CIndexUSRs.cpp revision 465d41b92b2c862f3062c412a0538db65c6a2661 
1//===- CIndexUSR.cpp - Clang-C Source Indexing Library --------------------===//
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 generation and use of USRs from CXEntities.
11//
12//===----------------------------------------------------------------------===//
13
14#include "CIndexer.h"
15#include "CXCursor.h"
16#include "clang/AST/DeclVisitor.h"
17#include "clang/Frontend/ASTUnit.h"
18#include "clang/Lex/PreprocessingRecord.h"
19#include "llvm/ADT/SmallString.h"
20#include "llvm/Support/raw_ostream.h"
21
22using namespace clang;
23using namespace clang::cxstring;
24
25//===----------------------------------------------------------------------===//
26// USR generation.
27//===----------------------------------------------------------------------===//
28
29namespace {
30class USRGenerator : public DeclVisitor<USRGenerator> {
31  llvm::SmallString<1024> Buf;
32  llvm::raw_svector_ostream Out;
33  bool IgnoreResults;
34  ASTUnit *AU;
35  bool generatedLoc;
36public:
37  USRGenerator(const CXCursor *C = 0)
38    : Out(Buf),
39      IgnoreResults(false),
40      AU(C ? cxcursor::getCursorASTUnit(*C) : 0),
41      generatedLoc(false)
42  {
43    // Add the USR space prefix.
44    Out << "c:";
45  }
46
47  llvm::StringRef str() {
48    return Out.str();
49  }
50
51  USRGenerator* operator->() { return this; }
52
53  template <typename T>
54  llvm::raw_svector_ostream &operator<<(const T &x) {
55    Out << x;
56    return Out;
57  }
58
59  bool ignoreResults() const { return IgnoreResults; }
60
61  // Visitation methods from generating USRs from AST elements.
62  void VisitDeclContext(DeclContext *D);
63  void VisitFieldDecl(FieldDecl *D);
64  void VisitFunctionDecl(FunctionDecl *D);
65  void VisitNamedDecl(NamedDecl *D);
66  void VisitNamespaceDecl(NamespaceDecl *D);
67  void VisitObjCClassDecl(ObjCClassDecl *CD);
68  void VisitObjCContainerDecl(ObjCContainerDecl *CD);
69  void VisitObjCForwardProtocolDecl(ObjCForwardProtocolDecl *P);
70  void VisitObjCMethodDecl(ObjCMethodDecl *MD);
71  void VisitObjCPropertyDecl(ObjCPropertyDecl *D);
72  void VisitObjCPropertyImplDecl(ObjCPropertyImplDecl *D);
73  void VisitTagDecl(TagDecl *D);
74  void VisitTypedefDecl(TypedefDecl *D);
75  void VisitVarDecl(VarDecl *D);
76  void VisitLinkageSpecDecl(LinkageSpecDecl *D) {
77    IgnoreResults = true;
78    return;
79  }
80
81  /// Generate the string component containing the location of the
82  ///  declaration.
83  bool GenLoc(const Decl *D);
84
85  /// String generation methods used both by the visitation methods
86  /// and from other clients that want to directly generate USRs.  These
87  /// methods do not construct complete USRs (which incorporate the parents
88  /// of an AST element), but only the fragments concerning the AST element
89  /// itself.
90
91  /// Generate a USR for an Objective-C class.
92  void GenObjCClass(llvm::StringRef cls);
93  /// Generate a USR for an Objective-C class category.
94  void GenObjCCategory(llvm::StringRef cls, llvm::StringRef cat);
95  /// Generate a USR fragment for an Objective-C instance variable.  The
96  /// complete USR can be created by concatenating the USR for the
97  /// encompassing class with this USR fragment.
98  void GenObjCIvar(llvm::StringRef ivar);
99  /// Generate a USR fragment for an Objective-C method.
100  void GenObjCMethod(llvm::StringRef sel, bool isInstanceMethod);
101  /// Generate a USR fragment for an Objective-C property.
