IdentifierTable.cpp revision 4d7e0ced7f16a04aabe2d8d91cbbb52fb1162810 
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
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 = 0;
48  Entry = 0;
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    virtual StringRef Next() { return StringRef(); }
66  };
67}
68
69IdentifierIterator *IdentifierInfoLookup::getIdentifiers() const {
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("__experimental_modules_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    KEYCXX0X = 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    KEYALL = (0xffff & ~KEYNOMS) // Because KEYNOMS is used to exclude.
111  };
112}
113
114/// AddKeyword - This method is used to associate a token ID with specific
115/// identifiers because they are language keywords.  This causes the lexer to
116/// automatically map matching identifiers to specialized token codes.
117///
118/// The C90/C99/CPP/CPP0x flags are set to 3 if the token is a keyword in a
119/// future language standard, set to 2 if the token should be enabled in the
120/// specified language, set to 1 if it is an extension in the specified
121/// language, and set to 0 if disabled in the specified language.
122static void AddKeyword(StringRef Keyword,
123                       tok::TokenKind TokenCode, unsigned Flags,
124                       const LangOptions &LangOpts, IdentifierTable &Table) {
125  unsigned AddResult = 0;
126  if (Flags == KEYALL) AddResult = 2;
127  else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2;
128  else if (LangOpts.CPlusPlus0x && (Flags & KEYCXX0X)) AddResult = 2;
129  else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2;
130  else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1;
131  else if (LangOpts.MicrosoftExt && (Flags & KEYMS)) AddResult = 1;
132  else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1;
133  else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2;
134  else if (LangOpts.WChar && (Flags & WCHARSUPPORT)) AddResult = 2;
135  else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2;
136  else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2;
137  else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2;
138  else if (LangOpts.C11 && (Flags & KEYC11)) AddResult = 2;
139  // We treat bridge casts as objective-C keywords so we can warn on them
140  // in non-arc mode.
141  else if (LangOpts.ObjC2 && (Flags & KEYARC)) AddResult = 2;
142  else if (LangOpts.CPlusPlus && (Flags & KEYCXX0X)) AddResult = 3;
143
144  // Don't add this keyword under MicrosoftMode.
145  if (LangOpts.MicrosoftMode && (Flags & KEYNOMS))
146     return;
147  // Don't add this keyword if disabled in this language.
148  if (AddResult == 0) return;
149
150  IdentifierInfo &Info =
151      Table.get(Keyword, AddResult == 3 ? tok::identifier : TokenCode);
152  Info.setIsExtensionToken(AddResult == 1);
153  Info.setIsCXX11CompatKeyword(AddResult == 3);
154}
155
156/// AddCXXOperatorKeyword - Register a C++ operator keyword alternative
157/// representations.
158static void AddCXXOperatorKeyword(StringRef Keyword,
159                                  tok::TokenKind TokenCode,
160                                  IdentifierTable &Table) {
161  IdentifierInfo &Info = Table.get(Keyword, TokenCode);
162  Info.setIsCPlusPlusOperatorKeyword();
163}
164
165/// AddObjCKeyword - Register an Objective-C \@keyword like "class" "selector"
166/// or "property".
167static void AddObjCKeyword(StringRef Name,
168                           tok::ObjCKeywordKind ObjCID,
169                           IdentifierTable &Table) {
170  Table.get(Name).setObjCKeywordID(ObjCID);
171}
172
173/// AddKeywords - Add all keywords to the symbol table.
174///
175void IdentifierTable::AddKeywords(const LangOptions &LangOpts) {
176  // Add keywords and tokens for the current language.
