xref: /freebsd/contrib/llvm-project/clang/lib/Tooling/Transformer/Stencil.cpp (revision 5e801ac66d24704442eba426ed13c3effb8a34e7)
1 //===--- Stencil.cpp - Stencil implementation -------------------*- C++ -*-===//
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 #include "clang/Tooling/Transformer/Stencil.h"
10 #include "clang/AST/ASTContext.h"
11 #include "clang/AST/ASTTypeTraits.h"
12 #include "clang/AST/Expr.h"
13 #include "clang/ASTMatchers/ASTMatchFinder.h"
14 #include "clang/ASTMatchers/ASTMatchers.h"
15 #include "clang/Basic/SourceLocation.h"
16 #include "clang/Lex/Lexer.h"
17 #include "clang/Tooling/Transformer/SourceCode.h"
18 #include "clang/Tooling/Transformer/SourceCodeBuilders.h"
19 #include "llvm/ADT/SmallVector.h"
20 #include "llvm/ADT/Twine.h"
21 #include "llvm/Support/Errc.h"
22 #include "llvm/Support/Error.h"
23 #include <atomic>
24 #include <memory>
25 #include <string>
26 
27 using namespace clang;
28 using namespace transformer;
29 
30 using ast_matchers::BoundNodes;
31 using ast_matchers::MatchFinder;
32 using llvm::errc;
33 using llvm::Error;
34 using llvm::Expected;
35 using llvm::StringError;
36 
37 static llvm::Expected<DynTypedNode> getNode(const BoundNodes &Nodes,
38                                             StringRef Id) {
39   auto &NodesMap = Nodes.getMap();
40   auto It = NodesMap.find(Id);
41   if (It == NodesMap.end())
42     return llvm::make_error<llvm::StringError>(llvm::errc::invalid_argument,
43                                                "Id not bound: " + Id);
44   return It->second;
45 }
46 
47 static Error printNode(StringRef Id, const MatchFinder::MatchResult &Match,
48                        std::string *Result) {
49   std::string Output;
50   llvm::raw_string_ostream Os(Output);
51   auto NodeOrErr = getNode(Match.Nodes, Id);
52   if (auto Err = NodeOrErr.takeError())
53     return Err;
54   NodeOrErr->print(Os, PrintingPolicy(Match.Context->getLangOpts()));
55   *Result += Os.str();
56   return Error::success();
57 }
58 
59 // FIXME: Consider memoizing this function using the `ASTContext`.
60 static bool isSmartPointerType(QualType Ty, ASTContext &Context) {
61   using namespace ::clang::ast_matchers;
62 
63   // Optimization: hard-code common smart-pointer types. This can/should be
64   // removed if we start caching the results of this function.
65   auto KnownSmartPointer =
66       cxxRecordDecl(hasAnyName("::std::unique_ptr", "::std::shared_ptr"));
67   const auto QuacksLikeASmartPointer = cxxRecordDecl(
68       hasMethod(cxxMethodDecl(hasOverloadedOperatorName("->"),
69                               returns(qualType(pointsTo(type()))))),
70       hasMethod(cxxMethodDecl(hasOverloadedOperatorName("*"),
71                               returns(qualType(references(type()))))));
72   const auto SmartPointer = qualType(hasDeclaration(
73       cxxRecordDecl(anyOf(KnownSmartPointer, QuacksLikeASmartPointer))));
74   return match(SmartPointer, Ty, Context).size() > 0;
75 }
76 
77 // Identifies use of `operator*` on smart pointers, and returns the underlying
78 // smart-pointer expression; otherwise, returns null.
79 static const Expr *isSmartDereference(const Expr &E, ASTContext &Context) {
80   using namespace ::clang::ast_matchers;
81 
82   const auto HasOverloadedArrow = cxxRecordDecl(hasMethod(cxxMethodDecl(
83       hasOverloadedOperatorName("->"), returns(qualType(pointsTo(type()))))));
84   // Verify it is a smart pointer by finding `operator->` in the class
85   // declaration.
86   auto Deref = cxxOperatorCallExpr(
87       hasOverloadedOperatorName("*"), hasUnaryOperand(expr().bind("arg")),
88       callee(cxxMethodDecl(ofClass(HasOverloadedArrow))));
89   return selectFirst<Expr>("arg", match(Deref, E, Context));
90 }
91 
92 namespace {
93 // An arbitrary fragment of code within a stencil.
