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/CharInfo.h"
16#include "clang/Basic/IdentifierTable.h"
17#include "clang/Basic/LangOptions.h"
18#include "clang/Basic/OperatorKinds.h"
19#include "llvm/ADT/DenseMap.h"
20#include "llvm/ADT/FoldingSet.h"
21#include "llvm/ADT/SmallString.h"
22#include "llvm/Support/ErrorHandling.h"
23#include "llvm/Support/raw_ostream.h"
24#include <cstdio>
25
26using namespace clang;
27
28//===----------------------------------------------------------------------===//
29// IdentifierInfo Implementation
30//===----------------------------------------------------------------------===//
31
32IdentifierInfo::IdentifierInfo() {
33  TokenID = tok::identifier;
34  ObjCOrBuiltinID = 0;
35  HasMacro = false;
36  HadMacro = false;
37  IsExtension = false;
38  IsCXX11CompatKeyword = false;
39  IsPoisoned = false;
40  IsCPPOperatorKeyword = false;
41  NeedsHandleIdentifier = false;
42  IsFromAST = false;
43  ChangedAfterLoad = false;
44  RevertedTokenID = false;
45  OutOfDate = false;
46  IsModulesImport = false;
47  FETokenInfo = nullptr;
48  Entry = nullptr;
49}
50
51//===----------------------------------------------------------------------===//
52// IdentifierTable Implementation
53//===----------------------------------------------------------------------===//
54
55IdentifierIterator::~IdentifierIterator() { }
56
57IdentifierInfoLookup::~IdentifierInfoLookup() {}
58
59namespace {
60  /// \brief A simple identifier lookup iterator that represents an
61  /// empty sequence of identifiers.
62  class EmptyLookupIterator : public IdentifierIterator
63  {
64  public:
65    StringRef Next() override { return StringRef(); }
66  };
67}
68
69IdentifierIterator *IdentifierInfoLookup::getIdentifiers() {
70  return new EmptyLookupIterator();
71}
72
73ExternalIdentifierLookup::~ExternalIdentifierLookup() {}
74
75IdentifierTable::IdentifierTable(const LangOptions &LangOpts,
76                                 IdentifierInfoLookup* externalLookup)
77  : HashTable(8192), // Start with space for 8K identifiers.
78    ExternalLookup(externalLookup) {
79
80  // Populate the identifier table with info about keywords for the current
81  // language.
82  AddKeywords(LangOpts);
83
84
85  // Add the '_experimental_modules_import' contextual keyword.
86  get("import").setModulesImport(true);
87}
88
89//===----------------------------------------------------------------------===//
90// Language Keyword Implementation
91//===----------------------------------------------------------------------===//
92
93// Constants for TokenKinds.def
94namespace {
95  enum {
96    KEYC99 = 0x1,
97    KEYCXX = 0x2,
98    KEYCXX11 = 0x4,
99    KEYGNU = 0x8,
100    KEYMS = 0x10,
101    BOOLSUPPORT = 0x20,
102    KEYALTIVEC = 0x40,
103    KEYNOCXX = 0x80,
104    KEYBORLAND = 0x100,
105    KEYOPENCL = 0x200,
106    KEYC11 = 0x400,
107    KEYARC = 0x800,
108    KEYNOMS = 0x01000,
109    WCHARSUPPORT = 0x02000,
110    HALFSUPPORT = 0x04000,
111    KEYALL = (0xffff & ~KEYNOMS) // Because KEYNOMS is used to exclude.
112  };
113}
114
115/// AddKeyword - This method is used to associate a token ID with specific
116/// identifiers because they are language keywords.  This causes the lexer to
117/// automatically map matching identifiers to specialized token codes.
118///
119/// The C90/C99/CPP/CPP0x flags are set to 3 if the token is a keyword in a
120/// future language standard, set to 2 if the token should be enabled in the
121/// specified language, set to 1 if it is an extension in the specified
122/// language, and set to 0 if disabled in the specified language.