177#define KEYWORD(NAME, FLAGS) \
178  AddKeyword(StringRef(#NAME), tok::kw_ ## NAME,  \
179             FLAGS, LangOpts, *this);
180#define ALIAS(NAME, TOK, FLAGS) \
181  AddKeyword(StringRef(NAME), tok::kw_ ## TOK,  \
182             FLAGS, LangOpts, *this);
183#define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \
184  if (LangOpts.CXXOperatorNames)          \
185    AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this);
186#define OBJC1_AT_KEYWORD(NAME) \
187  if (LangOpts.ObjC1)          \
188    AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
189#define OBJC2_AT_KEYWORD(NAME) \
190  if (LangOpts.ObjC2)          \
191    AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
192#define TESTING_KEYWORD(NAME, FLAGS)
193#include "clang/Basic/TokenKinds.def"
194
195  if (LangOpts.ParseUnknownAnytype)
196    AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL,
197               LangOpts, *this);
198}
199
200tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const {
201  // We use a perfect hash function here involving the length of the keyword,
202  // the first and third character.  For preprocessor ID's there are no
203  // collisions (if there were, the switch below would complain about duplicate
204  // case values).  Note that this depends on 'if' being null terminated.
205
206#define HASH(LEN, FIRST, THIRD) \
207  (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31)
208#define CASE(LEN, FIRST, THIRD, NAME) \
209  case HASH(LEN, FIRST, THIRD): \
210    return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME
211
212  unsigned Len = getLength();
213  if (Len < 2) return tok::pp_not_keyword;
214  const char *Name = getNameStart();
215  switch (HASH(Len, Name[0], Name[2])) {
216  default: return tok::pp_not_keyword;
217  CASE( 2, 'i', '\0', if);
218  CASE( 4, 'e', 'i', elif);
219  CASE( 4, 'e', 's', else);
220  CASE( 4, 'l', 'n', line);
221  CASE( 4, 's', 'c', sccs);
222  CASE( 5, 'e', 'd', endif);
223  CASE( 5, 'e', 'r', error);
224  CASE( 5, 'i', 'e', ident);
225  CASE( 5, 'i', 'd', ifdef);
226  CASE( 5, 'u', 'd', undef);
227
228  CASE( 6, 'a', 's', assert);
229  CASE( 6, 'd', 'f', define);
230  CASE( 6, 'i', 'n', ifndef);
231  CASE( 6, 'i', 'p', import);
232  CASE( 6, 'p', 'a', pragma);
233
234  CASE( 7, 'd', 'f', defined);
235  CASE( 7, 'i', 'c', include);
236  CASE( 7, 'w', 'r', warning);
237
238  CASE( 8, 'u', 'a', unassert);
239  CASE(12, 'i', 'c', include_next);
240
241  CASE(14, '_', 'p', __public_macro);
242
243  CASE(15, '_', 'p', __private_macro);
244
245  CASE(16, '_', 'i', __include_macros);
246#undef CASE
247#undef HASH
248  }
249}
250
251//===----------------------------------------------------------------------===//
252// Stats Implementation
253//===----------------------------------------------------------------------===//
254
255/// PrintStats - Print statistics about how well the identifier table is doing
256/// at hashing identifiers.
257void IdentifierTable::PrintStats() const {
258  unsigned NumBuckets = HashTable.getNumBuckets();
259  unsigned NumIdentifiers = HashTable.getNumItems();
260  unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers;
261  unsigned AverageIdentifierSize = 0;
262  unsigned MaxIdentifierLength = 0;
263
264  // TODO: Figure out maximum times an identifier had to probe for -stats.
265  for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator
266       I = HashTable.begin(), E = HashTable.end(); I != E; ++I) {
267    unsigned IdLen = I->getKeyLength();
268    AverageIdentifierSize += IdLen;
269    if (MaxIdentifierLength < IdLen)
270      MaxIdentifierLength = IdLen;
271  }
272
273  fprintf(stderr, "\n*** Identifier Table Stats:\n");
274  fprintf(stderr, "# Identifiers:   %d\n", NumIdentifiers);
275  fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets);
276  fprintf(stderr, "Hash density (#identifiers per bucket): %f\n",
277          NumIdentifiers/(double)NumBuckets);
278  fprintf(stderr, "Ave identifier length: %f\n",
279          (AverageIdentifierSize/(double)NumIdentifiers));
280  fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength);
281
282  // Compute statistics about the memory allocated for identifiers.