94 class RawTextStencil : public StencilInterface {
95   std::string Text;
96 
97 public:
98   explicit RawTextStencil(std::string T) : Text(std::move(T)) {}
99 
100   std::string toString() const override {
101     std::string Result;
102     llvm::raw_string_ostream OS(Result);
103     OS << "\"";
104     OS.write_escaped(Text);
105     OS << "\"";
106     OS.flush();
107     return Result;
108   }
109 
110   Error eval(const MatchFinder::MatchResult &Match,
111              std::string *Result) const override {
112     Result->append(Text);
113     return Error::success();
114   }
115 };
116 
117 // A debugging operation to dump the AST for a particular (bound) AST node.
118 class DebugPrintNodeStencil : public StencilInterface {
119   std::string Id;
120 
121 public:
122   explicit DebugPrintNodeStencil(std::string S) : Id(std::move(S)) {}
123 
124   std::string toString() const override {
125     return (llvm::Twine("dPrint(\"") + Id + "\")").str();
126   }
127 
128   Error eval(const MatchFinder::MatchResult &Match,
129              std::string *Result) const override {
130     return printNode(Id, Match, Result);
131   }
132 };
133 
134 // Operators that take a single node Id as an argument.
135 enum class UnaryNodeOperator {
136   Parens,
137   Deref,
138   MaybeDeref,
139   AddressOf,
140   MaybeAddressOf,
141   Describe,
142 };
143 
144 // Generic container for stencil operations with a (single) node-id argument.
145 class UnaryOperationStencil : public StencilInterface {
146   UnaryNodeOperator Op;
147   std::string Id;
148 
149 public:
150   UnaryOperationStencil(UnaryNodeOperator Op, std::string Id)
151       : Op(Op), Id(std::move(Id)) {}
152 
153   std::string toString() const override {
154     StringRef OpName;
155     switch (Op) {
156     case UnaryNodeOperator::Parens:
157       OpName = "expression";
158       break;
159     case UnaryNodeOperator::Deref:
160       OpName = "deref";
161       break;
162     case UnaryNodeOperator::MaybeDeref:
163       OpName = "maybeDeref";
164       break;
165     case UnaryNodeOperator::AddressOf:
166       OpName = "addressOf";
167       break;
168     case UnaryNodeOperator::MaybeAddressOf:
169       OpName = "maybeAddressOf";
170       break;
171     case UnaryNodeOperator::Describe:
172       OpName = "describe";
173       break;
174     }
175     return (OpName + "(\"" + Id + "\")").str();
176   }
177 
178   Error eval(const MatchFinder::MatchResult &Match,
179              std::string *Result) const override {
180     // The `Describe` operation can be applied to any node, not just
181     // expressions, so it is handled here, separately.
182     if (Op == UnaryNodeOperator::Describe)
183       return printNode(Id, Match, Result);
184 
185     const auto *E = Match.Nodes.getNodeAs<Expr>(Id);
186     if (E == nullptr)
187       return llvm::make_error<StringError>(errc::invalid_argument,
188                                            "Id not bound or not Expr: " + Id);
189     llvm::Optional<std::string> Source;
190     switch (Op) {
191     case UnaryNodeOperator::Parens:
192       Source = tooling::buildParens(*E, *Match.Context);
193       break;
194     case UnaryNodeOperator::Deref:
195       Source = tooling::buildDereference(*E, *Match.Context);
196       break;
197     case UnaryNodeOperator::MaybeDeref:
198       if (E->getType()->isAnyPointerType() ||
199           isSmartPointerType(E->getType(), *Match.Context)) {
200         // Strip off any operator->. This can only occur inside an actual arrow
201         // member access, so we treat it as equivalent to an actual object
202         // expression.
203         if (const auto *OpCall = dyn_cast<clang::CXXOperatorCallExpr>(E)) {
204           if (OpCall->getOperator() == clang::OO_Arrow &&
205               OpCall->getNumArgs() == 1) {
206             E = OpCall->getArg(0);
207           }
208         }
209         Source = tooling::buildDereference(*E, *Match.Context);
210         break;
211       }
212       *Result += tooling::getText(*E, *Match.Context);
213       return Error::success();
214     case UnaryNodeOperator::AddressOf:
215       Source = tooling::buildAddressOf(*E, *Match.Context);
216       break;
217     case UnaryNodeOperator::MaybeAddressOf:
218       if (E->getType()->isAnyPointerType() ||
219           isSmartPointerType(E->getType(), *Match.Context)) {
220         // Strip off any operator->. This can only occur inside an actual arrow
221         // member access, so we treat it as equivalent to an actual object
222         // expression.
