1 //===--- ASTMatchFinder.cpp - Structural query framework ------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 //  Implements an algorithm to efficiently search for matches on AST nodes.
10 //  Uses memoization to support recursive matches like HasDescendant.
11 //
12 //  The general idea is to visit all AST nodes with a RecursiveASTVisitor,
13 //  calling the Matches(...) method of each matcher we are running on each
14 //  AST node. The matcher can recurse via the ASTMatchFinder interface.
15 //
16 //===----------------------------------------------------------------------===//
17 
18 #include "clang/ASTMatchers/ASTMatchFinder.h"
19 #include "clang/AST/ASTConsumer.h"
20 #include "clang/AST/ASTContext.h"
21 #include "clang/AST/RecursiveASTVisitor.h"
22 #include "llvm/ADT/DenseMap.h"
23 #include "llvm/ADT/StringMap.h"
24 #include "llvm/Support/Timer.h"
25 #include <deque>
26 #include <memory>
27 #include <set>
28 
29 namespace clang {
30 namespace ast_matchers {
31 namespace internal {
32 namespace {
33 
34 typedef MatchFinder::MatchCallback MatchCallback;
35 
36 // The maximum number of memoization entries to store.
37 // 10k has been experimentally found to give a good trade-off
38 // of performance vs. memory consumption by running matcher
39 // that match on every statement over a very large codebase.
40 //
41 // FIXME: Do some performance optimization in general and
42 // revisit this number; also, put up micro-benchmarks that we can
43 // optimize this on.
44 static const unsigned MaxMemoizationEntries = 10000;
45 
46 // We use memoization to avoid running the same matcher on the same
47 // AST node twice.  This struct is the key for looking up match
48 // result.  It consists of an ID of the MatcherInterface (for
49 // identifying the matcher), a pointer to the AST node and the
50 // bound nodes before the matcher was executed.
51 //
52 // We currently only memoize on nodes whose pointers identify the
53 // nodes (\c Stmt and \c Decl, but not \c QualType or \c TypeLoc).
54 // For \c QualType and \c TypeLoc it is possible to implement
55 // generation of keys for each type.
56 // FIXME: Benchmark whether memoization of non-pointer typed nodes
57 // provides enough benefit for the additional amount of code.
58 struct MatchKey {
59   DynTypedMatcher::MatcherIDType MatcherID;
60   DynTypedNode Node;
61   BoundNodesTreeBuilder BoundNodes;
62   TraversalKind Traversal = TK_AsIs;
63 
64   bool operator<(const MatchKey &Other) const {
65     return std::tie(Traversal, MatcherID, Node, BoundNodes) <
66            std::tie(Other.Traversal, Other.MatcherID, Other.Node,
67                     Other.BoundNodes);
68   }
69 };
70 
71 // Used to store the result of a match and possibly bound nodes.
72 struct MemoizedMatchResult {
73   bool ResultOfMatch;
74   BoundNodesTreeBuilder Nodes;
75 };
76 
77 // A RecursiveASTVisitor that traverses all children or all descendants of
78 // a node.
79 class MatchChildASTVisitor
80     : public RecursiveASTVisitor<MatchChildASTVisitor> {
81 public:
82   typedef RecursiveASTVisitor<MatchChildASTVisitor> VisitorBase;
83 
84   // Creates an AST visitor that matches 'matcher' on all children or
85   // descendants of a traversed node. max_depth is the maximum depth
86   // to traverse: use 1 for matching the children and INT_MAX for
87   // matching the descendants.
88   MatchChildASTVisitor(const DynTypedMatcher *Matcher, ASTMatchFinder *Finder,
89                        BoundNodesTreeBuilder *Builder, int MaxDepth,
90                        TraversalKind Traversal, ASTMatchFinder::BindKind Bind)
91       : Matcher(Matcher), Finder(Finder), Builder(Builder), CurrentDepth(0),
92         MaxDepth(MaxDepth), Traversal(Traversal), Bind(Bind), Matches(false) {}
93 
94   // Returns true if a match is found in the subtree rooted at the
95   // given AST node. This is done via a set of mutually recursive
96   // functions. Here's how the recursion is done (the  *wildcard can
97   // actually be Decl, Stmt, or Type):
98   //
99   //   - Traverse(node) calls BaseTraverse(node) when it needs
100   //     to visit the descendants of node.
101   //   - BaseTraverse(node) then calls (via VisitorBase::Traverse*(node))
102   //     Traverse*(c) for each child c of 'node'.
103   //   - Traverse*(c) in turn calls Traverse(c), completing the
104   //     recursion.
105   bool findMatch(const DynTypedNode &DynNode) {
106     reset();
107     if (const Decl *D = DynNode.get<Decl>())
108       traverse(*D);
109     else if (const Stmt *S = DynNode.get<Stmt>())
110       traverse(*S);
111     else if (const NestedNameSpecifier *NNS =
112              DynNode.get<NestedNameSpecifier>())
113       traverse(*NNS);
114     else if (const NestedNameSpecifierLoc *NNSLoc =
115              DynNode.get<NestedNameSpecifierLoc>())
116       traverse(*NNSLoc);
117     else if (const QualType *Q = DynNode.get<QualType>())
118       traverse(*Q);
119     else if (const TypeLoc *T = DynNode.get<TypeLoc>())
120       traverse(*T);
121     else if (const auto *C = DynNode.get<CXXCtorInitializer>())
122       traverse(*C);
123     // FIXME: Add other base types after adding tests.
124 
125     // It's OK to always overwrite the bound nodes, as if there was
126     // no match in this recursive branch, the result set is empty
127     // anyway.
128     *Builder = ResultBindings;
129 
130     return Matches;
131   }
132 
133   // The following are overriding methods from the base visitor class.
134   // They are public only to allow CRTP to work. They are *not *part
135   // of the public API of this class.