283  HashTable.getAllocator().PrintStats();
284}
285
286//===----------------------------------------------------------------------===//
287// SelectorTable Implementation
288//===----------------------------------------------------------------------===//
289
290unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) {
291  return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr());
292}
293
294namespace clang {
295/// MultiKeywordSelector - One of these variable length records is kept for each
296/// selector containing more than one keyword. We use a folding set
297/// to unique aggregate names (keyword selectors in ObjC parlance). Access to
298/// this class is provided strictly through Selector.
299class MultiKeywordSelector
300  : public DeclarationNameExtra, public llvm::FoldingSetNode {
301  MultiKeywordSelector(unsigned nKeys) {
302    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
303  }
304public:
305  // Constructor for keyword selectors.
306  MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) {
307    assert((nKeys > 1) && "not a multi-keyword selector");
308    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
309
310    // Fill in the trailing keyword array.
311    IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1);
312    for (unsigned i = 0; i != nKeys; ++i)
313      KeyInfo[i] = IIV[i];
314  }
315
316  // getName - Derive the full selector name and return it.
317  std::string getName() const;
318
319  unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; }
320
321  typedef IdentifierInfo *const *keyword_iterator;
322  keyword_iterator keyword_begin() const {
323    return reinterpret_cast<keyword_iterator>(this+1);
324  }
325  keyword_iterator keyword_end() const {
326    return keyword_begin()+getNumArgs();
327  }
328  IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const {
329    assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index");
330    return keyword_begin()[i];
331  }
332  static void Profile(llvm::FoldingSetNodeID &ID,
333                      keyword_iterator ArgTys, unsigned NumArgs) {
334    ID.AddInteger(NumArgs);
335    for (unsigned i = 0; i != NumArgs; ++i)
336      ID.AddPointer(ArgTys[i]);
337  }
338  void Profile(llvm::FoldingSetNodeID &ID) {
339    Profile(ID, keyword_begin(), getNumArgs());
340  }
341};
342} // end namespace clang.
343
344unsigned Selector::getNumArgs() const {
345  unsigned IIF = getIdentifierInfoFlag();
346  if (IIF <= ZeroArg)
347    return 0;
348  if (IIF == OneArg)
349    return 1;
350  // We point to a MultiKeywordSelector.
351  MultiKeywordSelector *SI = getMultiKeywordSelector();
352  return SI->getNumArgs();
353}
354
355IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const {
356  if (getIdentifierInfoFlag() < MultiArg) {
357    assert(argIndex == 0 && "illegal keyword index");
358    return getAsIdentifierInfo();
359  }
360  // We point to a MultiKeywordSelector.
361  MultiKeywordSelector *SI = getMultiKeywordSelector();
362  return SI->getIdentifierInfoForSlot(argIndex);
363}
364
365StringRef Selector::getNameForSlot(unsigned int argIndex) const {
366  IdentifierInfo *II = getIdentifierInfoForSlot(argIndex);
367  return II? II->getName() : StringRef();
368}
369
370std::string MultiKeywordSelector::getName() const {
371  SmallString<256> Str;
372  llvm::raw_svector_ostream OS(Str);
373  for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) {
374    if (*I)
375      OS << (*I)->getName();
376    OS << ':';
377  }
378
379  return OS.str();
380}
381
382std::string Selector::getAsString() const {
383  if (InfoPtr == 0)
384    return "<null selector>";
385
386  if (getIdentifierInfoFlag() < MultiArg) {
387    IdentifierInfo *II = getAsIdentifierInfo();
388
389    // If the number of arguments is 0 then II is guaranteed to not be null.
390    if (getNumArgs() == 0)
391      return II->getName();
392
393    if (!II)
394      return ":";
395
396    return II->getName().str() + ":";
397  }
398
399  // We have a multiple keyword selector.
400  return getMultiKeywordSelector()->getName();
401}
402
403/// Interpreting the given string using the normal CamelCase
404/// conventions, determine whether the given string starts with the
405/// given "word", which is assumed to end in a lowercase letter.