223         if (const auto *OpCall = dyn_cast<clang::CXXOperatorCallExpr>(E)) {
224           if (OpCall->getOperator() == clang::OO_Arrow &&
225               OpCall->getNumArgs() == 1) {
226             E = OpCall->getArg(0);
227           }
228         }
229         *Result += tooling::getText(*E, *Match.Context);
230         return Error::success();
231       }
232       Source = tooling::buildAddressOf(*E, *Match.Context);
233       break;
234     case UnaryNodeOperator::Describe:
235       llvm_unreachable("This case is handled at the start of the function");
236     }
237     if (!Source)
238       return llvm::make_error<StringError>(
239           errc::invalid_argument,
240           "Could not construct expression source from ID: " + Id);
241     *Result += *Source;
242     return Error::success();
243   }
244 };
245 
246 // The fragment of code corresponding to the selected range.
247 class SelectorStencil : public StencilInterface {
248   RangeSelector Selector;
249 
250 public:
251   explicit SelectorStencil(RangeSelector S) : Selector(std::move(S)) {}
252 
253   std::string toString() const override { return "selection(...)"; }
254 
255   Error eval(const MatchFinder::MatchResult &Match,
256              std::string *Result) const override {
257     auto RawRange = Selector(Match);
258     if (!RawRange)
259       return RawRange.takeError();
260     CharSourceRange Range = Lexer::makeFileCharRange(
261         *RawRange, *Match.SourceManager, Match.Context->getLangOpts());
262     if (Range.isInvalid()) {
263       // Validate the original range to attempt to get a meaningful error
264       // message. If it's valid, then something else is the cause and we just
265       // return the generic failure message.
266       if (auto Err =
267               tooling::validateEditRange(*RawRange, *Match.SourceManager))
268         return handleErrors(std::move(Err), [](std::unique_ptr<StringError> E) {
269           assert(E->convertToErrorCode() ==
270                      llvm::make_error_code(errc::invalid_argument) &&
271                  "Validation errors must carry the invalid_argument code");
272           return llvm::createStringError(
273               errc::invalid_argument,
274               "selected range could not be resolved to a valid source range; " +
275                   E->getMessage());
276         });
277       return llvm::createStringError(
278           errc::invalid_argument,
279           "selected range could not be resolved to a valid source range");
280     }
281     // Validate `Range`, because `makeFileCharRange` accepts some ranges that
282     // `validateEditRange` rejects.
283     if (auto Err = tooling::validateEditRange(Range, *Match.SourceManager))
284       return joinErrors(
285           llvm::createStringError(errc::invalid_argument,
286                                   "selected range is not valid for editing"),
287           std::move(Err));
288     *Result += tooling::getText(Range, *Match.Context);
289     return Error::success();
290   }
291 };
292 
293 // A stencil operation to build a member access `e.m` or `e->m`, as appropriate.
294 class AccessStencil : public StencilInterface {
295   std::string BaseId;
296   Stencil Member;
297 
298 public:
299   AccessStencil(StringRef BaseId, Stencil Member)
300       : BaseId(std::string(BaseId)), Member(std::move(Member)) {}
301 
302   std::string toString() const override {
303     return (llvm::Twine("access(\"") + BaseId + "\", " + Member->toString() +
304             ")")
305         .str();
306   }
307 
308   Error eval(const MatchFinder::MatchResult &Match,
309              std::string *Result) const override {
310     const auto *E = Match.Nodes.getNodeAs<Expr>(BaseId);
311     if (E == nullptr)
312       return llvm::make_error<StringError>(errc::invalid_argument,
313                                            "Id not bound: " + BaseId);
314     if (!E->isImplicitCXXThis()) {
315       llvm::Optional<std::string> S;
316       if (E->getType()->isAnyPointerType() ||
317           isSmartPointerType(E->getType(), *Match.Context)) {
318         // Strip off any operator->. This can only occur inside an actual arrow
319         // member access, so we treat it as equivalent to an actual object
320         // expression.