102  void GenObjCProperty(llvm::StringRef prop);
103  /// Generate a USR for an Objective-C protocol.
104  void GenObjCProtocol(llvm::StringRef prot);
105
106  void VisitType(QualType T);
107
108  /// Emit a Decl's name using NamedDecl::printName() and return true if
109  ///  the decl had no name.
110  bool EmitDeclName(const NamedDecl *D);
111};
112
113} // end anonymous namespace
114
115//===----------------------------------------------------------------------===//
116// Generating USRs from ASTS.
117//===----------------------------------------------------------------------===//
118
119bool USRGenerator::EmitDeclName(const NamedDecl *D) {
120  Out.flush();
121  const unsigned startSize = Buf.size();
122  D->printName(Out);
123  Out.flush();
124  const unsigned endSize = Buf.size();
125  return startSize == endSize;
126}
127
128static bool InAnonymousNamespace(const Decl *D) {
129  if (const NamespaceDecl *ND = dyn_cast<NamespaceDecl>(D->getDeclContext()))
130    return ND->isAnonymousNamespace();
131  return false;
132}
133
134static inline bool ShouldGenerateLocation(const NamedDecl *D) {
135  return D->getLinkage() != ExternalLinkage && !InAnonymousNamespace(D);
136}
137
138void USRGenerator::VisitDeclContext(DeclContext *DC) {
139  if (NamedDecl *D = dyn_cast<NamedDecl>(DC))
140    Visit(D);
141}
142
143void USRGenerator::VisitFieldDecl(FieldDecl *D) {
144  VisitDeclContext(D->getDeclContext());
145  Out << (isa<ObjCIvarDecl>(D) ? "@" : "@FI@");
146  if (EmitDeclName(D)) {
147    // Bit fields can be anonymous.
148    IgnoreResults = true;
149    return;
150  }
151}
152
153void USRGenerator::VisitFunctionDecl(FunctionDecl *D) {
154  if (ShouldGenerateLocation(D) && GenLoc(D))
155    return;
156
157  VisitDeclContext(D->getDeclContext());
158  Out << "@F@";
159  D->printName(Out);
160
161  ASTContext &Ctx = AU->getASTContext();
162  if (!Ctx.getLangOptions().CPlusPlus || D->isExternC())
163    return;
164
165  // Mangle in type information for the arguments.
166  for (FunctionDecl::param_iterator I = D->param_begin(), E = D->param_end();
167       I != E; ++I) {
168    Out << '#';
169    if (ParmVarDecl *PD = *I)
170      VisitType(PD->getType());
171  }
172  if (D->isVariadic())
173    Out << '.';
174  Out << '#';
175  if (CXXMethodDecl *MD = dyn_cast<CXXMethodDecl>(D)) {
176    if (MD->isStatic())
177      Out << 'S';
178    if (unsigned quals = MD->getTypeQualifiers())
179      Out << (char)('0' + quals);
180  }
181}
182
183void USRGenerator::VisitNamedDecl(NamedDecl *D) {
184  VisitDeclContext(D->getDeclContext());
185  Out << "@";
186
187  if (EmitDeclName(D)) {
188    // The string can be empty if the declaration has no name; e.g., it is
189    // the ParmDecl with no name for declaration of a function pointer type,
190    // e.g.: void  (*f)(void *);
191    // In this case, don't generate a USR.
192    IgnoreResults = true;
193  }
194}
195
196void USRGenerator::VisitVarDecl(VarDecl *D) {
197  // VarDecls can be declared 'extern' within a function or method body,
198  // but their enclosing DeclContext is the function, not the TU.  We need
199  // to check the storage class to correctly generate the USR.
200  if (ShouldGenerateLocation(D) && GenLoc(D))
201    return;
202
203  VisitDeclContext(D->getDeclContext());
204
205  // Variables always have simple names.