406static bool startsWithWord(StringRef name, StringRef word) {
407  if (name.size() < word.size()) return false;
408  return ((name.size() == word.size() ||
409           !islower(name[word.size()]))
410          && name.startswith(word));
411}
412
413ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) {
414  IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
415  if (!first) return OMF_None;
416
417  StringRef name = first->getName();
418  if (sel.isUnarySelector()) {
419    if (name == "autorelease") return OMF_autorelease;
420    if (name == "dealloc") return OMF_dealloc;
421    if (name == "finalize") return OMF_finalize;
422    if (name == "release") return OMF_release;
423    if (name == "retain") return OMF_retain;
424    if (name == "retainCount") return OMF_retainCount;
425    if (name == "self") return OMF_self;
426  }
427
428  if (name == "performSelector") return OMF_performSelector;
429
430  // The other method families may begin with a prefix of underscores.
431  while (!name.empty() && name.front() == '_')
432    name = name.substr(1);
433
434  if (name.empty()) return OMF_None;
435  switch (name.front()) {
436  case 'a':
437    if (startsWithWord(name, "alloc")) return OMF_alloc;
438    break;
439  case 'c':
440    if (startsWithWord(name, "copy")) return OMF_copy;
441    break;
442  case 'i':
443    if (startsWithWord(name, "init")) return OMF_init;
444    break;
445  case 'm':
446    if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy;
447    break;
448  case 'n':
449    if (startsWithWord(name, "new")) return OMF_new;
450    break;
451  default:
452    break;
453  }
454
455  return OMF_None;
456}
457
458namespace {
459  struct SelectorTableImpl {
460    llvm::FoldingSet<MultiKeywordSelector> Table;
461    llvm::BumpPtrAllocator Allocator;
462  };
463} // end anonymous namespace.
464
465static SelectorTableImpl &getSelectorTableImpl(void *P) {
466  return *static_cast<SelectorTableImpl*>(P);
467}
468
469/*static*/ Selector
470SelectorTable::constructSetterName(IdentifierTable &Idents,
471                                   SelectorTable &SelTable,
472                                   const IdentifierInfo *Name) {
473  SmallString<100> SelectorName;
474  SelectorName = "set";
475  SelectorName += Name->getName();
476  SelectorName[3] = toupper(SelectorName[3]);
477  IdentifierInfo *SetterName = &Idents.get(SelectorName);
478  return SelTable.getUnarySelector(SetterName);
479}
480
481size_t SelectorTable::getTotalMemory() const {
482  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
483  return SelTabImpl.Allocator.getTotalMemory();
484}
485
486Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) {
487  if (nKeys < 2)
488    return Selector(IIV[0], nKeys);
489
490  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
491
492  // Unique selector, to guarantee there is one per name.
493  llvm::FoldingSetNodeID ID;
494  MultiKeywordSelector::Profile(ID, IIV, nKeys);
495
496  void *InsertPos = 0;
497  if (MultiKeywordSelector *SI =
498        SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos))
499    return Selector(SI);
500
501  // MultiKeywordSelector objects are not allocated with new because they have a
502  // variable size array (for parameter types) at the end of them.
503  unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *);
504  MultiKeywordSelector *SI =
505    (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size,
506                                         llvm::alignOf<MultiKeywordSelector>());
507  new (SI) MultiKeywordSelector(nKeys, IIV);
508  SelTabImpl.Table.InsertNode(SI, InsertPos);
509  return Selector(SI);
510}
511
512SelectorTable::SelectorTable() {
513  Impl = new SelectorTableImpl();
514}
515
516SelectorTable::~SelectorTable() {
517  delete &getSelectorTableImpl(Impl);
518}
519
520const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) {
521  switch (Operator) {
522  case OO_None:
523  case NUM_OVERLOADED_OPERATORS:
524    return 0;
525
526#define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \
527  case OO_##Name: return Spelling;
528#include "clang/Basic/OperatorKinds.def"
529  }
530
531  llvm_unreachable("Invalid OverloadedOperatorKind!");
532}
533