136   bool TraverseDecl(Decl *DeclNode) {
137     ScopedIncrement ScopedDepth(&CurrentDepth);
138     return (DeclNode == nullptr) || traverse(*DeclNode);
139   }
140 
141   Stmt *getStmtToTraverse(Stmt *StmtNode) {
142     Stmt *StmtToTraverse = StmtNode;
143     if (auto *ExprNode = dyn_cast_or_null<Expr>(StmtNode)) {
144       auto *LambdaNode = dyn_cast_or_null<LambdaExpr>(StmtNode);
145       if (LambdaNode &&
146           Finder->getASTContext().getParentMapContext().getTraversalKind() ==
147               TK_IgnoreUnlessSpelledInSource)
148         StmtToTraverse = LambdaNode;
149       else
150         StmtToTraverse =
151             Finder->getASTContext().getParentMapContext().traverseIgnored(
152                 ExprNode);
153     }
154     if (Traversal == TraversalKind::TK_IgnoreImplicitCastsAndParentheses) {
155       if (Expr *ExprNode = dyn_cast_or_null<Expr>(StmtNode))
156         StmtToTraverse = ExprNode->IgnoreParenImpCasts();
157     }
158     return StmtToTraverse;
159   }
160 
161   bool TraverseStmt(Stmt *StmtNode, DataRecursionQueue *Queue = nullptr) {
162     // If we need to keep track of the depth, we can't perform data recursion.
163     if (CurrentDepth == 0 || (CurrentDepth <= MaxDepth && MaxDepth < INT_MAX))
164       Queue = nullptr;
165 
166     ScopedIncrement ScopedDepth(&CurrentDepth);
167     Stmt *StmtToTraverse = getStmtToTraverse(StmtNode);
168     if (!StmtToTraverse)
169       return true;
170     if (!match(*StmtToTraverse))
171       return false;
172     return VisitorBase::TraverseStmt(StmtToTraverse, Queue);
173   }
174   // We assume that the QualType and the contained type are on the same
175   // hierarchy level. Thus, we try to match either of them.
176   bool TraverseType(QualType TypeNode) {
177     if (TypeNode.isNull())
178       return true;
179     ScopedIncrement ScopedDepth(&CurrentDepth);
180     // Match the Type.
181     if (!match(*TypeNode))
182       return false;
183     // The QualType is matched inside traverse.
184     return traverse(TypeNode);
185   }
186   // We assume that the TypeLoc, contained QualType and contained Type all are
187   // on the same hierarchy level. Thus, we try to match all of them.
188   bool TraverseTypeLoc(TypeLoc TypeLocNode) {
189     if (TypeLocNode.isNull())
190       return true;
191     ScopedIncrement ScopedDepth(&CurrentDepth);
192     // Match the Type.
193     if (!match(*TypeLocNode.getType()))
194       return false;
195     // Match the QualType.
196     if (!match(TypeLocNode.getType()))
197       return false;
198     // The TypeLoc is matched inside traverse.
199     return traverse(TypeLocNode);
200   }
201   bool TraverseNestedNameSpecifier(NestedNameSpecifier *NNS) {
202     ScopedIncrement ScopedDepth(&CurrentDepth);
203     return (NNS == nullptr) || traverse(*NNS);
204   }
205   bool TraverseNestedNameSpecifierLoc(NestedNameSpecifierLoc NNS) {
206     if (!NNS)
207       return true;
208     ScopedIncrement ScopedDepth(&CurrentDepth);
209     if (!match(*NNS.getNestedNameSpecifier()))
210       return false;
211     return traverse(NNS);
212   }
213   bool TraverseConstructorInitializer(CXXCtorInitializer *CtorInit) {
214     if (!CtorInit)
215       return true;
216     ScopedIncrement ScopedDepth(&CurrentDepth);
217     return traverse(*CtorInit);
218   }
219   bool TraverseLambdaExpr(LambdaExpr *Node) {
220     if (Finder->getASTContext().getParentMapContext().getTraversalKind() !=
221         TK_IgnoreUnlessSpelledInSource)
222       return VisitorBase::TraverseLambdaExpr(Node);
223     if (!Node)
224       return true;
225     ScopedIncrement ScopedDepth(&CurrentDepth);
226 
227     for (unsigned I = 0, N = Node->capture_size(); I != N; ++I) {
228       const auto *C = Node->capture_begin() + I;
229       if (!C->isExplicit())
230         continue;
231       if (Node->isInitCapture(C) && !match(*C->getCapturedVar()))
232         return false;
233       if (!match(*Node->capture_init_begin()[I]))
234         return false;
235     }
236 
237     if (const auto *TPL = Node->getTemplateParameterList()) {
238       for (const auto *TP : *TPL) {
239         if (!match(*TP))
240           return false;
241       }
242     }
243 
244     for (const auto *P : Node->getCallOperator()->parameters()) {
245       if (!match(*P))
246         return false;
247     }
248 
249     if (!match(*Node->getBody()))
250       return false;
251 
252     return true;
253   }
254 
255   bool shouldVisitTemplateInstantiations() const { return true; }
256   bool shouldVisitImplicitCode() const { return true; }
257 
258 private:
259   // Used for updating the depth during traversal.
260   struct ScopedIncrement {
261     explicit ScopedIncrement(int *Depth) : Depth(Depth) { ++(*Depth); }
262     ~ScopedIncrement() { --(*Depth); }
263 
264    private:
265     int *Depth;
266   };
267 
268   // Resets the state of this object.
269   void reset() {
270     Matches = false;
271     CurrentDepth = 0;
272   }
273 
274   // Forwards the call to the corresponding Traverse*() method in the
275   // base visitor class.
276   bool baseTraverse(const Decl &DeclNode) {
277     return VisitorBase::TraverseDecl(const_cast<Decl*>(&DeclNode));
278   }
279   bool baseTraverse(const Stmt &StmtNode) {
280     return VisitorBase::TraverseStmt(const_cast<Stmt*>(&StmtNode));
281   }
282   bool baseTraverse(QualType TypeNode) {
283     return VisitorBase::TraverseType(TypeNode);
284   }
285   bool baseTraverse(TypeLoc TypeLocNode) {
286     return VisitorBase::TraverseTypeLoc(TypeLocNode);
287   }
288   bool baseTraverse(const NestedNameSpecifier &NNS) {
289     return VisitorBase::TraverseNestedNameSpecifier(
290         const_cast<NestedNameSpecifier*>(&NNS));
291   }
292   bool baseTraverse(NestedNameSpecifierLoc NNS) {
293     return VisitorBase::TraverseNestedNameSpecifierLoc(NNS);
294   }
295   bool baseTraverse(const CXXCtorInitializer &CtorInit) {
296     return VisitorBase::TraverseConstructorInitializer(
297         const_cast<CXXCtorInitializer *>(&CtorInit));
298   }
299 
300   // Sets 'Matched' to true if 'Matcher' matches 'Node' and:
301   //   0 < CurrentDepth <= MaxDepth.