206  llvm::StringRef s = D->getName();
207
208  // The string can be empty if the declaration has no name; e.g., it is
209  // the ParmDecl with no name for declaration of a function pointer type, e.g.:
210  //  	void  (*f)(void *);
211  // In this case, don't generate a USR.
212  if (s.empty())
213    IgnoreResults = true;
214  else
215    Out << '@' << s;
216}
217
218void USRGenerator::VisitNamespaceDecl(NamespaceDecl *D) {
219  if (D->isAnonymousNamespace()) {
220    Out << "@aN";
221    return;
222  }
223
224  VisitDeclContext(D->getDeclContext());
225  if (!IgnoreResults)
226    Out << "@N@" << D->getName();
227}
228
229void USRGenerator::VisitObjCMethodDecl(ObjCMethodDecl *D) {
230  Visit(cast<Decl>(D->getDeclContext()));
231  // Ideally we would use 'GenObjCMethod', but this is such a hot path
232  // for Objective-C code that we don't want to use
233  // DeclarationName::getAsString().
234  Out << (D->isInstanceMethod() ? "(im)" : "(cm)");
235  DeclarationName N(D->getSelector());
236  N.printName(Out);
237}
238
239void USRGenerator::VisitObjCClassDecl(ObjCClassDecl *D) {
240  // FIXME: @class declarations can refer to multiple classes.  We need
241  //  to be able to traverse these.
242  IgnoreResults = true;
243}
244
245void USRGenerator::VisitObjCForwardProtocolDecl(ObjCForwardProtocolDecl *D) {
246  // FIXME: @protocol declarations can refer to multiple protocols.  We need
247  //  to be able to traverse these.
248  IgnoreResults = true;
249}
250
251void USRGenerator::VisitObjCContainerDecl(ObjCContainerDecl *D) {
252  switch (D->getKind()) {
253    default:
254      assert(false && "Invalid ObjC container.");
255    case Decl::ObjCInterface:
256    case Decl::ObjCImplementation:
257      GenObjCClass(D->getName());
258      break;
259    case Decl::ObjCCategory: {
260      ObjCCategoryDecl *CD = cast<ObjCCategoryDecl>(D);
261      ObjCInterfaceDecl *ID = CD->getClassInterface();
262      if (!ID) {
263        // Handle invalid code where the @interface might not
264        // have been specified.
265        // FIXME: We should be able to generate this USR even if the
266        // @interface isn't available.
267        IgnoreResults = true;
268        return;
269      }
270      GenObjCCategory(ID->getName(), CD->getName());
271      break;
272    }
273    case Decl::ObjCCategoryImpl: {
274      ObjCCategoryImplDecl *CD = cast<ObjCCategoryImplDecl>(D);
275      ObjCInterfaceDecl *ID = CD->getClassInterface();
276      if (!ID) {
277        // Handle invalid code where the @interface might not
278        // have been specified.
279        // FIXME: We should be able to generate this USR even if the
280        // @interface isn't available.
281        IgnoreResults = true;
282        return;
283      }
284      GenObjCCategory(ID->getName(), CD->getName());
285      break;
286    }
287    case Decl::ObjCProtocol:
288      GenObjCProtocol(cast<ObjCProtocolDecl>(D)->getName());
289      break;
290  }
291}
292
293void USRGenerator::VisitObjCPropertyDecl(ObjCPropertyDecl *D) {
294  Visit(cast<Decl>(D->getDeclContext()));
295  GenObjCProperty(D->getName());
296}
297
298void USRGenerator::VisitObjCPropertyImplDecl(ObjCPropertyImplDecl *D) {
299  if (ObjCPropertyDecl *PD = D->getPropertyDecl()) {
300    VisitObjCPropertyDecl(PD);
301    return;
302  }
303
304  IgnoreResults = true;
305}
306
307void USRGenerator::VisitTagDecl(TagDecl *D) {
308  // Add the location of the tag decl to handle resolution across
309  // translation units.