123static void AddKeyword(StringRef Keyword,
124                       tok::TokenKind TokenCode, unsigned Flags,
125                       const LangOptions &LangOpts, IdentifierTable &Table) {
126  unsigned AddResult = 0;
127  if (Flags == KEYALL) AddResult = 2;
128  else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2;
129  else if (LangOpts.CPlusPlus11 && (Flags & KEYCXX11)) AddResult = 2;
130  else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2;
131  else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1;
132  else if (LangOpts.MicrosoftExt && (Flags & KEYMS)) AddResult = 1;
133  else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1;
134  else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2;
135  else if (LangOpts.Half && (Flags & HALFSUPPORT)) AddResult = 2;
136  else if (LangOpts.WChar && (Flags & WCHARSUPPORT)) AddResult = 2;
137  else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2;
138  else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2;
139  else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2;
140  else if (LangOpts.C11 && (Flags & KEYC11)) AddResult = 2;
141  // We treat bridge casts as objective-C keywords so we can warn on them
142  // in non-arc mode.
143  else if (LangOpts.ObjC2 && (Flags & KEYARC)) AddResult = 2;
144  else if (LangOpts.CPlusPlus && (Flags & KEYCXX11)) AddResult = 3;
145
146  // Don't add this keyword under MSVCCompat.
147  if (LangOpts.MSVCCompat && (Flags & KEYNOMS))
148     return;
149  // Don't add this keyword if disabled in this language.
150  if (AddResult == 0) return;
151
152  IdentifierInfo &Info =
153      Table.get(Keyword, AddResult == 3 ? tok::identifier : TokenCode);
154  Info.setIsExtensionToken(AddResult == 1);
155  Info.setIsCXX11CompatKeyword(AddResult == 3);
156}
157
158/// AddCXXOperatorKeyword - Register a C++ operator keyword alternative
159/// representations.
160static void AddCXXOperatorKeyword(StringRef Keyword,
161                                  tok::TokenKind TokenCode,
162                                  IdentifierTable &Table) {
163  IdentifierInfo &Info = Table.get(Keyword, TokenCode);
164  Info.setIsCPlusPlusOperatorKeyword();
165}
166
167/// AddObjCKeyword - Register an Objective-C \@keyword like "class" "selector"
168/// or "property".
169static void AddObjCKeyword(StringRef Name,
170                           tok::ObjCKeywordKind ObjCID,
171                           IdentifierTable &Table) {
172  Table.get(Name).setObjCKeywordID(ObjCID);
173}
174
175/// AddKeywords - Add all keywords to the symbol table.
176///
177void IdentifierTable::AddKeywords(const LangOptions &LangOpts) {
178  // Add keywords and tokens for the current language.
179#define KEYWORD(NAME, FLAGS) \
180  AddKeyword(StringRef(#NAME), tok::kw_ ## NAME,  \
181             FLAGS, LangOpts, *this);
182#define ALIAS(NAME, TOK, FLAGS) \
183  AddKeyword(StringRef(NAME), tok::kw_ ## TOK,  \
184             FLAGS, LangOpts, *this);
185#define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \
186  if (LangOpts.CXXOperatorNames)          \
187    AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this);
188#define OBJC1_AT_KEYWORD(NAME) \
189  if (LangOpts.ObjC1)          \
190    AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
191#define OBJC2_AT_KEYWORD(NAME) \
192  if (LangOpts.ObjC2)          \
193    AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
194#define TESTING_KEYWORD(NAME, FLAGS)
195#include "clang/Basic/TokenKinds.def"
196
197  if (LangOpts.ParseUnknownAnytype)
198    AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL,
199               LangOpts, *this);
200}
201
202tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const {
203  // We use a perfect hash function here involving the length of the keyword,
204  // the first and third character.  For preprocessor ID's there are no
205  // collisions (if there were, the switch below would complain about duplicate
206  // case values).  Note that this depends on 'if' being null terminated.