321         if (const auto *OpCall = dyn_cast<clang::CXXOperatorCallExpr>(E)) {
322           if (OpCall->getOperator() == clang::OO_Arrow &&
323               OpCall->getNumArgs() == 1) {
324             E = OpCall->getArg(0);
325           }
326         }
327         S = tooling::buildArrow(*E, *Match.Context);
328       } else if (const auto *Operand = isSmartDereference(*E, *Match.Context)) {
329         // `buildDot` already handles the built-in dereference operator, so we
330         // only need to catch overloaded `operator*`.
331         S = tooling::buildArrow(*Operand, *Match.Context);
332       } else {
333         S = tooling::buildDot(*E, *Match.Context);
334       }
335       if (S.hasValue())
336         *Result += *S;
337       else
338         return llvm::make_error<StringError>(
339             errc::invalid_argument,
340             "Could not construct object text from ID: " + BaseId);
341     }
342     return Member->eval(Match, Result);
343   }
344 };
345 
346 class IfBoundStencil : public StencilInterface {
347   std::string Id;
348   Stencil TrueStencil;
349   Stencil FalseStencil;
350 
351 public:
352   IfBoundStencil(StringRef Id, Stencil TrueStencil, Stencil FalseStencil)
353       : Id(std::string(Id)), TrueStencil(std::move(TrueStencil)),
354         FalseStencil(std::move(FalseStencil)) {}
355 
356   std::string toString() const override {
357     return (llvm::Twine("ifBound(\"") + Id + "\", " + TrueStencil->toString() +
358             ", " + FalseStencil->toString() + ")")
359         .str();
360   }
361 
362   Error eval(const MatchFinder::MatchResult &Match,
363              std::string *Result) const override {
364     auto &M = Match.Nodes.getMap();
365     return (M.find(Id) != M.end() ? TrueStencil : FalseStencil)
366         ->eval(Match, Result);
367   }
368 };
369 
370 class SelectBoundStencil : public clang::transformer::StencilInterface {
371   static bool containsNoNullStencils(
372       const std::vector<std::pair<std::string, Stencil>> &Cases) {
373     for (const auto &S : Cases)
374       if (S.second == nullptr)
375         return false;
376     return true;
377   }
378 
379 public:
380   SelectBoundStencil(std::vector<std::pair<std::string, Stencil>> Cases,
381                      Stencil Default)
382       : CaseStencils(std::move(Cases)), DefaultStencil(std::move(Default)) {
383     assert(containsNoNullStencils(CaseStencils) &&
384            "cases of selectBound may not be null");
385   }
386   ~SelectBoundStencil() override{};
387 
388   llvm::Error eval(const MatchFinder::MatchResult &match,
389                    std::string *result) const override {
390     const BoundNodes::IDToNodeMap &NodeMap = match.Nodes.getMap();
391     for (const auto &S : CaseStencils) {
392       if (NodeMap.count(S.first) > 0) {
393         return S.second->eval(match, result);
394       }
395     }
396 
397     if (DefaultStencil != nullptr) {
398       return DefaultStencil->eval(match, result);
399     }
400 
401     llvm::SmallVector<llvm::StringRef, 2> CaseIDs;
402     CaseIDs.reserve(CaseStencils.size());
403     for (const auto &S : CaseStencils)
404       CaseIDs.emplace_back(S.first);
405 
406     return llvm::createStringError(
407         errc::result_out_of_range,
408         llvm::Twine("selectBound failed: no cases bound and no default: {") +
409             llvm::join(CaseIDs, ", ") + "}");
410   }
411 
412   std::string toString() const override {
413     std::string Buffer;
414     llvm::raw_string_ostream Stream(Buffer);
415     Stream << "selectBound({";
416     bool First = true;
417     for (const auto &S : CaseStencils) {
418       if (First)
419         First = false;
420       else
421         Stream << "}, ";
422       Stream << "{\"" << S.first << "\", " << S.second->toString();
423     }
424     Stream << "}}";
425     if (DefaultStencil != nullptr) {
426       Stream << ", " << DefaultStencil->toString();
427     }
428     Stream << ")";
429     return Stream.str();
430   }
431 
432 private:
433   std::vector<std::pair<std::string, Stencil>> CaseStencils;
434   Stencil DefaultStencil;
435 };
436 
437 class SequenceStencil : public StencilInterface {