302   //
303   // Returns 'true' if traversal should continue after this function
304   // returns, i.e. if no match is found or 'Bind' is 'BK_All'.
305   template <typename T>
306   bool match(const T &Node) {
307     if (CurrentDepth == 0 || CurrentDepth > MaxDepth) {
308       return true;
309     }
310     if (Bind != ASTMatchFinder::BK_All) {
311       BoundNodesTreeBuilder RecursiveBuilder(*Builder);
312       if (Matcher->matches(DynTypedNode::create(Node), Finder,
313                            &RecursiveBuilder)) {
314         Matches = true;
315         ResultBindings.addMatch(RecursiveBuilder);
316         return false; // Abort as soon as a match is found.
317       }
318     } else {
319       BoundNodesTreeBuilder RecursiveBuilder(*Builder);
320       if (Matcher->matches(DynTypedNode::create(Node), Finder,
321                            &RecursiveBuilder)) {
322         // After the first match the matcher succeeds.
323         Matches = true;
324         ResultBindings.addMatch(RecursiveBuilder);
325       }
326     }
327     return true;
328   }
329 
330   // Traverses the subtree rooted at 'Node'; returns true if the
331   // traversal should continue after this function returns.
332   template <typename T>
333   bool traverse(const T &Node) {
334     static_assert(IsBaseType<T>::value,
335                   "traverse can only be instantiated with base type");
336     if (!match(Node))
337       return false;
338     return baseTraverse(Node);
339   }
340 
341   const DynTypedMatcher *const Matcher;
342   ASTMatchFinder *const Finder;
343   BoundNodesTreeBuilder *const Builder;
344   BoundNodesTreeBuilder ResultBindings;
345   int CurrentDepth;
346   const int MaxDepth;
347   const TraversalKind Traversal;
348   const ASTMatchFinder::BindKind Bind;
349   bool Matches;
350 };
351 
352 // Controls the outermost traversal of the AST and allows to match multiple
353 // matchers.
354 class MatchASTVisitor : public RecursiveASTVisitor<MatchASTVisitor>,
355                         public ASTMatchFinder {
356 public:
357   MatchASTVisitor(const MatchFinder::MatchersByType *Matchers,
358                   const MatchFinder::MatchFinderOptions &Options)
359       : Matchers(Matchers), Options(Options), ActiveASTContext(nullptr) {}
360 
361   ~MatchASTVisitor() override {
362     if (Options.CheckProfiling) {
363       Options.CheckProfiling->Records = std::move(TimeByBucket);
364     }
365   }
366 
367   void onStartOfTranslationUnit() {
368     const bool EnableCheckProfiling = Options.CheckProfiling.hasValue();
369     TimeBucketRegion Timer;
370     for (MatchCallback *MC : Matchers->AllCallbacks) {
371       if (EnableCheckProfiling)
372         Timer.setBucket(&TimeByBucket[MC->getID()]);
373       MC->onStartOfTranslationUnit();
374     }
375   }
376 
377   void onEndOfTranslationUnit() {
378     const bool EnableCheckProfiling = Options.CheckProfiling.hasValue();
379     TimeBucketRegion Timer;
380     for (MatchCallback *MC : Matchers->AllCallbacks) {
381       if (EnableCheckProfiling)
382         Timer.setBucket(&TimeByBucket[MC->getID()]);
383       MC->onEndOfTranslationUnit();
384     }
385   }
386 
387   void set_active_ast_context(ASTContext *NewActiveASTContext) {
388     ActiveASTContext = NewActiveASTContext;
389   }
390 
391   // The following Visit*() and Traverse*() functions "override"
392   // methods in RecursiveASTVisitor.
393 
394   bool VisitTypedefNameDecl(TypedefNameDecl *DeclNode) {
395     // When we see 'typedef A B', we add name 'B' to the set of names
396     // A's canonical type maps to.  This is necessary for implementing
397     // isDerivedFrom(x) properly, where x can be the name of the base
398     // class or any of its aliases.
399     //
400     // In general, the is-alias-of (as defined by typedefs) relation
401     // is tree-shaped, as you can typedef a type more than once.  For
402     // example,
403     //
404     //   typedef A B;
405     //   typedef A C;
406     //   typedef C D;
407     //   typedef C E;
408     //
409     // gives you
410     //
411     //   A
412     //   |- B
413     //   `- C
414     //      |- D
415     //      `- E
416     //
417     // It is wrong to assume that the relation is a chain.  A correct
418     // implementation of isDerivedFrom() needs to recognize that B and
419     // E are aliases, even though neither is a typedef of the other.
420     // Therefore, we cannot simply walk through one typedef chain to
421     // find out whether the type name matches.
422     const Type *TypeNode = DeclNode->getUnderlyingType().getTypePtr();
423     const Type *CanonicalType =  // root of the typedef tree
424         ActiveASTContext->getCanonicalType(TypeNode);
425     TypeAliases[CanonicalType].insert(DeclNode);
426     return true;
427   }
428 
429   bool VisitObjCCompatibleAliasDecl(ObjCCompatibleAliasDecl *CAD) {
430     const ObjCInterfaceDecl *InterfaceDecl = CAD->getClassInterface();
431     CompatibleAliases[InterfaceDecl].insert(CAD);
432     return true;
433   }
434 
435   bool TraverseDecl(Decl *DeclNode);
436   bool TraverseStmt(Stmt *StmtNode, DataRecursionQueue *Queue = nullptr);
437   bool TraverseType(QualType TypeNode);
438   bool TraverseTypeLoc(TypeLoc TypeNode);
439   bool TraverseNestedNameSpecifier(NestedNameSpecifier *NNS);
440   bool TraverseNestedNameSpecifierLoc(NestedNameSpecifierLoc NNS);
441   bool TraverseConstructorInitializer(CXXCtorInitializer *CtorInit);
442 
443   // Matches children or descendants of 'Node' with 'BaseMatcher'.
444   bool memoizedMatchesRecursively(const DynTypedNode &Node, ASTContext &Ctx,
445                                   const DynTypedMatcher &Matcher,
446                                   BoundNodesTreeBuilder *Builder, int MaxDepth,
447                                   TraversalKind Traversal, BindKind Bind) {
448     // For AST-nodes that don't have an identity, we can't memoize.
