else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1;
    132   else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2;
    133   else if (LangOpts.WChar && (Flags & WCHARSUPPORT)) AddResult = 2;
    134   else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2;
    135   else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2;
    136   else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2;
    137   else if (LangOpts.C11 && (Flags & KEYC11)) AddResult = 2;
    138   // We treat bridge casts as objective-C keywords so we can warn on them
    139   // in non-arc mode.
    140   else if (LangOpts.ObjC2 && (Flags & KEYARC)) AddResult = 2;
    141   else if (LangOpts.CPlusPlus && (Flags & KEYCXX11)) AddResult = 3;
    142 
    143   // Don't add this keyword under MicrosoftMode.
    144   if (LangOpts.MicrosoftMode && (Flags & KEYNOMS))
    145      return;
    146   // Don't add this keyword if disabled in this language.
    147   if (AddResult == 0) return;
    148 
    149   IdentifierInfo &Info =
    150       Table.get(Keyword, AddResult == 3 ? tok::identifier : TokenCode);
    151   Info.setIsExtensionToken(AddResult == 1);
    152   Info.setIsCXX11CompatKeyword(AddResult == 3);
    153 }
    154 
    155 /// AddCXXOperatorKeyword - Register a C++ operator keyword alternative
    156 /// representations.
    157 static void AddCXXOperatorKeyword(StringRef Keyword,
    158                                   tok::TokenKind TokenCode,
    159                                   IdentifierTable &Table) {
    160   IdentifierInfo &Info = Table.get(Keyword, TokenCode);
    161   Info.setIsCPlusPlusOperatorKeyword();
    162 }
    163 
    164 /// AddObjCKeyword - Register an Objective-C \@keyword like "class" "selector"
    165 /// or "property".
    166 static void AddObjCKeyword(StringRef Name,
    167                            tok::ObjCKeywordKind ObjCID,
    168                            IdentifierTable &Table) {
    169   Table.get(Name).setObjCKeywordID(ObjCID);
    170 }
    171 
    172 /// AddKeywords - Add all keywords to the symbol table.
    173 ///
    174 void IdentifierTable::AddKeywords(const LangOptions &LangOpts) {
    175   // Add keywords and tokens for the current language.
    176 #define KEYWORD(NAME, FLAGS) \
    177   AddKeyword(StringRef(#NAME), tok::kw_ ## NAME,  \
    178              FLAGS, LangOpts, *this);
    179 #define ALIAS(NAME, TOK, FLAGS) \
    180   AddKeyword(StringRef(NAME), tok::kw_ ## TOK,  \
    181              FLAGS, LangOpts, *this);
    182 #define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \
    183   if (LangOpts.CXXOperatorNames)          \
    184     AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this);
    185 #define OBJC1_AT_KEYWORD(NAME) \
    186   if (LangOpts.ObjC1)          \
    187     AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
    188 #define OBJC2_AT_KEYWORD(NAME) \
    189   if (LangOpts.ObjC2)          \
    190     AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
    191 #define TESTING_KEYWORD(NAME, FLAGS)
    192 #include "clang/Basic/TokenKinds.def"
    193 
    194   if (LangOpts.ParseUnknownAnytype)
    195     AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL,
    196                LangOpts, *this);
    197 }
    198 
    199 tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const {
    200   // We use a perfect hash function here involving the length of the keyword,
    201   // the first and third character.  For preprocessor ID's there are no
    202   // collisions (if there were, the switch below would complain about duplicate
    203   // case values).  Note that this depends on 'if' being null terminated.
