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::MatchFinder;
31 using llvm::errc;
32 using llvm::Error;
33 using llvm::Expected;
34 using llvm::StringError;
35
36 static llvm::Expected<DynTypedNode>
getNode(const ast_matchers::BoundNodes & Nodes,StringRef Id)37 getNode(const ast_matchers::BoundNodes &Nodes, StringRef Id) {
38 auto &NodesMap = Nodes.getMap();
39 auto It = NodesMap.find(Id);
40 if (It == NodesMap.end())
41 return llvm::make_error<llvm::StringError>(llvm::errc::invalid_argument,
42 "Id not bound: " + Id);
43 return It->second;
44 }
45
46 namespace {
47 // An arbitrary fragment of code within a stencil.
48 struct RawTextData {
RawTextData__anon50cfd1840111::RawTextData49 explicit RawTextData(std::string T) : Text(std::move(T)) {}
50 std::string Text;
51 };
52
53 // A debugging operation to dump the AST for a particular (bound) AST node.
54 struct DebugPrintNodeData {
DebugPrintNodeData__anon50cfd1840111::DebugPrintNodeData55 explicit DebugPrintNodeData(std::string S) : Id(std::move(S)) {}
56 std::string Id;
57 };
58
59 // Operators that take a single node Id as an argument.
60 enum class UnaryNodeOperator {
61 Parens,
62 Deref,
63 MaybeDeref,
64 AddressOf,
65 MaybeAddressOf,
66 Describe,
67 };
68
69 // Generic container for stencil operations with a (single) node-id argument.
70 struct UnaryOperationData {
UnaryOperationData__anon50cfd1840111::UnaryOperationData71 UnaryOperationData(UnaryNodeOperator Op, std::string Id)
72 : Op(Op), Id(std::move(Id)) {}
73 UnaryNodeOperator Op;
74 std::string Id;
75 };
76
77 // The fragment of code corresponding to the selected range.
78 struct SelectorData {
SelectorData__anon50cfd1840111::SelectorData79 explicit SelectorData(RangeSelector S) : Selector(std::move(S)) {}
80 RangeSelector Selector;
81 };
82
83 // A stencil operation to build a member access `e.m` or `e->m`, as appropriate.
84 struct AccessData {
AccessData__anon50cfd1840111::AccessData85 AccessData(StringRef BaseId, Stencil Member)
86 : BaseId(std::string(BaseId)), Member(std::move(Member)) {}
87 std::string BaseId;
88 Stencil Member;
89 };
90
91 struct IfBoundData {
IfBoundData__anon50cfd1840111::IfBoundData92 IfBoundData(StringRef Id, Stencil TrueStencil, Stencil FalseStencil)
93 : Id(std::string(Id)), TrueStencil(std::move(TrueStencil)),
94 FalseStencil(std::move(FalseStencil)) {}
95 std::string Id;
96 Stencil TrueStencil;
97 Stencil FalseStencil;
98 };
99
100 struct SequenceData {
SequenceData__anon50cfd1840111::SequenceData101 SequenceData(std::vector<Stencil> Stencils) : Stencils(std::move(Stencils)) {}
102 std::vector<Stencil> Stencils;
103 };
104
toStringData(const RawTextData & Data)105 std::string toStringData(const RawTextData &Data) {
106 std::string Result;
107 llvm::raw_string_ostream OS(Result);
108 OS << "\"";
109 OS.write_escaped(Data.Text);
110 OS << "\"";
111 OS.flush();
112 return Result;
113 }
114
toStringData(const DebugPrintNodeData & Data)115 std::string toStringData(const DebugPrintNodeData &Data) {
116 return (llvm::Twine("dPrint(\"") + Data.Id + "\")").str();
117 }
118
toStringData(const UnaryOperationData & Data)119 std::string toStringData(const UnaryOperationData &Data) {
120 StringRef OpName;
121 switch (Data.Op) {
122 case UnaryNodeOperator::Parens:
123 OpName = "expression";
124 break;
125 case UnaryNodeOperator::Deref:
126 OpName = "deref";
127 break;
128 case UnaryNodeOperator::MaybeDeref:
129 OpName = "maybeDeref";
130 break;
131 case UnaryNodeOperator::AddressOf:
132 OpName = "addressOf";
133 break;
134 case UnaryNodeOperator::MaybeAddressOf:
135 OpName = "maybeAddressOf";
136 break;
137 case UnaryNodeOperator::Describe:
138 OpName = "describe";
139 break;
140 }
141 return (OpName + "(\"" + Data.Id + "\")").str();
142 }
143
toStringData(const SelectorData &)144 std::string toStringData(const SelectorData &) { return "selection(...)"; }
145
toStringData(const AccessData & Data)146 std::string toStringData(const AccessData &Data) {
147 return (llvm::Twine("access(\"") + Data.BaseId + "\", " +
148 Data.Member->toString() + ")")
149 .str();
150 }
151
toStringData(const IfBoundData & Data)152 std::string toStringData(const IfBoundData &Data) {
153 return (llvm::Twine("ifBound(\"") + Data.Id + "\", " +
154 Data.TrueStencil->toString() + ", " + Data.FalseStencil->toString() +
155 ")")
156 .str();
157 }
158
toStringData(const MatchConsumer<std::string> &)159 std::string toStringData(const MatchConsumer<std::string> &) {
160 return "run(...)";
161 }
162
toStringData(const SequenceData & Data)163 std::string toStringData(const SequenceData &Data) {
164 llvm::SmallVector<std::string, 2> Parts;
165 Parts.reserve(Data.Stencils.size());
166 for (const auto &S : Data.Stencils)
167 Parts.push_back(S->toString());
168 return (llvm::Twine("seq(") + llvm::join(Parts, ", ") + ")").str();
169 }
170
171 // The `evalData()` overloads evaluate the given stencil data to a string, given
172 // the match result, and append it to `Result`. We define an overload for each
173 // type of stencil data.