449     if (!Node.getMemoizationData() || !Builder->isComparable())
450       return matchesRecursively(Node, Matcher, Builder, MaxDepth, Traversal,
451                                 Bind);
452 
453     MatchKey Key;
454     Key.MatcherID = Matcher.getID();
455     Key.Node = Node;
456     // Note that we key on the bindings *before* the match.
457     Key.BoundNodes = *Builder;
458     Key.Traversal = Ctx.getParentMapContext().getTraversalKind();
459 
460     MemoizationMap::iterator I = ResultCache.find(Key);
461     if (I != ResultCache.end()) {
462       *Builder = I->second.Nodes;
463       return I->second.ResultOfMatch;
464     }
465 
466     MemoizedMatchResult Result;
467     Result.Nodes = *Builder;
468     Result.ResultOfMatch = matchesRecursively(Node, Matcher, &Result.Nodes,
469                                               MaxDepth, Traversal, Bind);
470 
471     MemoizedMatchResult &CachedResult = ResultCache[Key];
472     CachedResult = std::move(Result);
473 
474     *Builder = CachedResult.Nodes;
475     return CachedResult.ResultOfMatch;
476   }
477 
478   // Matches children or descendants of 'Node' with 'BaseMatcher'.
479   bool matchesRecursively(const DynTypedNode &Node,
480                           const DynTypedMatcher &Matcher,
481                           BoundNodesTreeBuilder *Builder, int MaxDepth,
482                           TraversalKind Traversal, BindKind Bind) {
483     MatchChildASTVisitor Visitor(
484       &Matcher, this, Builder, MaxDepth, Traversal, Bind);
485     return Visitor.findMatch(Node);
486   }
487 
488   bool classIsDerivedFrom(const CXXRecordDecl *Declaration,
489                           const Matcher<NamedDecl> &Base,
490                           BoundNodesTreeBuilder *Builder,
491                           bool Directly) override;
492 
493   bool objcClassIsDerivedFrom(const ObjCInterfaceDecl *Declaration,
494                               const Matcher<NamedDecl> &Base,
495                               BoundNodesTreeBuilder *Builder,
496                               bool Directly) override;
497 
498   // Implements ASTMatchFinder::matchesChildOf.
499   bool matchesChildOf(const DynTypedNode &Node, ASTContext &Ctx,
500                       const DynTypedMatcher &Matcher,
501                       BoundNodesTreeBuilder *Builder, TraversalKind Traversal,
502                       BindKind Bind) override {
503     if (ResultCache.size() > MaxMemoizationEntries)
504       ResultCache.clear();
505     return memoizedMatchesRecursively(Node, Ctx, Matcher, Builder, 1, Traversal,
506                                       Bind);
507   }
508   // Implements ASTMatchFinder::matchesDescendantOf.
509   bool matchesDescendantOf(const DynTypedNode &Node, ASTContext &Ctx,
510                            const DynTypedMatcher &Matcher,
511                            BoundNodesTreeBuilder *Builder,
512                            BindKind Bind) override {
513     if (ResultCache.size() > MaxMemoizationEntries)
514       ResultCache.clear();
515     return memoizedMatchesRecursively(Node, Ctx, Matcher, Builder, INT_MAX,
516                                       TraversalKind::TK_AsIs, Bind);
517   }
518   // Implements ASTMatchFinder::matchesAncestorOf.
519   bool matchesAncestorOf(const DynTypedNode &Node, ASTContext &Ctx,
520                          const DynTypedMatcher &Matcher,
521                          BoundNodesTreeBuilder *Builder,
522                          AncestorMatchMode MatchMode) override {
523     // Reset the cache outside of the recursive call to make sure we
524     // don't invalidate any iterators.
525     if (ResultCache.size() > MaxMemoizationEntries)
526       ResultCache.clear();
527     return memoizedMatchesAncestorOfRecursively(Node, Ctx, Matcher, Builder,
528                                                 MatchMode);
529   }
530 
531   // Matches all registered matchers on the given node and calls the
532   // result callback for every node that matches.
533   void match(const DynTypedNode &Node) {
534     // FIXME: Improve this with a switch or a visitor pattern.
535     if (auto *N = Node.get<Decl>()) {
536       match(*N);
537     } else if (auto *N = Node.get<Stmt>()) {
538       match(*N);
539     } else if (auto *N = Node.get<Type>()) {
540       match(*N);
541     } else if (auto *N = Node.get<QualType>()) {
542       match(*N);
543     } else if (auto *N = Node.get<NestedNameSpecifier>()) {
544       match(*N);
545     } else if (auto *N = Node.get<NestedNameSpecifierLoc>()) {
546       match(*N);
547     } else if (auto *N = Node.get<TypeLoc>()) {
548       match(*N);
549     } else if (auto *N = Node.get<CXXCtorInitializer>()) {
550       match(*N);
551     }
552   }
553 
554   template <typename T> void match(const T &Node) {
555     matchDispatch(&Node);
556   }
557 
558   // Implements ASTMatchFinder::getASTContext.
559   ASTContext &getASTContext() const override { return *ActiveASTContext; }
560 
561   bool shouldVisitTemplateInstantiations() const { return true; }
562   bool shouldVisitImplicitCode() const { return true; }
563 
564 private:
565   class TimeBucketRegion {
566   public:
567     TimeBucketRegion() : Bucket(nullptr) {}
568     ~TimeBucketRegion() { setBucket(nullptr); }
569 
570     /// Start timing for \p NewBucket.
571     ///
572     /// If there was a bucket already set, it will finish the timing for that
573     /// other bucket.
574     /// \p NewBucket will be timed until the next call to \c setBucket() or
575     /// until the \c TimeBucketRegion is destroyed.
576     /// If \p NewBucket is the same as the currently timed bucket, this call
577     /// does nothing.
578     void setBucket(llvm::TimeRecord *NewBucket) {
579       if (Bucket != NewBucket) {
580         auto Now = llvm::TimeRecord::getCurrentTime(true);
581         if (Bucket)
582           *Bucket += Now;
583         if (NewBucket)
584           *NewBucket -= Now;
585         Bucket = NewBucket;
586       }
587     }
588 
589   private:
590     llvm::TimeRecord *Bucket;
591   };
592 
593   /// Runs all the \p Matchers on \p Node.