    204 
    205 #define HASH(LEN, FIRST, THIRD) \
    206   (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31)
    207 #define CASE(LEN, FIRST, THIRD, NAME) \
    208   case HASH(LEN, FIRST, THIRD): \
    209     return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME
    210 
    211   unsigned Len = getLength();
    212   if (Len < 2) return tok::pp_not_keyword;
    213   const char *Name = getNameStart();
    214   switch (HASH(Len, Name[0], Name[2])) {
    215   default: return tok::pp_not_keyword;
    216   CASE( 2, 'i', '\0', if);
    217   CASE( 4, 'e', 'i', elif);
    218   CASE( 4, 'e', 's', else);
    219   CASE( 4, 'l', 'n', line);
    220   CASE( 4, 's', 'c', sccs);
    221   CASE( 5, 'e', 'd', endif);
    222   CASE( 5, 'e', 'r', error);
    223   CASE( 5, 'i', 'e', ident);
    224   CASE( 5, 'i', 'd', ifdef);
    225   CASE( 5, 'u', 'd', undef);
    226 
    227   CASE( 6, 'a', 's', assert);
    228   CASE( 6, 'd', 'f', define);
    229   CASE( 6, 'i', 'n', ifndef);
    230   CASE( 6, 'i', 'p', import);
    231   CASE( 6, 'p', 'a', pragma);
    232 
    233   CASE( 7, 'd', 'f', defined);
    234   CASE( 7, 'i', 'c', include);
    235   CASE( 7, 'w', 'r', warning);
    236 
    237   CASE( 8, 'u', 'a', unassert);
    238   CASE(12, 'i', 'c', include_next);
    239 
    240   CASE(14, '_', 'p', __public_macro);
    241 
    242   CASE(15, '_', 'p', __private_macro);
    243 
    244   CASE(16, '_', 'i', __include_macros);
    245 #undef CASE
    246 #undef HASH
    247   }
    248 }
    249 
    250 //===----------------------------------------------------------------------===//
    251 // Stats Implementation
    252 //===----------------------------------------------------------------------===//
    253 
    254 /// PrintStats - Print statistics about how well the identifier table is doing
    255 /// at hashing identifiers.
    256 void IdentifierTable::PrintStats() const {
    257   unsigned NumBuckets = HashTable.getNumBuckets();
    258   unsigned NumIdentifiers = HashTable.getNumItems();
    259   unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers;
    260   unsigned AverageIdentifierSize = 0;
    261   unsigned MaxIdentifierLength = 0;
    262 
    263   // TODO: Figure out maximum times an identifier had to probe for -stats.
    264   for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator
    265        I = HashTable.begin(), E = HashTable.end(); I != E; ++I) {
    266     unsigned IdLen = I->getKeyLength();
    267     AverageIdentifierSize += IdLen;
    268     if (MaxIdentifierLength < IdLen)
    269       MaxIdentifierLength = IdLen;
    270   }
    271 
    272   fprintf(stderr, "\n*** Identifier Table Stats:\n");
    273   fprintf(stderr, "# Identifiers:   %d\n", NumIdentifiers);
    274   fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets);
    275   fprintf(stderr, "Hash density (#identifiers per bucket): %f\n",
    276           NumIdentifiers/(double)NumBuckets);
    277   fprintf(stderr, "Ave identifier length: %f\n",
    278           (AverageIdentifierSize/(double)NumIdentifiers));
    279   fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength);
    280 
    281   // Compute statistics about the memory allocated for identifiers.
    282   HashTable.getAllocator().PrintStats();
    283 }
    284 
    285 //===----------------------------------------------------------------------===//
    286 // SelectorTable Implementation
    287 //===----------------------------------------------------------------------===//
    288 
    289 unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) {
    290   return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr());
    291 }
    292 
    293 namespace clang {
    294 /// MultiKeywordSelector - One of these variable length records is kept for each
    295 /// selector containing more than one keyword. We use a folding set
    296 /// to unique aggregate names (keyword selectors in ObjC parlance). Access to
    297 /// this class is provided strictly through Selector.
    298 class MultiKeywordSelector
    299   : public DeclarationNameExtra, public llvm::FoldingSetNode {
    300   MultiKeywordSelector(unsigned nKeys) {
    301     ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
    302   }
    303 public:
    304   // Constructor for keyword selectors.
    305   MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) {
    306     assert((nKeys > 1) && "not a multi-keyword selector");
    307     ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
    308 
    309     // Fill in the trailing keyword array.