207
208#define HASH(LEN, FIRST, THIRD) \
209  (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31)
210#define CASE(LEN, FIRST, THIRD, NAME) \
211  case HASH(LEN, FIRST, THIRD): \
212    return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME
213
214  unsigned Len = getLength();
215  if (Len < 2) return tok::pp_not_keyword;
216  const char *Name = getNameStart();
217  switch (HASH(Len, Name[0], Name[2])) {
218  default: return tok::pp_not_keyword;
219  CASE( 2, 'i', '\0', if);
220  CASE( 4, 'e', 'i', elif);
221  CASE( 4, 'e', 's', else);
222  CASE( 4, 'l', 'n', line);
223  CASE( 4, 's', 'c', sccs);
224  CASE( 5, 'e', 'd', endif);
225  CASE( 5, 'e', 'r', error);
226  CASE( 5, 'i', 'e', ident);
227  CASE( 5, 'i', 'd', ifdef);
228  CASE( 5, 'u', 'd', undef);
229
230  CASE( 6, 'a', 's', assert);
231  CASE( 6, 'd', 'f', define);
232  CASE( 6, 'i', 'n', ifndef);
233  CASE( 6, 'i', 'p', import);
234  CASE( 6, 'p', 'a', pragma);
235
236  CASE( 7, 'd', 'f', defined);
237  CASE( 7, 'i', 'c', include);
238  CASE( 7, 'w', 'r', warning);
239
240  CASE( 8, 'u', 'a', unassert);
241  CASE(12, 'i', 'c', include_next);
242
243  CASE(14, '_', 'p', __public_macro);
244
245  CASE(15, '_', 'p', __private_macro);
246
247  CASE(16, '_', 'i', __include_macros);
248#undef CASE
249#undef HASH
250  }
251}
252
253//===----------------------------------------------------------------------===//
254// Stats Implementation
255//===----------------------------------------------------------------------===//
256
257/// PrintStats - Print statistics about how well the identifier table is doing
258/// at hashing identifiers.
259void IdentifierTable::PrintStats() const {
260  unsigned NumBuckets = HashTable.getNumBuckets();
261  unsigned NumIdentifiers = HashTable.getNumItems();
262  unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers;
263  unsigned AverageIdentifierSize = 0;
264  unsigned MaxIdentifierLength = 0;
265
266  // TODO: Figure out maximum times an identifier had to probe for -stats.
267  for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator
268       I = HashTable.begin(), E = HashTable.end(); I != E; ++I) {
269    unsigned IdLen = I->getKeyLength();
270    AverageIdentifierSize += IdLen;
271    if (MaxIdentifierLength < IdLen)
272      MaxIdentifierLength = IdLen;
273  }
274
275  fprintf(stderr, "\n*** Identifier Table Stats:\n");
276  fprintf(stderr, "# Identifiers:   %d\n", NumIdentifiers);
277  fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets);
278  fprintf(stderr, "Hash density (#identifiers per bucket): %f\n",
279          NumIdentifiers/(double)NumBuckets);
280  fprintf(stderr, "Ave identifier length: %f\n",
281          (AverageIdentifierSize/(double)NumIdentifiers));
282  fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength);
283
284  // Compute statistics about the memory allocated for identifiers.
285  HashTable.getAllocator().PrintStats();
286}
287
288//===----------------------------------------------------------------------===//
289// SelectorTable Implementation
290//===----------------------------------------------------------------------===//
291
292unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) {
293  return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr());
294}
295
296namespace clang {
297/// MultiKeywordSelector - One of these variable length records is kept for each
298/// selector containing more than one keyword. We use a folding set
299/// to unique aggregate names (keyword selectors in ObjC parlance). Access to
300/// this class is provided strictly through Selector.
301class MultiKeywordSelector
302  : public DeclarationNameExtra, public llvm::FoldingSetNode {
303  MultiKeywordSelector(unsigned nKeys) {
304    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
305  }
306public:
307  // Constructor for keyword selectors.
308  MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) {
309    assert((nKeys > 1) && "not a multi-keyword selector");
310    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
311
312    // Fill in the trailing keyword array.
313    IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1);
314    for (unsigned i = 0; i != nKeys; ++i)
315      KeyInfo[i] = IIV[i];
316  }
317
318  // getName - Derive the full selector name and return it.
