IdentifierTable.cpp revision 926df6cfabf3eaa4afc990c097fa4619b76a9b57 
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/Support/raw_ostream.h"
21#include <cstdio>
22
23using namespace clang;
24
25//===----------------------------------------------------------------------===//
26// IdentifierInfo Implementation
27//===----------------------------------------------------------------------===//
28
29IdentifierInfo::IdentifierInfo() {
30  TokenID = tok::identifier;
31  ObjCOrBuiltinID = 0;
32  HasMacro = false;
33  IsExtension = false;
34  IsPoisoned = false;
35  IsCPPOperatorKeyword = false;
36  NeedsHandleIdentifier = false;
37  IsFromAST = false;
38  RevertedTokenID = false;
39  FETokenInfo = 0;
40  Entry = 0;
41}
42
43//===----------------------------------------------------------------------===//
44// IdentifierTable Implementation
45//===----------------------------------------------------------------------===//
46
47IdentifierIterator::~IdentifierIterator() { }
48
49IdentifierInfoLookup::~IdentifierInfoLookup() {}
50
51namespace {
52  /// \brief A simple identifier lookup iterator that represents an
53  /// empty sequence of identifiers.
54  class EmptyLookupIterator : public IdentifierIterator
55  {
56  public:
57    virtual llvm::StringRef Next() { return llvm::StringRef(); }
58  };
59}
60
61IdentifierIterator *IdentifierInfoLookup::getIdentifiers() const {
62  return new EmptyLookupIterator();
63}
64
65ExternalIdentifierLookup::~ExternalIdentifierLookup() {}
66
67IdentifierTable::IdentifierTable(const LangOptions &LangOpts,
68                                 IdentifierInfoLookup* externalLookup)
69  : HashTable(8192), // Start with space for 8K identifiers.
70    ExternalLookup(externalLookup) {
71
72  // Populate the identifier table with info about keywords for the current
73  // language.
74  AddKeywords(LangOpts);
75}
76
77//===----------------------------------------------------------------------===//
78// Language Keyword Implementation
79//===----------------------------------------------------------------------===//
80
81// Constants for TokenKinds.def
82namespace {
83  enum {
84    KEYC99 = 0x1,
85    KEYCXX = 0x2,
86    KEYCXX0X = 0x4,
87    KEYGNU = 0x8,
88    KEYMS = 0x10,
89    BOOLSUPPORT = 0x20,
90    KEYALTIVEC = 0x40,
91    KEYNOCXX = 0x80,
92    KEYBORLAND = 0x100,
93    KEYOPENCL = 0x200,
94    KEYC1X = 0x400,
95    KEYALL = 0x7ff
96  };
97}
98
99/// AddKeyword - This method is used to associate a token ID with specific
100/// identifiers because they are language keywords.  This causes the lexer to
101/// automatically map matching identifiers to specialized token codes.
102///
103/// The C90/C99/CPP/CPP0x flags are set to 2 if the token should be
104/// enabled in the specified langauge, set to 1 if it is an extension
105/// in the specified language, and set to 0 if disabled in the
106/// specified language.
107static void AddKeyword(llvm::StringRef Keyword,
108                       tok::TokenKind TokenCode, unsigned Flags,
109                       const LangOptions &LangOpts, IdentifierTable &Table) {
110  unsigned AddResult = 0;
111  if (Flags == KEYALL) AddResult = 2;
112  else if (LangOpts.CPlusPlus && (Flags & KEYCXX)) AddResult = 2;
113  else if (LangOpts.CPlusPlus0x && (Flags & KEYCXX0X)) AddResult = 2;
114  else if (LangOpts.C99 && (Flags & KEYC99)) AddResult = 2;
115  else if (LangOpts.GNUKeywords && (Flags & KEYGNU)) AddResult = 1;
116  else if (LangOpts.Microsoft && (Flags & KEYMS)) AddResult = 1;
117  else if (LangOpts.Borland && (Flags & KEYBORLAND)) AddResult = 1;
118  else if (LangOpts.Bool && (Flags & BOOLSUPPORT)) AddResult = 2;
119  else if (LangOpts.AltiVec && (Flags & KEYALTIVEC)) AddResult = 2;
120  else if (LangOpts.OpenCL && (Flags & KEYOPENCL)) AddResult = 2;
121  else if (!LangOpts.CPlusPlus && (Flags & KEYNOCXX)) AddResult = 2;
122  else if (LangOpts.C1X && (Flags & KEYC1X)) AddResult = 2;
123
124  // Don't add this keyword if disabled in this language.
125  if (AddResult == 0) return;
126
127  IdentifierInfo &Info = Table.get(Keyword, TokenCode);
128  Info.setIsExtensionToken(AddResult == 1);
129}
130
131/// AddCXXOperatorKeyword - Register a C++ operator keyword alternative
132/// representations.