    310     IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1);
    311     for (unsigned i = 0; i != nKeys; ++i)
    312       KeyInfo[i] = IIV[i];
    313   }
    314 
    315   // getName - Derive the full selector name and return it.
    316   std::string getName() const;
    317 
    318   unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; }
    319 
    320   typedef IdentifierInfo *const *keyword_iterator;
    321   keyword_iterator keyword_begin() const {
    322     return reinterpret_cast<keyword_iterator>(this+1);
    323   }
    324   keyword_iterator keyword_end() const {
    325     return keyword_begin()+getNumArgs();
    326   }
    327   IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const {
    328     assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index");
    329     return keyword_begin()[i];
    330   }
    331   static void Profile(llvm::FoldingSetNodeID &ID,
    332                       keyword_iterator ArgTys, unsigned NumArgs) {
    333     ID.AddInteger(NumArgs);
    334     for (unsigned i = 0; i != NumArgs; ++i)
    335       ID.AddPointer(ArgTys[i]);
    336   }
    337   void Profile(llvm::FoldingSetNodeID &ID) {
    338     Profile(ID, keyword_begin(), getNumArgs());
    339   }
    340 };
    341 } // end namespace clang.
    342 
    343 unsigned Selector::getNumArgs() const {
    344   unsigned IIF = getIdentifierInfoFlag();
    345   if (IIF <= ZeroArg)
    346     return 0;
    347   if (IIF == OneArg)
    348     return 1;
    349   // We point to a MultiKeywordSelector.
    350   MultiKeywordSelector *SI = getMultiKeywordSelector();
    351   return SI->getNumArgs();
    352 }
    353 
    354 IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const {
    355   if (getIdentifierInfoFlag() < MultiArg) {
    356     assert(argIndex == 0 && "illegal keyword index");
    357     return getAsIdentifierInfo();
    358   }
    359   // We point to a MultiKeywordSelector.
    360   MultiKeywordSelector *SI = getMultiKeywordSelector();
    361   return SI->getIdentifierInfoForSlot(argIndex);
    362 }
    363 
    364 StringRef Selector::getNameForSlot(unsigned int argIndex) const {
    365   IdentifierInfo *II = getIdentifierInfoForSlot(argIndex);
    366   return II? II->getName() : StringRef();
    367 }
    368 
    369 std::string MultiKeywordSelector::getName() const {
    370   SmallString<256> Str;
    371   llvm::raw_svector_ostream OS(Str);
    372   for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) {
    373     if (*I)
    374       OS << (*I)->getName();
    375     OS << ':';
    376   }
    377 
    378   return OS.str();
    379 }
    380 
    381 std::string Selector::getAsString() const {
    382   if (InfoPtr == 0)
    383     return "<null selector>";
    384 
    385   if (getIdentifierInfoFlag() < MultiArg) {
    386     IdentifierInfo *II = getAsIdentifierInfo();
    387 
    388     // If the number of arguments is 0 then II is guaranteed to not be null.
    389     if (getNumArgs() == 0)
    390       return II->getName();
    391 
    392     if (!II)
    393       return ":";
    394 
    395     return II->getName().str() + ":";
    396   }
    397 
    398   // We have a multiple keyword selector.
    399   return getMultiKeywordSelector()->getName();
    400 }
    401 
    402 /// Interpreting the given string using the normal CamelCase
    403 /// conventions, determine whether the given string starts with the
    404 /// given "word", which is assumed to end in a lowercase letter.
    405 static bool startsWithWord(StringRef name, StringRef word) {
    406   if (name.size() < word.size()) return false;
    407   return ((name.size() == word.size() || !isLowercase(name[word.size()])) &&
    408           name.startswith(word));
    409 }
    410 
    411 ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) {
    412   IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
    413   if (!first) return OMF_None;
    414 
    415   StringRef name = first->getName();
    416   if (sel.isUnarySelector()) {
    417     if (name == "autorelease") return OMF_autorelease;
    418     if (name == "dealloc") return OMF_dealloc;
    419     if (name == "finalize") return OMF_finalize;
    420     if (name == "release") return OMF_release;
    421     if (name == "retain") return OMF_retain;
    422     if (name == "retainCount") return OMF_retainCount;
    423     if (name == "self") return OMF_self;
    424   }
    425 
    426   if (name == "performSelector") return OMF_performSelector;
    427 
    428   // The other method families may begin with a prefix of underscores.
