IdentifierTable.cpp revision eee242ff426bf79149f221798966e58688383c1e 
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
24using namespace clang;
25
26//===----------------------------------------------------------------------===//
27// IdentifierInfo Implementation
28//===----------------------------------------------------------------------===//
29
30IdentifierInfo::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  FETokenInfo = 0;
44  Entry = 0;
45}
46
47//===----------------------------------------------------------------------===//
48// IdentifierTable Implementation
49//===----------------------------------------------------------------------===//
50
51IdentifierIterator::~IdentifierIterator() { }
52
53IdentifierInfoLookup::~IdentifierInfoLookup() {}
54
55namespace {
56  /// \brief A simple identifier lookup iterator that represents an
57  /// empty sequence of identifiers.
58  class EmptyLookupIterator : public IdentifierIterator
59  {
60  public:
61    virtual StringRef Next() { return StringRef(); }
62  };
63}
64
65IdentifierIterator *IdentifierInfoLookup::getIdentifiers() const {
66  return new EmptyLookupIterator();
67}
68
69ExternalIdentifierLookup::~ExternalIdentifierLookup() {}
70
71IdentifierTable::IdentifierTable(const LangOptions &LangOpts,
72                                 IdentifierInfoLookup* externalLookup)
73  : HashTable(8192), // Start with space for 8K identifiers.
74    ExternalLookup(externalLookup) {
75
76  // Populate the identifier table with info about keywords for the current
77  // language.
78  AddKeywords(LangOpts);
79}
80
81//===----------------------------------------------------------------------===//
82// Language Keyword Implementation
83//===----------------------------------------------------------------------===//
84
85// Constants for TokenKinds.def
86namespace {
87  enum {
88    KEYC99 = 0x1,
89    KEYCXX = 0x2,
90    KEYCXX0X = 0x4,
91    KEYGNU = 0x8,
92    KEYMS = 0x10,
93    BOOLSUPPORT = 0x20,
94    KEYALTIVEC = 0x40,
95    KEYNOCXX = 0x80,
96    KEYBORLAND = 0x100,
97    KEYOPENCL = 0x200,
98    KEYC1X = 0x400,
99    KEYARC = 0x800,
100    KEYALL = 0x0fff
101  };
102}
103
104/// AddKeyword - This method is used to associate a token ID with specific
105/// identifiers because they are language keywords.  This causes the lexer to
106/// automatically map matching identifiers to specialized token codes.
107///
108/// The C90/C99/CPP/CPP0x flags are set to 3 if the token is a keyword in a
109/// future language standard, set to 2 if the token should be enabled in the
110/// specified langauge, set to 1 if it is an extension in the specified
111/// language, and set to 0 if disabled in the specified language.
112static void AddKeyword(StringRef Keyword,
113                       tok::TokenKind TokenCode, unsigned Flags,
114                       const LangOptions &LangOpts, IdentifierTable &Table) {
115  unsigned AddResult = 0;
116  if (Flags == KEYALL) AddResult = 2;
117  else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2;
118  else if (LangOpts.CPlusPlus0x && (Flags & KEYCXX0X)) AddResult = 2;
119  else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2;
120  else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1;
121  else if (LangOpts.MicrosoftExt && (Flags & KEYMS)) AddResult = 1;
122  else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1;
123  else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2;
124  else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2;
125  else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2;
126  else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2;
127  else if (LangOpts.C1X && (Flags & KEYC1X)) AddResult = 2;
128  else if (LangOpts.ObjCAutoRefCount && (Flags & KEYARC)) AddResult = 2;
129  else if (LangOpts.CPlusPlus && (Flags & KEYCXX0X)) AddResult = 3;
130
131  // Don't add this keyword if disabled in this language.
132  if (AddResult == 0) return;
133
134  IdentifierInfo &Info =
135      Table.get(Keyword, AddResult == 3 ? tok::identifier : TokenCode);
136  Info.setIsExtensionToken(AddResult == 1);
137  Info.setIsCXX11CompatKeyword(AddResult == 3);
138}
139
140/// AddCXXOperatorKeyword - Register a C++ operator keyword alternative
141/// representations.