133static void AddCXXOperatorKeyword(llvm::StringRef Keyword,
134                                  tok::TokenKind TokenCode,
135                                  IdentifierTable &Table) {
136  IdentifierInfo &Info = Table.get(Keyword, TokenCode);
137  Info.setIsCPlusPlusOperatorKeyword();
138}
139
140/// AddObjCKeyword - Register an Objective-C @keyword like "class" "selector" or
141/// "property".
142static void AddObjCKeyword(llvm::StringRef Name,
143                           tok::ObjCKeywordKind ObjCID,
144                           IdentifierTable &Table) {
145  Table.get(Name).setObjCKeywordID(ObjCID);
146}
147
148/// AddKeywords - Add all keywords to the symbol table.
149///
150void IdentifierTable::AddKeywords(const LangOptions &LangOpts) {
151  // Add keywords and tokens for the current language.
152#define KEYWORD(NAME, FLAGS) \
153  AddKeyword(llvm::StringRef(#NAME), tok::kw_ ## NAME,  \
154             FLAGS, LangOpts, *this);
155#define ALIAS(NAME, TOK, FLAGS) \
156  AddKeyword(llvm::StringRef(NAME), tok::kw_ ## TOK,  \
157             FLAGS, LangOpts, *this);
158#define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \
159  if (LangOpts.CXXOperatorNames)          \
160    AddCXXOperatorKeyword(llvm::StringRef(#NAME), tok::ALIAS, *this);
161#define OBJC1_AT_KEYWORD(NAME) \
162  if (LangOpts.ObjC1)          \
163    AddObjCKeyword(llvm::StringRef(#NAME), tok::objc_##NAME, *this);
164#define OBJC2_AT_KEYWORD(NAME) \
165  if (LangOpts.ObjC2)          \
166    AddObjCKeyword(llvm::StringRef(#NAME), tok::objc_##NAME, *this);
167#define TESTING_KEYWORD(NAME, FLAGS)
168#include "clang/Basic/TokenKinds.def"
169
170  if (LangOpts.ParseUnknownAnytype)
171    AddKeyword("__unknown_anytype", tok::kw___unknown_anytype, KEYALL,
172               LangOpts, *this);
173}
174
175tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const {
176  // We use a perfect hash function here involving the length of the keyword,
177  // the first and third character.  For preprocessor ID's there are no
178  // collisions (if there were, the switch below would complain about duplicate
179  // case values).  Note that this depends on 'if' being null terminated.
180
181#define HASH(LEN, FIRST, THIRD) \
182  (LEN << 5) + (((FIRST-'a') + (THIRD-'a')) & 31)
183#define CASE(LEN, FIRST, THIRD, NAME) \
184  case HASH(LEN, FIRST, THIRD): \
185    return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME
186
187  unsigned Len = getLength();
188  if (Len < 2) return tok::pp_not_keyword;
189  const char *Name = getNameStart();
190  switch (HASH(Len, Name[0], Name[2])) {
191  default: return tok::pp_not_keyword;
192  CASE( 2, 'i', '\0', if);
193  CASE( 4, 'e', 'i', elif);
194  CASE( 4, 'e', 's', else);
195  CASE( 4, 'l', 'n', line);
196  CASE( 4, 's', 'c', sccs);
197  CASE( 5, 'e', 'd', endif);
198  CASE( 5, 'e', 'r', error);
199  CASE( 5, 'i', 'e', ident);
200  CASE( 5, 'i', 'd', ifdef);
201  CASE( 5, 'u', 'd', undef);
202
203  CASE( 6, 'a', 's', assert);
204  CASE( 6, 'd', 'f', define);
205  CASE( 6, 'i', 'n', ifndef);
206  CASE( 6, 'i', 'p', import);
207  CASE( 6, 'p', 'a', pragma);
208
209  CASE( 7, 'd', 'f', defined);
210  CASE( 7, 'i', 'c', include);
211  CASE( 7, 'w', 'r', warning);
212
213  CASE( 8, 'u', 'a', unassert);
214  CASE(12, 'i', 'c', include_next);
215
216  CASE(16, '_', 'i', __include_macros);
217#undef CASE
218#undef HASH
219  }
220}
221
222//===----------------------------------------------------------------------===//
223// Stats Implementation
224//===----------------------------------------------------------------------===//
225
226/// PrintStats - Print statistics about how well the identifier table is doing
227/// at hashing identifiers.
228void IdentifierTable::PrintStats() const {
229  unsigned NumBuckets = HashTable.getNumBuckets();
230  unsigned NumIdentifiers = HashTable.getNumItems();
231  unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers;
232  unsigned AverageIdentifierSize = 0;
233  unsigned MaxIdentifierLength = 0;
234
235  // TODO: Figure out maximum times an identifier had to probe for -stats.