    429   while (!name.empty() && name.front() == '_')
    430     name = name.substr(1);
    431 
    432   if (name.empty()) return OMF_None;
    433   switch (name.front()) {
    434   case 'a':
    435     if (startsWithWord(name, "alloc")) return OMF_alloc;
    436     break;
    437   case 'c':
    438     if (startsWithWord(name, "copy")) return OMF_copy;
    439     break;
    440   case 'i':
    441     if (startsWithWord(name, "init")) return OMF_init;
    442     break;
    443   case 'm':
    444     if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy;
    445     break;
    446   case 'n':
    447     if (startsWithWord(name, "new")) return OMF_new;
    448     break;
    449   default:
    450     break;
    451   }
    452 
    453   return OMF_None;
    454 }
    455 
    456 namespace {
    457   struct SelectorTableImpl {
    458     llvm::FoldingSet<MultiKeywordSelector> Table;
    459     llvm::BumpPtrAllocator Allocator;
    460   };
    461 } // end anonymous namespace.
    462 
    463 static SelectorTableImpl &getSelectorTableImpl(void *P) {
    464   return *static_cast<SelectorTableImpl*>(P);
    465 }
    466 
    467 /*static*/ Selector
    468 SelectorTable::constructSetterName(IdentifierTable &Idents,
    469                                    SelectorTable &SelTable,
    470                                    const IdentifierInfo *Name) {
    471   SmallString<100> SelectorName;
    472   SelectorName = "set";
    473   SelectorName += Name->getName();
    474   SelectorName[3] = toUppercase(SelectorName[3]);
    475   IdentifierInfo *SetterName = &Idents.get(SelectorName);
    476   return SelTable.getUnarySelector(SetterName);
    477 }
    478 
    479 size_t SelectorTable::getTotalMemory() const {
    480   SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
    481   return SelTabImpl.Allocator.getTotalMemory();
    482 }
    483 
    484 Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) {
    485   if (nKeys < 2)
    486     return Selector(IIV[0], nKeys);
    487 
    488   SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
    489 
    490   // Unique selector, to guarantee there is one per name.
    491   llvm::FoldingSetNodeID ID;
    492   MultiKeywordSelector::Profile(ID, IIV, nKeys);
    493 
    494   void *InsertPos = 0;
    495   if (MultiKeywordSelector *SI =
    496         SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos))
    497     return Selector(SI);
    498 
    499   // MultiKeywordSelector objects are not allocated with new because they have a
    500   // variable size array (for parameter types) at the end of them.
    501   unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *);
    502   MultiKeywordSelector *SI =
    503     (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size,
    504                                          llvm::alignOf<MultiKeywordSelector>());
    505   new (SI) MultiKeywordSelector(nKeys, IIV);
    506   SelTabImpl.Table.InsertNode(SI, InsertPos);
    507   return Selector(SI);
    508 }
    509 
    510 SelectorTable::SelectorTable() {
    511   Impl = new SelectorTableImpl();
    512 }
    513 
    514 SelectorTable::~SelectorTable() {
    515   delete &getSelectorTableImpl(Impl);
    516 }
    517 
    518 const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) {
    519   switch (Operator) {
    520   case OO_None:
    521   case NUM_OVERLOADED_OPERATORS:
    522     return 0;
    523 
    524 #define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \
    525   case OO_##Name: return Spelling;
    526 #include "clang/Basic/OperatorKinds.def"
    527   }
    528 
    529   llvm_unreachable("Invalid OverloadedOperatorKind!");
    530 }
    531