142static void AddCXXOperatorKeyword(StringRef Keyword,
143                                  tok::TokenKind TokenCode,
144                                  IdentifierTable &Table) {
145  IdentifierInfo &Info = Table.get(Keyword, TokenCode);
146  Info.setIsCPlusPlusOperatorKeyword();
147}
148
149/// AddObjCKeyword - Register an Objective-C @keyword like "class" "selector" or
150/// "property".
151static void AddObjCKeyword(StringRef Name,
152                           tok::ObjCKeywordKind ObjCID,
153                           IdentifierTable &Table) {
154  Table.get(Name).setObjCKeywordID(ObjCID);
155}
156
157/// AddKeywords - Add all keywords to the symbol table.
158///
159void IdentifierTable::AddKeywords(const LangOptions &LangOpts) {
160  // Add keywords and tokens for the current language.
161#define KEYWORD(NAME, FLAGS) \
162  AddKeyword(StringRef(#NAME), tok::kw_ ## NAME,  \
163             FLAGS, LangOpts, *this);
164#define ALIAS(NAME, TOK, FLAGS) \
165  AddKeyword(StringRef(NAME), tok::kw_ ## TOK,  \
166             FLAGS, LangOpts, *this);
167#define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \
168  if (LangOpts.CXXOperatorNames)          \
169    AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this);
170#define OBJC1_AT_KEYWORD(NAME) \
171  if (LangOpts.ObjC1)          \
172    AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
173#define OBJC2_AT_KEYWORD(NAME) \
174  if (LangOpts.ObjC2)          \
175    AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this);
176#define TESTING_KEYWORD(NAME, FLAGS)
177#include "clang/Basic/TokenKinds.def"
178
179  if (LangOpts.ParseUnknownAnytype)
180    AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL,
181               LangOpts, *this);
182}
183
184tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const {
185  // We use a perfect hash function here involving the length of the keyword,
186  // the first and third character.  For preprocessor ID's there are no
187  // collisions (if there were, the switch below would complain about duplicate
188  // case values).  Note that this depends on 'if' being null terminated.
189
190#define HASH(LEN, FIRST, THIRD) \
191  (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31)
192#define CASE(LEN, FIRST, THIRD, NAME) \
193  case HASH(LEN, FIRST, THIRD): \
194    return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME
195
196  unsigned Len = getLength();
197  if (Len < 2) return tok::pp_not_keyword;
198  const char *Name = getNameStart();
199  switch (HASH(Len, Name[0], Name[2])) {
200  default: return tok::pp_not_keyword;
201  CASE( 2, 'i', '\0', if);
202  CASE( 4, 'e', 'i', elif);
203  CASE( 4, 'e', 's', else);
204  CASE( 4, 'l', 'n', line);
205  CASE( 4, 's', 'c', sccs);
206  CASE( 5, 'e', 'd', endif);
207  CASE( 5, 'e', 'r', error);
208  CASE( 5, 'i', 'e', ident);
209  CASE( 5, 'i', 'd', ifdef);
210  CASE( 5, 'u', 'd', undef);
211
212  CASE( 6, 'a', 's', assert);
213  CASE( 6, 'd', 'f', define);
214  CASE( 6, 'i', 'n', ifndef);
215  CASE( 6, 'i', 'p', import);
216  CASE( 6, 'p', 'a', pragma);
217
218  CASE( 7, 'd', 'f', defined);
219  CASE( 7, 'i', 'c', include);
220  CASE( 7, 'w', 'r', warning);
221
222  CASE( 8, 'u', 'a', unassert);
223  CASE(12, 'i', 'c', include_next);
224
225  CASE(16, '_', 'i', __include_macros);
226  CASE(16, '_', 'e', __export_macro__);
227
228  CASE(17, '_', 'p', __private_macro__);
229#undef CASE
230#undef HASH
231  }
232}
233
234//===----------------------------------------------------------------------===//
235// Stats Implementation
236//===----------------------------------------------------------------------===//
237
238/// PrintStats - Print statistics about how well the identifier table is doing
239/// at hashing identifiers.
240void IdentifierTable::PrintStats() const {
241  unsigned NumBuckets = HashTable.getNumBuckets();
242  unsigned NumIdentifiers = HashTable.getNumItems();
243  unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers;
244  unsigned AverageIdentifierSize = 0;
245  unsigned MaxIdentifierLength = 0;
246
247  // TODO: Figure out maximum times an identifier had to probe for -stats.