236  for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator
237       I = HashTable.begin(), E = HashTable.end(); I != E; ++I) {
238    unsigned IdLen = I->getKeyLength();
239    AverageIdentifierSize += IdLen;
240    if (MaxIdentifierLength < IdLen)
241      MaxIdentifierLength = IdLen;
242  }
243
244  fprintf(stderr, "\n*** Identifier Table Stats:\n");
245  fprintf(stderr, "# Identifiers:   %d\n", NumIdentifiers);
246  fprintf(stderr, "# Empty Buckets: %d\n", NumEmptyBuckets);
247  fprintf(stderr, "Hash density (#identifiers per bucket): %f\n",
248          NumIdentifiers/(double)NumBuckets);
249  fprintf(stderr, "Ave identifier length: %f\n",
250          (AverageIdentifierSize/(double)NumIdentifiers));
251  fprintf(stderr, "Max identifier length: %d\n", MaxIdentifierLength);
252
253  // Compute statistics about the memory allocated for identifiers.
254  HashTable.getAllocator().PrintStats();
255}
256
257//===----------------------------------------------------------------------===//
258// SelectorTable Implementation
259//===----------------------------------------------------------------------===//
260
261unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) {
262  return DenseMapInfo<void*>::getHashValue(S.getAsOpaquePtr());
263}
264
265namespace clang {
266/// MultiKeywordSelector - One of these variable length records is kept for each
267/// selector containing more than one keyword. We use a folding set
268/// to unique aggregate names (keyword selectors in ObjC parlance). Access to
269/// this class is provided strictly through Selector.
270class MultiKeywordSelector
271  : public DeclarationNameExtra, public llvm::FoldingSetNode {
272  MultiKeywordSelector(unsigned nKeys) {
273    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
274  }
275public:
276  // Constructor for keyword selectors.
277  MultiKeywordSelector(unsigned nKeys, IdentifierInfo **IIV) {
278    assert((nKeys > 1) && "not a multi-keyword selector");
279    ExtraKindOrNumArgs = NUM_EXTRA_KINDS + nKeys;
280
281    // Fill in the trailing keyword array.
282    IdentifierInfo **KeyInfo = reinterpret_cast<IdentifierInfo **>(this+1);
283    for (unsigned i = 0; i != nKeys; ++i)
284      KeyInfo[i] = IIV[i];
285  }
286
287  // getName - Derive the full selector name and return it.
288  std::string getName() const;
289
290  unsigned getNumArgs() const { return ExtraKindOrNumArgs - NUM_EXTRA_KINDS; }
291
292  typedef IdentifierInfo *const *keyword_iterator;
293  keyword_iterator keyword_begin() const {
294    return reinterpret_cast<keyword_iterator>(this+1);
295  }
296  keyword_iterator keyword_end() const {
297    return keyword_begin()+getNumArgs();
298  }
299  IdentifierInfo *getIdentifierInfoForSlot(unsigned i) const {
300    assert(i < getNumArgs() && "getIdentifierInfoForSlot(): illegal index");
301    return keyword_begin()[i];
302  }
303  static void Profile(llvm::FoldingSetNodeID &ID,
304                      keyword_iterator ArgTys, unsigned NumArgs) {
305    ID.AddInteger(NumArgs);
306    for (unsigned i = 0; i != NumArgs; ++i)
307      ID.AddPointer(ArgTys[i]);
308  }
309  void Profile(llvm::FoldingSetNodeID &ID) {
310    Profile(ID, keyword_begin(), getNumArgs());
311  }
312};
313} // end namespace clang.
314
315unsigned Selector::getNumArgs() const {
316  unsigned IIF = getIdentifierInfoFlag();
317  if (IIF == ZeroArg)
318    return 0;
319  if (IIF == OneArg)
320    return 1;
321  // We point to a MultiKeywordSelector (pointer doesn't contain any flags).
322  MultiKeywordSelector *SI = reinterpret_cast<MultiKeywordSelector *>(InfoPtr);
323  return SI->getNumArgs();
324}
325
326IdentifierInfo *Selector::getIdentifierInfoForSlot(unsigned argIndex) const {
327  if (getIdentifierInfoFlag()) {
328    assert(argIndex == 0 && "illegal keyword index");
329    return getAsIdentifierInfo();
330  }
331  // We point to a MultiKeywordSelector (pointer doesn't contain any flags).