310  if (ShouldGenerateLocation(D) && GenLoc(D))
311    return;
312
313  D = D->getCanonicalDecl();
314  VisitDeclContext(D->getDeclContext());
315
316  switch (D->getTagKind()) {
317    case TTK_Struct: Out << "@S"; break;
318    case TTK_Class:  Out << "@C"; break;
319    case TTK_Union:  Out << "@U"; break;
320    case TTK_Enum:   Out << "@E"; break;
321  }
322
323  Out << '@';
324  Out.flush();
325  assert(Buf.size() > 0);
326  const unsigned off = Buf.size() - 1;
327
328  if (EmitDeclName(D)) {
329    if (const TypedefDecl *TD = D->getTypedefForAnonDecl()) {
330      Buf[off] = 'A';
331      Out << '@' << TD;
332    }
333    else
334      Buf[off] = 'a';
335  }
336}
337
338void USRGenerator::VisitTypedefDecl(TypedefDecl *D) {
339  if (ShouldGenerateLocation(D) && GenLoc(D))
340    return;
341  DeclContext *DC = D->getDeclContext();
342  if (NamedDecl *DCN = dyn_cast<NamedDecl>(DC))
343    Visit(DCN);
344  Out << "@T@";
345  Out << D->getName();
346}
347
348bool USRGenerator::GenLoc(const Decl *D) {
349  if (generatedLoc)
350    return IgnoreResults;
351  generatedLoc = true;
352
353  const SourceManager &SM = AU->getSourceManager();
354  SourceLocation L = D->getLocStart();
355  if (L.isInvalid()) {
356    IgnoreResults = true;
357    return true;
358  }
359  L = SM.getInstantiationLoc(L);
360  const std::pair<FileID, unsigned> &Decomposed = SM.getDecomposedLoc(L);
361  const FileEntry *FE = SM.getFileEntryForID(Decomposed.first);
362  if (FE) {
363    llvm::sys::Path P(FE->getName());
364    Out << P.getLast();
365  }
366  else {
367    // This case really isn't interesting.
368    IgnoreResults = true;
369    return true;
370  }
371  Out << '@'
372      << SM.getLineNumber(Decomposed.first, Decomposed.second) << ':'
373      << SM.getColumnNumber(Decomposed.first, Decomposed.second);
374
375  return IgnoreResults;
376}
377
378void USRGenerator::VisitType(QualType T) {
379  // This method mangles in USR information for types.  It can possibly
380  // just reuse the naming-mangling logic used by codegen, although the
381  // requirements for USRs might not be the same.
382  ASTContext &Ctx = AU->getASTContext();
383
384  do {
385    T = Ctx.getCanonicalType(T);
386    Qualifiers Q = T.getQualifiers();
387    unsigned qVal = 0;
388    if (Q.hasConst())
389      qVal |= 0x1;
390    if (Q.hasVolatile())
391      qVal |= 0x2;
392    if (Q.hasRestrict())
393      qVal |= 0x4;
394    if(qVal)
395      Out << ((char) ('0' + qVal));
396
397    // Mangle in ObjC GC qualifiers?