438   std::vector<Stencil> Stencils;
439 
440 public:
441   SequenceStencil(std::vector<Stencil> Stencils)
442       : Stencils(std::move(Stencils)) {}
443 
444   std::string toString() const override {
445     llvm::SmallVector<std::string, 2> Parts;
446     Parts.reserve(Stencils.size());
447     for (const auto &S : Stencils)
448       Parts.push_back(S->toString());
449     return (llvm::Twine("seq(") + llvm::join(Parts, ", ") + ")").str();
450   }
451 
452   Error eval(const MatchFinder::MatchResult &Match,
453              std::string *Result) const override {
454     for (const auto &S : Stencils)
455       if (auto Err = S->eval(Match, Result))
456         return Err;
457     return Error::success();
458   }
459 };
460 
461 class RunStencil : public StencilInterface {
462   MatchConsumer<std::string> Consumer;
463 
464 public:
465   explicit RunStencil(MatchConsumer<std::string> C) : Consumer(std::move(C)) {}
466 
467   std::string toString() const override { return "run(...)"; }
468 
469   Error eval(const MatchFinder::MatchResult &Match,
470              std::string *Result) const override {
471 
472     Expected<std::string> Value = Consumer(Match);
473     if (!Value)
474       return Value.takeError();
475     *Result += *Value;
476     return Error::success();
477   }
478 };
479 } // namespace
480 
481 Stencil transformer::detail::makeStencil(StringRef Text) {
482   return std::make_shared<RawTextStencil>(std::string(Text));
483 }
484 
485 Stencil transformer::detail::makeStencil(RangeSelector Selector) {
486   return std::make_shared<SelectorStencil>(std::move(Selector));
487 }
488 
489 Stencil transformer::dPrint(StringRef Id) {
490   return std::make_shared<DebugPrintNodeStencil>(std::string(Id));
491 }
492 
493 Stencil transformer::expression(llvm::StringRef Id) {
494   return std::make_shared<UnaryOperationStencil>(UnaryNodeOperator::Parens,
495                                                  std::string(Id));
496 }
497 
498 Stencil transformer::deref(llvm::StringRef ExprId) {
499   return std::make_shared<UnaryOperationStencil>(UnaryNodeOperator::Deref,
500                                                  std::string(ExprId));
501 }
502 
503 Stencil transformer::maybeDeref(llvm::StringRef ExprId) {
504   return std::make_shared<UnaryOperationStencil>(UnaryNodeOperator::MaybeDeref,
505                                                  std::string(ExprId));
506 }
507 
508 Stencil transformer::addressOf(llvm::StringRef ExprId) {
509   return std::make_shared<UnaryOperationStencil>(UnaryNodeOperator::AddressOf,
510                                                  std::string(ExprId));
511 }
512 
513 Stencil transformer::maybeAddressOf(llvm::StringRef ExprId) {
514   return std::make_shared<UnaryOperationStencil>(
515       UnaryNodeOperator::MaybeAddressOf, std::string(ExprId));
516 }
517 
518 Stencil transformer::describe(StringRef Id) {
519   return std::make_shared<UnaryOperationStencil>(UnaryNodeOperator::Describe,
520                                                  std::string(Id));
521 }
522 
523 Stencil transformer::access(StringRef BaseId, Stencil Member) {
524   return std::make_shared<AccessStencil>(BaseId, std::move(Member));
525 }
526 
527 Stencil transformer::ifBound(StringRef Id, Stencil TrueStencil,
528                              Stencil FalseStencil) {
529   return std::make_shared<IfBoundStencil>(Id, std::move(TrueStencil),
530                                           std::move(FalseStencil));
531 }
532 
533 Stencil transformer::selectBound(
534     std::vector<std::pair<std::string, Stencil>> CaseStencils,
535     Stencil DefaultStencil) {
536   return std::make_shared<SelectBoundStencil>(std::move(CaseStencils),
537                                               std::move(DefaultStencil));
538 }
539 
540 Stencil transformer::run(MatchConsumer<std::string> Fn) {
541   return std::make_shared<RunStencil>(std::move(Fn));
542 }
543 
544 Stencil transformer::catVector(std::vector<Stencil> Parts) {
545   // Only one argument, so don't wrap in sequence.
546   if (Parts.size() == 1)
547     return std::move(Parts[0]);
548   return std::make_shared<SequenceStencil>(std::move(Parts));
549 }
550