332  MultiKeywordSelector *SI = reinterpret_cast<MultiKeywordSelector *>(InfoPtr);
333  return SI->getIdentifierInfoForSlot(argIndex);
334}
335
336llvm::StringRef Selector::getNameForSlot(unsigned int argIndex) const {
337  IdentifierInfo *II = getIdentifierInfoForSlot(argIndex);
338  return II? II->getName() : llvm::StringRef();
339}
340
341std::string MultiKeywordSelector::getName() const {
342  llvm::SmallString<256> Str;
343  llvm::raw_svector_ostream OS(Str);
344  for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) {
345    if (*I)
346      OS << (*I)->getName();
347    OS << ':';
348  }
349
350  return OS.str();
351}
352
353std::string Selector::getAsString() const {
354  if (InfoPtr == 0)
355    return "<null selector>";
356
357  if (InfoPtr & ArgFlags) {
358    IdentifierInfo *II = getAsIdentifierInfo();
359
360    // If the number of arguments is 0 then II is guaranteed to not be null.
361    if (getNumArgs() == 0)
362      return II->getName();
363
364    if (!II)
365      return ":";
366
367    return II->getName().str() + ":";
368  }
369
370  // We have a multiple keyword selector (no embedded flags).
371  return reinterpret_cast<MultiKeywordSelector *>(InfoPtr)->getName();
372}
373
374/// Interpreting the given string using the normal CamelCase
375/// conventions, determine whether the given string starts with the
376/// given "word", which is assumed to end in a lowercase letter.
377static bool startsWithWord(llvm::StringRef name, llvm::StringRef word) {
378  if (name.size() < word.size()) return false;
379  return ((name.size() == word.size() ||
380           !islower(name[word.size()]))
381          && name.startswith(word));
382}
383
384ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) {
385  IdentifierInfo *first = sel.getIdentifierInfoForSlot(0);
386  if (!first) return OMF_None;
387
388  llvm::StringRef name = first->getName();
389  if (sel.isUnarySelector()) {
390    if (name == "autorelease") return OMF_autorelease;
391    if (name == "dealloc") return OMF_dealloc;
392    if (name == "release") return OMF_release;
393    if (name == "retain") return OMF_retain;
394    if (name == "retainCount") return OMF_retainCount;
395    if (name == "self") return OMF_self;
396  }
397
398  // The other method families may begin with a prefix of underscores.
399  while (!name.empty() && name.front() == '_')
400    name = name.substr(1);
401
402  if (name.empty()) return OMF_None;
403  switch (name.front()) {
404  case 'a':
405    if (startsWithWord(name, "alloc")) return OMF_alloc;
406    break;
407  case 'c':
408    if (startsWithWord(name, "copy")) return OMF_copy;
409    break;
410  case 'i':
411    if (startsWithWord(name, "init")) return OMF_init;
412    break;
413  case 'm':
414    if (startsWithWord(name, "mutableCopy")) return OMF_mutableCopy;
415    break;
416  case 'n':
417    if (startsWithWord(name, "new")) return OMF_new;
418    break;
419  default:
420    break;
421  }
422
423  return OMF_None;
424}
425
426namespace {
427  struct SelectorTableImpl {
428    llvm::FoldingSet<MultiKeywordSelector> Table;
429    llvm::BumpPtrAllocator Allocator;
430  };
431} // end anonymous namespace.
432
433static SelectorTableImpl &getSelectorTableImpl(void *P) {
434  return *static_cast<SelectorTableImpl*>(P);
435}
436
437size_t SelectorTable::getTotalMemory() const {
438  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
439  return SelTabImpl.Allocator.getTotalMemory();
440}
441
442Selector SelectorTable::getSelector(unsigned nKeys, IdentifierInfo **IIV) {
443  if (nKeys < 2)
444    return Selector(IIV[0], nKeys);
445
446  SelectorTableImpl &SelTabImpl = getSelectorTableImpl(Impl);
447
448  // Unique selector, to guarantee there is one per name.
449  llvm::FoldingSetNodeID ID;
450  MultiKeywordSelector::Profile(ID, IIV, nKeys);
451
452  void *InsertPos = 0;
453  if (MultiKeywordSelector *SI =
454        SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos))
455    return Selector(SI);
456
457  // MultiKeywordSelector objects are not allocated with new because they have a
458  // variable size array (for parameter types) at the end of them.
459  unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *);
460  MultiKeywordSelector *SI =
461    (MultiKeywordSelector*)SelTabImpl.Allocator.Allocate(Size,
462                                         llvm::alignOf<MultiKeywordSelector>());
463  new (SI) MultiKeywordSelector(nKeys, IIV);
464  SelTabImpl.Table.InsertNode(SI, InsertPos);
465  return Selector(SI);
466}
467
468SelectorTable::SelectorTable() {
469  Impl = new SelectorTableImpl();
470}
471
472SelectorTable::~SelectorTable() {
473  delete &getSelectorTableImpl(Impl);
474}
475
476const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) {
477  switch (Operator) {
478  case OO_None:
479  case NUM_OVERLOADED_OPERATORS:
480    return 0;
481
482#define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \
483  case OO_##Name: return Spelling;
484#include "clang/Basic/OperatorKinds.def"
485  }
486
487  return 0;
488}
489
490