594   ///
595   /// Used by \c matchDispatch() below.
596   template <typename T, typename MC>
597   void matchWithoutFilter(const T &Node, const MC &Matchers) {
598     const bool EnableCheckProfiling = Options.CheckProfiling.hasValue();
599     TimeBucketRegion Timer;
600     for (const auto &MP : Matchers) {
601       if (EnableCheckProfiling)
602         Timer.setBucket(&TimeByBucket[MP.second->getID()]);
603       BoundNodesTreeBuilder Builder;
604       if (MP.first.matches(Node, this, &Builder)) {
605         MatchVisitor Visitor(ActiveASTContext, MP.second);
606         Builder.visitMatches(&Visitor);
607       }
608     }
609   }
610 
611   void matchWithFilter(const DynTypedNode &DynNode) {
612     auto Kind = DynNode.getNodeKind();
613     auto it = MatcherFiltersMap.find(Kind);
614     const auto &Filter =
615         it != MatcherFiltersMap.end() ? it->second : getFilterForKind(Kind);
616 
617     if (Filter.empty())
618       return;
619 
620     const bool EnableCheckProfiling = Options.CheckProfiling.hasValue();
621     TimeBucketRegion Timer;
622     auto &Matchers = this->Matchers->DeclOrStmt;
623     for (unsigned short I : Filter) {
624       auto &MP = Matchers[I];
625       if (EnableCheckProfiling)
626         Timer.setBucket(&TimeByBucket[MP.second->getID()]);
627       BoundNodesTreeBuilder Builder;
628       if (MP.first.matches(DynNode, this, &Builder)) {
629         MatchVisitor Visitor(ActiveASTContext, MP.second);
630         Builder.visitMatches(&Visitor);
631       }
632     }
633   }
634 
635   const std::vector<unsigned short> &getFilterForKind(ASTNodeKind Kind) {
636     auto &Filter = MatcherFiltersMap[Kind];
637     auto &Matchers = this->Matchers->DeclOrStmt;
638     assert((Matchers.size() < USHRT_MAX) && "Too many matchers.");
639     for (unsigned I = 0, E = Matchers.size(); I != E; ++I) {
640       if (Matchers[I].first.canMatchNodesOfKind(Kind)) {
641         Filter.push_back(I);
642       }
643     }
644     return Filter;
645   }
646 
647   /// @{
648   /// Overloads to pair the different node types to their matchers.
649   void matchDispatch(const Decl *Node) {
650     return matchWithFilter(DynTypedNode::create(*Node));
651   }
652   void matchDispatch(const Stmt *Node) {
653     return matchWithFilter(DynTypedNode::create(*Node));
654   }
655 
656   void matchDispatch(const Type *Node) {
657     matchWithoutFilter(QualType(Node, 0), Matchers->Type);
658   }
659   void matchDispatch(const TypeLoc *Node) {
660     matchWithoutFilter(*Node, Matchers->TypeLoc);
661   }
662   void matchDispatch(const QualType *Node) {
663     matchWithoutFilter(*Node, Matchers->Type);
664   }
665   void matchDispatch(const NestedNameSpecifier *Node) {
666     matchWithoutFilter(*Node, Matchers->NestedNameSpecifier);
667   }
668   void matchDispatch(const NestedNameSpecifierLoc *Node) {
669     matchWithoutFilter(*Node, Matchers->NestedNameSpecifierLoc);
670   }
671   void matchDispatch(const CXXCtorInitializer *Node) {
672     matchWithoutFilter(*Node, Matchers->CtorInit);
673   }
674   void matchDispatch(const void *) { /* Do nothing. */ }
675   /// @}
676 
677   // Returns whether an ancestor of \p Node matches \p Matcher.
678   //
679   // The order of matching ((which can lead to different nodes being bound in
680   // case there are multiple matches) is breadth first search.
681   //
682   // To allow memoization in the very common case of having deeply nested
683   // expressions inside a template function, we first walk up the AST, memoizing
684   // the result of the match along the way, as long as there is only a single
685   // parent.
686   //
687   // Once there are multiple parents, the breadth first search order does not
688   // allow simple memoization on the ancestors. Thus, we only memoize as long
689   // as there is a single parent.
690   bool memoizedMatchesAncestorOfRecursively(const DynTypedNode &Node,
691                                             ASTContext &Ctx,
692                                             const DynTypedMatcher &Matcher,
693                                             BoundNodesTreeBuilder *Builder,
694                                             AncestorMatchMode MatchMode) {
695     // For AST-nodes that don't have an identity, we can't memoize.
696     if (!Builder->isComparable())
697       return matchesAncestorOfRecursively(Node, Ctx, Matcher, Builder,
698                                           MatchMode);
699 
700     MatchKey Key;
701     Key.MatcherID = Matcher.getID();
702     Key.Node = Node;
703     Key.BoundNodes = *Builder;
704     Key.Traversal = Ctx.getParentMapContext().getTraversalKind();
705 
706     // Note that we cannot use insert and reuse the iterator, as recursive
707     // calls to match might invalidate the result cache iterators.
708     MemoizationMap::iterator I = ResultCache.find(Key);
709     if (I != ResultCache.end()) {
710       *Builder = I->second.Nodes;
711       return I->second.ResultOfMatch;
712     }
713 
714     MemoizedMatchResult Result;
715     Result.Nodes = *Builder;
716     Result.ResultOfMatch = matchesAncestorOfRecursively(
717         Node, Ctx, Matcher, &Result.Nodes, MatchMode);
718 
719     MemoizedMatchResult &CachedResult = ResultCache[Key];
720     CachedResult = std::move(Result);
721 
722     *Builder = CachedResult.Nodes;
723     return CachedResult.ResultOfMatch;
724   }
725 
726   bool matchesAncestorOfRecursively(const DynTypedNode &Node, ASTContext &Ctx,
727                                     const DynTypedMatcher &Matcher,
728                                     BoundNodesTreeBuilder *Builder,
729                                     AncestorMatchMode MatchMode) {
730     const auto &Parents = ActiveASTContext->getParents(Node);
731     if (Parents.empty()) {
732       // Nodes may have no parents if:
733       //  a) the node is the TranslationUnitDecl
734       //  b) we have a limited traversal scope that excludes the parent edges
735       //  c) there is a bug in the AST, and the node is not reachable
736       // Usually the traversal scope is the whole AST, which precludes b.