319  std::string getName() const;
320
321  unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; }
322
323  typedef IdentifierInfo *const *keyword_iterator;
324  keyword_iterator keyword_begin() const {
325    return reinterpret_cast<keyword_iterator>(this+1);
326  }
327  keyword_iterator keyword_end() const {
328    return keyword_begin()+getNumArgs();
329  }
330  IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const {
331    assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index");
332    return keyword_begin()[i];
333  }
334  static void Profile(llvm::FoldingSetNodeID &ID,
335                      keyword_iterator ArgTys, unsigned NumArgs) {
336    ID.AddInteger(NumArgs);
337    for (unsigned i = 0; i != NumArgs; ++i)
338      ID.AddPointer(ArgTys[i]);
339  }
340  void Profile(llvm::FoldingSetNodeID &ID) {
341    Profile(ID, keyword_begin(), getNumArgs());
342  }
343};
344} // end namespace clang.
345
346unsigned Selector::getNumArgs() const {
347  unsigned IIF = getIdentifierInfoFlag();
348  if (IIF <= ZeroArg)
349    return 0;
350  if (IIF == OneArg)
351    return 1;
352  // We point to a MultiKeywordSelector.
353  MultiKeywordSelector *SI = getMultiKeywordSelector();
354  return SI->getNumArgs();
355}
356
357IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const {
358  if (getIdentifierInfoFlag() < MultiArg) {
359    assert(argIndex == 0 && "illegal keyword index");
360    return getAsIdentifierInfo();
361  }
362  // We point to a MultiKeywordSelector.
363  MultiKeywordSelector *SI = getMultiKeywordSelector();
364  return SI->getIdentifierInfoForSlot(argIndex);
365}
366
367StringRef Selector::getNameForSlot(unsigned int argIndex) const {
368  IdentifierInfo *II = getIdentifierInfoForSlot(argIndex);
369  return II? II->getName() : StringRef();
370}
371
372std::string MultiKeywordSelector::getName() const {
373  SmallString<256> Str;
374  llvm::raw_svector_ostream OS(Str);
375  for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) {
376    if (*I)
377      OS << (*I)->getName();
378    OS << ':';
379  }
380
381  return OS.str();
382}
383
384std::string Selector::getAsString() const {
385  if (InfoPtr == 0)
386    return "<null selector>";
387
388  if (getIdentifierInfoFlag() < MultiArg) {
389    IdentifierInfo *II = getAsIdentifierInfo();
390
391    // If the number of arguments is 0 then II is guaranteed to not be null.
392    if (getNumArgs() == 0)
393      return II->getName();
394
395    if (!II)
396      return ":";
397
398    return II->getName().str() + ":";
399  }
400
401  // We have a multiple keyword selector.
402  return getMultiKeywordSelector()->getName();
403}
404
405void Selector::print(llvm::raw_ostream &OS) const {
406  OS << getAsString();
407}
408
409/// Interpreting the given string using the normal CamelCase
410/// conventions, determine whether the given string starts with the
411/// given "word", which is assumed to end in a lowercase letter.
412static bool startsWithWord(StringRef name, StringRef word) {
413  if (name.size() < word.size()) return false;
414  return ((name.size() == word.size() || !isLowercase(name[word.size()])) &&
415          name.startswith(word));
416}
417
418ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) {
419  IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
420  if (!first) return OMF_None;
421
422  StringRef name = first->getName();
423  if (sel.isUnarySelector()) {
424    if (name == "autorelease") return OMF_autorelease;
425    if (name == "dealloc") return OMF_dealloc;
426    if (name == "finalize") return OMF_finalize;
427    if (name == "release") return OMF_release;
428    if (name == "retain") return OMF_retain;
429    if (name == "retainCount") return OMF_retainCount;
430    if (name == "self") return OMF_self;
431  }
432
433  if (name == "performSelector") return OMF_performSelector;
434
435  // The other method families may begin with a prefix of underscores.