248  for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator
249       I = HashTable.begin(), E = HashTable.end(); I != E; ++I) {
250    unsigned IdLen = I->getKeyLength();
251    AverageIdentifierSize += IdLen;
252    if (MaxIdentifierLength < IdLen)
253      MaxIdentifierLength = IdLen;
254  }
255
256  fprintf(stderr, "\n*** Identifier Table Stats:\n");
257  fprintf(stderr, "# Identifiers:   %d\n", NumIdentifiers);
258  fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets);
259  fprintf(stderr, "Hash density (#identifiers per bucket): %f\n",
260          NumIdentifiers/(double)NumBuckets);
261  fprintf(stderr, "Ave identifier length: %f\n",
262          (AverageIdentifierSize/(double)NumIdentifiers));
263  fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength);
264
265  // Compute statistics about the memory allocated for identifiers.
266  HashTable.getAllocator().PrintStats();
267}
268
269//===----------------------------------------------------------------------===//
270// SelectorTable Implementation
271//===----------------------------------------------------------------------===//
272
273unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) {
274  return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr());
275}
276
277namespace clang {
278/// MultiKeywordSelector - One of these variable length records is kept for each
279/// selector containing more than one keyword. We use a folding set
280/// to unique aggregate names (keyword selectors in ObjC parlance). Access to
281/// this class is provided strictly through Selector.
282class MultiKeywordSelector
283  : public DeclarationNameExtra, public llvm::FoldingSetNode {
284  MultiKeywordSelector(unsigned nKeys) {
285    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
286  }
287public:
288  // Constructor for keyword selectors.
289  MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) {
290    assert((nKeys > 1) && "not a multi-keyword selector");
291    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
292
293    // Fill in the trailing keyword array.
294    IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1);
295    for (unsigned i = 0; i != nKeys; ++i)
296      KeyInfo[i] = IIV[i];
297  }
298
299  // getName - Derive the full selector name and return it.
300  std::string getName() const;
301
302  unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; }
303
304  typedef IdentifierInfo *const *keyword_iterator;
305  keyword_iterator keyword_begin() const {
306    return reinterpret_cast<keyword_iterator>(this+1);
307  }
308  keyword_iterator keyword_end() const {
309    return keyword_begin()+getNumArgs();
310  }
311  IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const {
312    assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index");
313    return keyword_begin()[i];
314  }
315  static void Profile(llvm::FoldingSetNodeID &ID,
316                      keyword_iterator ArgTys, unsigned NumArgs) {
317    ID.AddInteger(NumArgs);
318    for (unsigned i = 0; i != NumArgs; ++i)
319      ID.AddPointer(ArgTys[i]);
320  }
321  void Profile(llvm::FoldingSetNodeID &ID) {
322    Profile(ID, keyword_begin(), getNumArgs());
323  }
324};
325} // end namespace clang.
326
327unsigned Selector::getNumArgs() const {
328  unsigned IIF = getIdentifierInfoFlag();
329  if (IIF == ZeroArg)
330    return 0;
331  if (IIF == OneArg)
332    return 1;
333  // We point to a MultiKeywordSelector (pointer doesn't contain any flags).
334  MultiKeywordSelector *SI = reinterpret_cast<MultiKeywordSelector *>(InfoPtr);
335  return SI->getNumArgs();
336}
337
338IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const {
339  if (getIdentifierInfoFlag()) {
340    assert(argIndex == 0 && "illegal keyword index");
341    return getAsIdentifierInfo();
342  }
343  // We point to a MultiKeywordSelector (pointer doesn't contain any flags).
344  MultiKeywordSelector *SI = reinterpret_cast<MultiKeywordSelector *>(InfoPtr);
345  return SI->getIdentifierInfoForSlot(argIndex);
346}
347
348StringRef Selector::getNameForSlot(unsigned int argIndex) const {
349  IdentifierInfo *II = getIdentifierInfoForSlot(argIndex);
350  return II? II->getName() : StringRef();
351}
352
353std::string MultiKeywordSelector::getName() const {
354  llvm::SmallString<256> Str;
355  llvm::raw_svector_ostream OS(Str);
356  for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) {
357    if (*I)
358      OS << (*I)->getName();
359    OS << ':';
360  }
361
362  return OS.str();
363}
364
365std::string Selector::getAsString() const {
366  if (InfoPtr == 0)
367    return "<null selector>";
368
369  if (InfoPtr & ArgFlags) {
370    IdentifierInfo *II = getAsIdentifierInfo();
371
372    // If the number of arguments is 0 then II is guaranteed to not be null.