737       // Bugs are common enough that it's worthwhile asserting when we can.
738 #ifndef NDEBUG
739       if (!Node.get<TranslationUnitDecl>() &&
740           /* Traversal scope is full AST if any of the bounds are the TU */
741           llvm::any_of(ActiveASTContext->getTraversalScope(), [](Decl *D) {
742             return D->getKind() == Decl::TranslationUnit;
743           })) {
744         llvm::errs() << "Tried to match orphan node:\n";
745         Node.dump(llvm::errs(), ActiveASTContext->getSourceManager());
746         llvm_unreachable("Parent map should be complete!");
747       }
748 #endif
749       return false;
750     }
751     if (Parents.size() == 1) {
752       // Only one parent - do recursive memoization.
753       const DynTypedNode Parent = Parents[0];
754       BoundNodesTreeBuilder BuilderCopy = *Builder;
755       if (Matcher.matches(Parent, this, &BuilderCopy)) {
756         *Builder = std::move(BuilderCopy);
757         return true;
758       }
759       if (MatchMode != ASTMatchFinder::AMM_ParentOnly) {
760         return memoizedMatchesAncestorOfRecursively(Parent, Ctx, Matcher,
761                                                     Builder, MatchMode);
762         // Once we get back from the recursive call, the result will be the
763         // same as the parent's result.
764       }
765     } else {
766       // Multiple parents - BFS over the rest of the nodes.
767       llvm::DenseSet<const void *> Visited;
768       std::deque<DynTypedNode> Queue(Parents.begin(), Parents.end());
769       while (!Queue.empty()) {
770         BoundNodesTreeBuilder BuilderCopy = *Builder;
771         if (Matcher.matches(Queue.front(), this, &BuilderCopy)) {
772           *Builder = std::move(BuilderCopy);
773           return true;
774         }
775         if (MatchMode != ASTMatchFinder::AMM_ParentOnly) {
776           for (const auto &Parent :
777                ActiveASTContext->getParents(Queue.front())) {
778             // Make sure we do not visit the same node twice.
779             // Otherwise, we'll visit the common ancestors as often as there
780             // are splits on the way down.
781             if (Visited.insert(Parent.getMemoizationData()).second)
782               Queue.push_back(Parent);
783           }
784         }
785         Queue.pop_front();
786       }
787     }
788     return false;
789   }
790 
791   // Implements a BoundNodesTree::Visitor that calls a MatchCallback with
792   // the aggregated bound nodes for each match.
793   class MatchVisitor : public BoundNodesTreeBuilder::Visitor {
794   public:
795     MatchVisitor(ASTContext* Context,
796                  MatchFinder::MatchCallback* Callback)
797       : Context(Context),
798         Callback(Callback) {}
799 
800     void visitMatch(const BoundNodes& BoundNodesView) override {
801       Callback->run(MatchFinder::MatchResult(BoundNodesView, Context));
802     }
803 
804   private:
805     ASTContext* Context;
806     MatchFinder::MatchCallback* Callback;
807   };
808 
809   // Returns true if 'TypeNode' has an alias that matches the given matcher.
810   bool typeHasMatchingAlias(const Type *TypeNode,
811                             const Matcher<NamedDecl> &Matcher,
812                             BoundNodesTreeBuilder *Builder) {
813     const Type *const CanonicalType =
814       ActiveASTContext->getCanonicalType(TypeNode);
815     auto Aliases = TypeAliases.find(CanonicalType);
816     if (Aliases == TypeAliases.end())
817       return false;
818     for (const TypedefNameDecl *Alias : Aliases->second) {
819       BoundNodesTreeBuilder Result(*Builder);
820       if (Matcher.matches(*Alias, this, &Result)) {
821         *Builder = std::move(Result);
822         return true;
823       }
824     }
825     return false;
826   }
827 
828   bool
829   objcClassHasMatchingCompatibilityAlias(const ObjCInterfaceDecl *InterfaceDecl,
830                                          const Matcher<NamedDecl> &Matcher,
831                                          BoundNodesTreeBuilder *Builder) {
832     auto Aliases = CompatibleAliases.find(InterfaceDecl);
833     if (Aliases == CompatibleAliases.end())
834       return false;
835     for (const ObjCCompatibleAliasDecl *Alias : Aliases->second) {
836       BoundNodesTreeBuilder Result(*Builder);
837       if (Matcher.matches(*Alias, this, &Result)) {
838         *Builder = std::move(Result);
839         return true;
840       }
841     }
842     return false;
843   }
844 
845   /// Bucket to record map.
846   ///
847   /// Used to get the appropriate bucket for each matcher.
848   llvm::StringMap<llvm::TimeRecord> TimeByBucket;
849 
850   const MatchFinder::MatchersByType *Matchers;
851 
852   /// Filtered list of matcher indices for each matcher kind.
853   ///
854   /// \c Decl and \c Stmt toplevel matchers usually apply to a specific node
855   /// kind (and derived kinds) so it is a waste to try every matcher on every
856   /// node.
857   /// We precalculate a list of matchers that pass the toplevel restrict check.
858   llvm::DenseMap<ASTNodeKind, std::vector<unsigned short>> MatcherFiltersMap;
859 
860   const MatchFinder::MatchFinderOptions &Options;
861   ASTContext *ActiveASTContext;
862 
863   // Maps a canonical type to its TypedefDecls.
864   llvm::DenseMap<const Type*, std::set<const TypedefNameDecl*> > TypeAliases;
865 
866   // Maps an Objective-C interface to its ObjCCompatibleAliasDecls.
867   llvm::DenseMap<const ObjCInterfaceDecl *,
868                  llvm::SmallPtrSet<const ObjCCompatibleAliasDecl *, 2>>
869       CompatibleAliases;
870 
871   // Maps (matcher, node) -> the match result for memoization.
872   typedef std::map<MatchKey, MemoizedMatchResult> MemoizationMap;
873   MemoizationMap ResultCache;
874 };
875 
876 static CXXRecordDecl *
877 getAsCXXRecordDeclOrPrimaryTemplate(const Type *TypeNode) {
878   if (auto *RD = TypeNode->getAsCXXRecordDecl())
879     return RD;
880 
881   // Find the innermost TemplateSpecializationType that isn't an alias template.