398
399    if (const PointerType *PT = T->getAs<PointerType>()) {
400      Out << '*';
401      T = PT->getPointeeType();
402      continue;
403    }
404    if (const ReferenceType *RT = T->getAs<ReferenceType>()) {
405      Out << '&';
406      T = RT->getPointeeType();
407      continue;
408    }
409    if (const FunctionProtoType *FT = T->getAs<FunctionProtoType>()) {
410      Out << 'F';
411      VisitType(FT->getResultType());
412      for (FunctionProtoType::arg_type_iterator
413            I = FT->arg_type_begin(), E = FT->arg_type_end(); I!=E; ++I) {
414        VisitType(*I);
415      }
416      if (FT->isVariadic())
417        Out << '.';
418      return;
419    }
420    if (const BlockPointerType *BT = T->getAs<BlockPointerType>()) {
421      Out << 'B';
422      T = BT->getPointeeType();
423      continue;
424    }
425    if (const BuiltinType *BT = T->getAs<BuiltinType>()) {
426      unsigned char c = '\0';
427      switch (BT->getKind()) {
428        case BuiltinType::Void:
429          c = 'v'; break;
430        case BuiltinType::Bool:
431          c = 'b'; break;
432        case BuiltinType::Char_U:
433        case BuiltinType::UChar:
434          c = 'c'; break;
435        case BuiltinType::Char16:
436          c = 'q'; break;
437        case BuiltinType::Char32:
438          c = 'w'; break;
439        case BuiltinType::UShort:
440          c = 's'; break;
441        case BuiltinType::UInt:
442          c = 'i'; break;
443        case BuiltinType::ULong:
444          c = 'l'; break;
445        case BuiltinType::ULongLong:
446          c = 'k'; break;
447        case BuiltinType::UInt128:
448          c = 'j'; break;
449        case BuiltinType::Char_S:
450        case BuiltinType::SChar:
451          c = 'C'; break;
452        case BuiltinType::WChar:
453          c = 'W'; break;
454        case BuiltinType::Short:
455          c = 'S'; break;
456        case BuiltinType::Int:
457          c = 'I'; break;
458        case BuiltinType::Long:
459          c = 'L'; break;
460        case BuiltinType::LongLong:
461          c = 'K'; break;
462        case BuiltinType::Int128:
463          c = 'J'; break;
464        case BuiltinType::Float:
465          c = 'f'; break;
466        case BuiltinType::Double:
467          c = 'd'; break;
468        case BuiltinType::LongDouble:
469          c = 'D'; break;
470        case BuiltinType::NullPtr:
471          c = 'n'; break;
472        case BuiltinType::Overload:
473        case BuiltinType::Dependent:
474        case BuiltinType::UndeducedAuto:
475          IgnoreResults = true;
476          return;
477        case BuiltinType::ObjCId:
478          c = 'o'; break;
479        case BuiltinType::ObjCClass:
480          c = 'O'; break;
481        case BuiltinType::ObjCSel:
482          c = 'e'; break;
483      }
484      Out << c;
485      return;
486    }
487    if (const ComplexType *CT = T->getAs<ComplexType>()) {
488      Out << '<';
489      T = CT->getElementType();
490      continue;
491    }
492    if (const TagType *TT = T->getAs<TagType>()) {
493      Out << '$';
494      VisitTagDecl(TT->getDecl());
495      return;
496    }
497
498    // Unhandled type.
499    Out << ' ';
500    break;
501  } while (true);
502}
503
504//===----------------------------------------------------------------------===//
505// General purpose USR generation methods.
506//===----------------------------------------------------------------------===//
507
508void USRGenerator::GenObjCClass(llvm::StringRef cls) {
509  Out << "objc(cs)" << cls;
510}
511
512void USRGenerator::GenObjCCategory(llvm::StringRef cls, llvm::StringRef cat) {
513  Out << "objc(cy)" << cls << '@' << cat;
514}
515
516void USRGenerator::GenObjCIvar(llvm::StringRef ivar) {
517  Out << '@' << ivar;
518}
519
520void USRGenerator::GenObjCMethod(llvm::StringRef meth, bool isInstanceMethod) {
521  Out << (isInstanceMethod ? "(im)" : "(cm)") << meth;
522}
523
524void USRGenerator::GenObjCProperty(llvm::StringRef prop) {
525  Out << "(py)" << prop;
526}
527
528void USRGenerator::GenObjCProtocol(llvm::StringRef prot) {
529  Out << "objc(pl)" << prot;
530}
531
532//===----------------------------------------------------------------------===//
533// API hooks.