436  while (!name.empty() && name.front() == '_')
437    name = name.substr(1);
438
439  if (name.empty()) return OMF_None;
440  switch (name.front()) {
441  case 'a':
442    if (startsWithWord(name, "alloc")) return OMF_alloc;
443    break;
444  case 'c':
445    if (startsWithWord(name, "copy")) return OMF_copy;
446    break;
447  case 'i':
448    if (startsWithWord(name, "init")) return OMF_init;
449    break;
450  case 'm':
451    if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy;
452    break;
453  case 'n':
454    if (startsWithWord(name, "new")) return OMF_new;
455    break;
456  default:
457    break;
458  }
459
460  return OMF_None;
461}
462
463ObjCInstanceTypeFamily Selector::getInstTypeMethodFamily(Selector sel) {
464  IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
465  if (!first) return OIT_None;
466
467  StringRef name = first->getName();
468
469  if (name.empty()) return OIT_None;
470  switch (name.front()) {
471    case 'a':
472      if (startsWithWord(name, "array")) return OIT_Array;
473      break;
474    case 'd':
475      if (startsWithWord(name, "default")) return OIT_ReturnsSelf;
476      if (startsWithWord(name, "dictionary")) return OIT_Dictionary;
477      break;
478    case 's':
479      if (startsWithWord(name, "shared")) return OIT_ReturnsSelf;
480      if (startsWithWord(name, "standard")) return OIT_Singleton;
481    case 'i':
482      if (startsWithWord(name, "init")) return OIT_Init;
483    default:
484      break;
485  }
486  return OIT_None;
487}
488
489namespace {
490  struct SelectorTableImpl {
491    llvm::FoldingSet<MultiKeywordSelector> Table;
492    llvm::BumpPtrAllocator Allocator;
493  };
494} // end anonymous namespace.
495
496static SelectorTableImpl &getSelectorTableImpl(void *P) {
497  return *static_cast<SelectorTableImpl*>(P);
498}
499
500SmallString<64>
501SelectorTable::constructSetterName(StringRef Name) {
502  SmallString<64> SetterName("set");
503  SetterName += Name;
504  SetterName[3] = toUppercase(SetterName[3]);
505  return SetterName;
506}
507
508Selector
509SelectorTable::constructSetterSelector(IdentifierTable &Idents,
510                                       SelectorTable &SelTable,
511                                       const IdentifierInfo *Name) {
512  IdentifierInfo *SetterName =
513    &Idents.get(constructSetterName(Name->getName()));
514  return SelTable.getUnarySelector(SetterName);
515}
516
517size_t SelectorTable::getTotalMemory() const {
518  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
519  return SelTabImpl.Allocator.getTotalMemory();
520}
521
522Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) {
523  if (nKeys < 2)
524    return Selector(IIV[0], nKeys);
525
526  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
527
528  // Unique selector, to guarantee there is one per name.
529  llvm::FoldingSetNodeID ID;
530  MultiKeywordSelector::Profile(ID, IIV, nKeys);
531
532  void *InsertPos = nullptr;
533  if (MultiKeywordSelector *SI =
534        SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos))
535    return Selector(SI);
536
537  // MultiKeywordSelector objects are not allocated with new because they have a
538  // variable size array (for parameter types) at the end of them.
539  unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *);
540  MultiKeywordSelector *SI =
541    (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size,
542                                         llvm::alignOf<MultiKeywordSelector>());
543  new (SI) MultiKeywordSelector(nKeys, IIV);
544  SelTabImpl.Table.InsertNode(SI, InsertPos);
545  return Selector(SI);
546}
547
548SelectorTable::SelectorTable() {
549  Impl = new SelectorTableImpl();
550}
551
552SelectorTable::~SelectorTable() {
553  delete &getSelectorTableImpl(Impl);
554}
555
556const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) {
557  switch (Operator) {
558  case OO_None:
559  case NUM_OVERLOADED_OPERATORS:
560    return nullptr;
561
562#define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \
563  case OO_##Name: return Spelling;
564#include "clang/Basic/OperatorKinds.def"
565  }
566
567  llvm_unreachable("Invalid OverloadedOperatorKind!");
568}
569