882   auto *TemplateType = TypeNode->getAs<TemplateSpecializationType>();
883   while (TemplateType && TemplateType->isTypeAlias())
884     TemplateType =
885         TemplateType->getAliasedType()->getAs<TemplateSpecializationType>();
886 
887   // If this is the name of a (dependent) template specialization, use the
888   // definition of the template, even though it might be specialized later.
889   if (TemplateType)
890     if (auto *ClassTemplate = dyn_cast_or_null<ClassTemplateDecl>(
891           TemplateType->getTemplateName().getAsTemplateDecl()))
892       return ClassTemplate->getTemplatedDecl();
893 
894   return nullptr;
895 }
896 
897 // Returns true if the given C++ class is directly or indirectly derived
898 // from a base type with the given name.  A class is not considered to be
899 // derived from itself.
900 bool MatchASTVisitor::classIsDerivedFrom(const CXXRecordDecl *Declaration,
901                                          const Matcher<NamedDecl> &Base,
902                                          BoundNodesTreeBuilder *Builder,
903                                          bool Directly) {
904   if (!Declaration->hasDefinition())
905     return false;
906   for (const auto &It : Declaration->bases()) {
907     const Type *TypeNode = It.getType().getTypePtr();
908 
909     if (typeHasMatchingAlias(TypeNode, Base, Builder))
910       return true;
911 
912     // FIXME: Going to the primary template here isn't really correct, but
913     // unfortunately we accept a Decl matcher for the base class not a Type
914     // matcher, so it's the best thing we can do with our current interface.
915     CXXRecordDecl *ClassDecl = getAsCXXRecordDeclOrPrimaryTemplate(TypeNode);
916     if (!ClassDecl)
917       continue;
918     if (ClassDecl == Declaration) {
919       // This can happen for recursive template definitions; if the
920       // current declaration did not match, we can safely return false.
921       return false;
922     }
923     BoundNodesTreeBuilder Result(*Builder);
924     if (Base.matches(*ClassDecl, this, &Result)) {
925       *Builder = std::move(Result);
926       return true;
927     }
928     if (!Directly && classIsDerivedFrom(ClassDecl, Base, Builder, Directly))
929       return true;
930   }
931   return false;
932 }
933 
934 // Returns true if the given Objective-C class is directly or indirectly
935 // derived from a matching base class. A class is not considered to be derived
936 // from itself.
937 bool MatchASTVisitor::objcClassIsDerivedFrom(
938     const ObjCInterfaceDecl *Declaration, const Matcher<NamedDecl> &Base,
939     BoundNodesTreeBuilder *Builder, bool Directly) {
940   // Check if any of the superclasses of the class match.
941   for (const ObjCInterfaceDecl *ClassDecl = Declaration->getSuperClass();
942        ClassDecl != nullptr; ClassDecl = ClassDecl->getSuperClass()) {
943     // Check if there are any matching compatibility aliases.
944     if (objcClassHasMatchingCompatibilityAlias(ClassDecl, Base, Builder))
945       return true;
946 
947     // Check if there are any matching type aliases.
948     const Type *TypeNode = ClassDecl->getTypeForDecl();
949     if (typeHasMatchingAlias(TypeNode, Base, Builder))
950       return true;
951 
952     if (Base.matches(*ClassDecl, this, Builder))
953       return true;
954 
955     // Not `return false` as a temporary workaround for PR43879.
956     if (Directly)
957       break;
958   }
959 
960   return false;
961 }
962 
963 bool MatchASTVisitor::TraverseDecl(Decl *DeclNode) {
964   if (!DeclNode) {
965     return true;
966   }
967   match(*DeclNode);
968   return RecursiveASTVisitor<MatchASTVisitor>::TraverseDecl(DeclNode);
969 }
970 
971 bool MatchASTVisitor::TraverseStmt(Stmt *StmtNode, DataRecursionQueue *Queue) {
972   if (!StmtNode) {
973     return true;
974   }
975   match(*StmtNode);
976   return RecursiveASTVisitor<MatchASTVisitor>::TraverseStmt(StmtNode, Queue);
977 }
978 
979 bool MatchASTVisitor::TraverseType(QualType TypeNode) {
980   match(TypeNode);
981   return RecursiveASTVisitor<MatchASTVisitor>::TraverseType(TypeNode);
982 }
983 
984 bool MatchASTVisitor::TraverseTypeLoc(TypeLoc TypeLocNode) {
985   // The RecursiveASTVisitor only visits types if they're not within TypeLocs.
986   // We still want to find those types via matchers, so we match them here. Note
987   // that the TypeLocs are structurally a shadow-hierarchy to the expressed
988   // type, so we visit all involved parts of a compound type when matching on
989   // each TypeLoc.
990   match(TypeLocNode);
991   match(TypeLocNode.getType());
992   return RecursiveASTVisitor<MatchASTVisitor>::TraverseTypeLoc(TypeLocNode);
993 }
994 
995 bool MatchASTVisitor::TraverseNestedNameSpecifier(NestedNameSpecifier *NNS) {
996   match(*NNS);
997   return RecursiveASTVisitor<MatchASTVisitor>::TraverseNestedNameSpecifier(NNS);
998 }
999 
1000 bool MatchASTVisitor::TraverseNestedNameSpecifierLoc(
1001     NestedNameSpecifierLoc NNS) {
1002   if (!NNS)
1003     return true;
1004 
1005   match(NNS);
1006 
1007   // We only match the nested name specifier here (as opposed to traversing it)
1008   // because the traversal is already done in the parallel "Loc"-hierarchy.