534//===----------------------------------------------------------------------===//
535
536static inline llvm::StringRef extractUSRSuffix(llvm::StringRef s) {
537  return s.startswith("c:") ? s.substr(2) : "";
538}
539
540static CXString getDeclCursorUSR(const CXCursor &C) {
541  Decl *D = cxcursor::getCursorDecl(C);
542
543  // Don't generate USRs for things with invalid locations.
544  if (!D || D->getLocStart().isInvalid())
545    return createCXString("");
546
547  // Check if the cursor has 'NoLinkage'.
548  if (const NamedDecl *ND = dyn_cast<NamedDecl>(D))
549    switch (ND->getLinkage()) {
550      case ExternalLinkage:
551        // Generate USRs for all entities with external linkage.
552        break;
553      case NoLinkage:
554      case UniqueExternalLinkage:
555        // We allow enums, typedefs, and structs that have no linkage to
556        // have USRs that are anchored to the file they were defined in
557        // (e.g., the header).  This is a little gross, but in principal
558        // enums/anonymous structs/etc. defined in a common header file
559        // are referred to across multiple translation units.
560        if (isa<TagDecl>(ND) || isa<TypedefDecl>(ND) ||
561            isa<EnumConstantDecl>(ND) || isa<FieldDecl>(ND) ||
562            isa<VarDecl>(ND) || isa<NamespaceDecl>(ND))
563          break;
564        // Fall-through.
565      case InternalLinkage:
566        if (isa<FunctionDecl>(ND))
567          break;
568    }
569
570  USRGenerator UG(&C);
571  UG->Visit(D);
572
573  if (UG->ignoreResults())
574    return createCXString("");
575
576#if 0
577  // For development testing.
578  assert(UG.str().size() > 2);
579#endif
580
581    // Return a copy of the string that must be disposed by the caller.
582  return createCXString(UG.str(), true);
583}
584
585extern "C" {
586
587CXString clang_getCursorUSR(CXCursor C) {
588  const CXCursorKind &K = clang_getCursorKind(C);
589
590  if (clang_isDeclaration(K))
591      return getDeclCursorUSR(C);
592
593  if (K == CXCursor_MacroDefinition) {
594    USRGenerator UG(&C);
595    UG << "macro@"
596       << cxcursor::getCursorMacroDefinition(C)->getName()->getNameStart();
597    return createCXString(UG.str(), true);
598  }
599
600  return createCXString("");
601}
602
603CXString clang_constructUSR_ObjCIvar(const char *name, CXString classUSR) {
604  USRGenerator UG;
605  UG << extractUSRSuffix(clang_getCString(classUSR));
606  UG->GenObjCIvar(name);
607  return createCXString(UG.str(), true);
608}
609
610CXString clang_constructUSR_ObjCMethod(const char *name,
611                                       unsigned isInstanceMethod,
612                                       CXString classUSR) {
613  USRGenerator UG;
614  UG << extractUSRSuffix(clang_getCString(classUSR));
615  UG->GenObjCMethod(name, isInstanceMethod);
616  return createCXString(UG.str(), true);
617}
618
619CXString clang_constructUSR_ObjCClass(const char *name) {
620  USRGenerator UG;
621  UG->GenObjCClass(name);
622  return createCXString(UG.str(), true);
623}
624
625CXString clang_constructUSR_ObjCProtocol(const char *name) {
626  USRGenerator UG;
627  UG->GenObjCProtocol(name);
628  return createCXString(UG.str(), true);
629}
630
631CXString clang_constructUSR_ObjCCategory(const char *class_name,
632                                         const char *category_name) {
633  USRGenerator UG;
634  UG->GenObjCCategory(class_name, category_name);
635  return createCXString(UG.str(), true);
636}
637
638CXString clang_constructUSR_ObjCProperty(const char *property,
639                                         CXString classUSR) {
640  USRGenerator UG;
641  UG << extractUSRSuffix(clang_getCString(classUSR));
642  UG->GenObjCProperty(property);
643  return createCXString(UG.str(), true);
644}
645
646} // end extern "C"
647