373    if (getNumArgs() == 0)
374      return II->getName();
375
376    if (!II)
377      return ":";
378
379    return II->getName().str() + ":";
380  }
381
382  // We have a multiple keyword selector (no embedded flags).
383  return reinterpret_cast<MultiKeywordSelector *>(InfoPtr)->getName();
384}
385
386/// Interpreting the given string using the normal CamelCase
387/// conventions, determine whether the given string starts with the
388/// given "word", which is assumed to end in a lowercase letter.
389static bool startsWithWord(StringRef name, StringRef word) {
390  if (name.size() < word.size()) return false;
391  return ((name.size() == word.size() ||
392           !islower(name[word.size()]))
393          && name.startswith(word));
394}
395
396ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) {
397  IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
398  if (!first) return OMF_None;
399
400  StringRef name = first->getName();
401  if (sel.isUnarySelector()) {
402    if (name == "autorelease") return OMF_autorelease;
403    if (name == "dealloc") return OMF_dealloc;
404    if (name == "finalize") return OMF_finalize;
405    if (name == "release") return OMF_release;
406    if (name == "retain") return OMF_retain;
407    if (name == "retainCount") return OMF_retainCount;
408    if (name == "self") return OMF_self;
409  }
410
411  if (name == "performSelector") return OMF_performSelector;
412
413  // The other method families may begin with a prefix of underscores.
414  while (!name.empty() && name.front() == '_')
415    name = name.substr(1);
416
417  if (name.empty()) return OMF_None;
418  switch (name.front()) {
419  case 'a':
420    if (startsWithWord(name, "alloc")) return OMF_alloc;
421    break;
422  case 'c':
423    if (startsWithWord(name, "copy")) return OMF_copy;
424    break;
425  case 'i':
426    if (startsWithWord(name, "init")) return OMF_init;
427    break;
428  case 'm':
429    if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy;
430    break;
431  case 'n':
432    if (startsWithWord(name, "new")) return OMF_new;
433    break;
434  default:
435    break;
436  }
437
438  return OMF_None;
439}
440
441namespace {
442  struct SelectorTableImpl {
443    llvm::FoldingSet<MultiKeywordSelector> Table;
444    llvm::BumpPtrAllocator Allocator;
445  };
446} // end anonymous namespace.
447
448static SelectorTableImpl &getSelectorTableImpl(void *P) {
449  return *static_cast<SelectorTableImpl*>(P);
450}
451
452size_t SelectorTable::getTotalMemory() const {
453  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
454  return SelTabImpl.Allocator.getTotalMemory();
455}
456
457Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) {
458  if (nKeys < 2)
459    return Selector(IIV[0], nKeys);
460
461  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
462
463  // Unique selector, to guarantee there is one per name.
464  llvm::FoldingSetNodeID ID;
465  MultiKeywordSelector::Profile(ID, IIV, nKeys);
466
467  void *InsertPos = 0;
468  if (MultiKeywordSelector *SI =
469        SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos))
470    return Selector(SI);
471
472  // MultiKeywordSelector objects are not allocated with new because they have a
473  // variable size array (for parameter types) at the end of them.
474  unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *);
475  MultiKeywordSelector *SI =
476    (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size,
477                                         llvm::alignOf<MultiKeywordSelector>());
478  new (SI) MultiKeywordSelector(nKeys, IIV);
479  SelTabImpl.Table.InsertNode(SI, InsertPos);
480  return Selector(SI);
481}
482
483SelectorTable::SelectorTable() {
484  Impl = new SelectorTableImpl();
485}
486
487SelectorTable::~SelectorTable() {
488  delete &getSelectorTableImpl(Impl);
489}
490
491const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) {
492  switch (Operator) {
493  case OO_None:
494  case NUM_OVERLOADED_OPERATORS:
495    return 0;
496
497#define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \
498  case OO_##Name: return Spelling;
499#include "clang/Basic/OperatorKinds.def"
500  }
501
502  return 0;
503}
504