1009   if (NNS.hasQualifier())
1010     match(*NNS.getNestedNameSpecifier());
1011   return
1012       RecursiveASTVisitor<MatchASTVisitor>::TraverseNestedNameSpecifierLoc(NNS);
1013 }
1014 
1015 bool MatchASTVisitor::TraverseConstructorInitializer(
1016     CXXCtorInitializer *CtorInit) {
1017   if (!CtorInit)
1018     return true;
1019 
1020   match(*CtorInit);
1021 
1022   return RecursiveASTVisitor<MatchASTVisitor>::TraverseConstructorInitializer(
1023       CtorInit);
1024 }
1025 
1026 class MatchASTConsumer : public ASTConsumer {
1027 public:
1028   MatchASTConsumer(MatchFinder *Finder,
1029                    MatchFinder::ParsingDoneTestCallback *ParsingDone)
1030       : Finder(Finder), ParsingDone(ParsingDone) {}
1031 
1032 private:
1033   void HandleTranslationUnit(ASTContext &Context) override {
1034     if (ParsingDone != nullptr) {
1035       ParsingDone->run();
1036     }
1037     Finder->matchAST(Context);
1038   }
1039 
1040   MatchFinder *Finder;
1041   MatchFinder::ParsingDoneTestCallback *ParsingDone;
1042 };
1043 
1044 } // end namespace
1045 } // end namespace internal
1046 
1047 MatchFinder::MatchResult::MatchResult(const BoundNodes &Nodes,
1048                                       ASTContext *Context)
1049   : Nodes(Nodes), Context(Context),
1050     SourceManager(&Context->getSourceManager()) {}
1051 
1052 MatchFinder::MatchCallback::~MatchCallback() {}
1053 MatchFinder::ParsingDoneTestCallback::~ParsingDoneTestCallback() {}
1054 
1055 MatchFinder::MatchFinder(MatchFinderOptions Options)
1056     : Options(std::move(Options)), ParsingDone(nullptr) {}
1057 
1058 MatchFinder::~MatchFinder() {}
1059 
1060 void MatchFinder::addMatcher(const DeclarationMatcher &NodeMatch,
1061                              MatchCallback *Action) {
1062   Matchers.DeclOrStmt.emplace_back(NodeMatch, Action);
1063   Matchers.AllCallbacks.insert(Action);
1064 }
1065 
1066 void MatchFinder::addMatcher(const TypeMatcher &NodeMatch,
1067                              MatchCallback *Action) {
1068   Matchers.Type.emplace_back(NodeMatch, Action);
1069   Matchers.AllCallbacks.insert(Action);
1070 }
1071 
1072 void MatchFinder::addMatcher(const StatementMatcher &NodeMatch,
1073                              MatchCallback *Action) {
1074   Matchers.DeclOrStmt.emplace_back(NodeMatch, Action);
1075   Matchers.AllCallbacks.insert(Action);
1076 }
1077 
1078 void MatchFinder::addMatcher(const NestedNameSpecifierMatcher &NodeMatch,
1079                              MatchCallback *Action) {
1080   Matchers.NestedNameSpecifier.emplace_back(NodeMatch, Action);
1081   Matchers.AllCallbacks.insert(Action);
1082 }
1083 
1084 void MatchFinder::addMatcher(const NestedNameSpecifierLocMatcher &NodeMatch,
1085                              MatchCallback *Action) {
1086   Matchers.NestedNameSpecifierLoc.emplace_back(NodeMatch, Action);
1087   Matchers.AllCallbacks.insert(Action);
1088 }
1089 
1090 void MatchFinder::addMatcher(const TypeLocMatcher &NodeMatch,
1091                              MatchCallback *Action) {
1092   Matchers.TypeLoc.emplace_back(NodeMatch, Action);
1093   Matchers.AllCallbacks.insert(Action);
1094 }
1095 
1096 void MatchFinder::addMatcher(const CXXCtorInitializerMatcher &NodeMatch,
1097                              MatchCallback *Action) {
1098   Matchers.CtorInit.emplace_back(NodeMatch, Action);
1099   Matchers.AllCallbacks.insert(Action);
1100 }
1101 
1102 bool MatchFinder::addDynamicMatcher(const internal::DynTypedMatcher &NodeMatch,
1103                                     MatchCallback *Action) {
1104   if (NodeMatch.canConvertTo<Decl>()) {
1105     addMatcher(NodeMatch.convertTo<Decl>(), Action);
1106     return true;
1107   } else if (NodeMatch.canConvertTo<QualType>()) {
1108     addMatcher(NodeMatch.convertTo<QualType>(), Action);
1109     return true;
1110   } else if (NodeMatch.canConvertTo<Stmt>()) {
1111     addMatcher(NodeMatch.convertTo<Stmt>(), Action);
1112     return true;
1113   } else if (NodeMatch.canConvertTo<NestedNameSpecifier>()) {
1114     addMatcher(NodeMatch.convertTo<NestedNameSpecifier>(), Action);
1115     return true;
1116   } else if (NodeMatch.canConvertTo<NestedNameSpecifierLoc>()) {
1117     addMatcher(NodeMatch.convertTo<NestedNameSpecifierLoc>(), Action);
1118     return true;
1119   } else if (NodeMatch.canConvertTo<TypeLoc>()) {
1120     addMatcher(NodeMatch.convertTo<TypeLoc>(), Action);
1121     return true;
1122   } else if (NodeMatch.canConvertTo<CXXCtorInitializer>()) {
1123     addMatcher(NodeMatch.convertTo<CXXCtorInitializer>(), Action);
1124     return true;
1125   }
1126   return false;
1127 }
1128 
1129 std::unique_ptr<ASTConsumer> MatchFinder::newASTConsumer() {
1130   return std::make_unique<internal::MatchASTConsumer>(this, ParsingDone);
1131 }
1132 
1133 void MatchFinder::match(const clang::DynTypedNode &Node, ASTContext &Context) {
1134   internal::MatchASTVisitor Visitor(&Matchers, Options);
1135   Visitor.set_active_ast_context(&Context);
1136   Visitor.match(Node);
1137 }
1138 
1139 void MatchFinder::matchAST(ASTContext &Context) {
1140   internal::MatchASTVisitor Visitor(&Matchers, Options);
1141   Visitor.set_active_ast_context(&Context);
1142   Visitor.onStartOfTranslationUnit();
1143   Visitor.TraverseAST(Context);
1144   Visitor.onEndOfTranslationUnit();
1145 }
1146 
1147 void MatchFinder::registerTestCallbackAfterParsing(
1148     MatchFinder::ParsingDoneTestCallback *NewParsingDone) {
1149   ParsingDone = NewParsingDone;
1150 }
1151 
1152 StringRef MatchFinder::MatchCallback::getID() const { return "<unknown>"; }
1153 
1154 } // end namespace ast_matchers
1155 } // end namespace clang
1156