174
evalData(const RawTextData & Data,const MatchFinder::MatchResult &,std::string * Result)175 Error evalData(const RawTextData &Data, const MatchFinder::MatchResult &,
176 std::string *Result) {
177 Result->append(Data.Text);
178 return Error::success();
179 }
180
printNode(StringRef Id,const MatchFinder::MatchResult & Match,std::string * Result)181 static Error printNode(StringRef Id, const MatchFinder::MatchResult &Match,
182 std::string *Result) {
183 std::string Output;
184 llvm::raw_string_ostream Os(Output);
185 auto NodeOrErr = getNode(Match.Nodes, Id);
186 if (auto Err = NodeOrErr.takeError())
187 return Err;
188 NodeOrErr->print(Os, PrintingPolicy(Match.Context->getLangOpts()));
189 *Result += Os.str();
190 return Error::success();
191 }
192
evalData(const DebugPrintNodeData & Data,const MatchFinder::MatchResult & Match,std::string * Result)193 Error evalData(const DebugPrintNodeData &Data,
194 const MatchFinder::MatchResult &Match, std::string *Result) {
195 return printNode(Data.Id, Match, Result);
196 }
197
198 // FIXME: Consider memoizing this function using the `ASTContext`.
isSmartPointerType(QualType Ty,ASTContext & Context)199 static bool isSmartPointerType(QualType Ty, ASTContext &Context) {
200 using namespace ::clang::ast_matchers;
201
202 // Optimization: hard-code common smart-pointer types. This can/should be
203 // removed if we start caching the results of this function.
204 auto KnownSmartPointer =
205 cxxRecordDecl(hasAnyName("::std::unique_ptr", "::std::shared_ptr"));
206 const auto QuacksLikeASmartPointer = cxxRecordDecl(
207 hasMethod(cxxMethodDecl(hasOverloadedOperatorName("->"),
208 returns(qualType(pointsTo(type()))))),
209 hasMethod(cxxMethodDecl(hasOverloadedOperatorName("*"),
210 returns(qualType(references(type()))))));
211 const auto SmartPointer = qualType(hasDeclaration(
212 cxxRecordDecl(anyOf(KnownSmartPointer, QuacksLikeASmartPointer))));
213 return match(SmartPointer, Ty, Context).size() > 0;
214 }
215
evalData(const UnaryOperationData & Data,const MatchFinder::MatchResult & Match,std::string * Result)216 Error evalData(const UnaryOperationData &Data,
217 const MatchFinder::MatchResult &Match, std::string *Result) {
218 // The `Describe` operation can be applied to any node, not just expressions,
219 // so it is handled here, separately.
220 if (Data.Op == UnaryNodeOperator::Describe)
221 return printNode(Data.Id, Match, Result);
222
223 const auto *E = Match.Nodes.getNodeAs<Expr>(Data.Id);
224 if (E == nullptr)
225 return llvm::make_error<StringError>(
226 errc::invalid_argument, "Id not bound or not Expr: " + Data.Id);
227 llvm::Optional<std::string> Source;
228 switch (Data.Op) {
229 case UnaryNodeOperator::Parens:
230 Source = tooling::buildParens(*E, *Match.Context);
231 break;
232 case UnaryNodeOperator::Deref:
233 Source = tooling::buildDereference(*E, *Match.Context);
234 break;
235 case UnaryNodeOperator::MaybeDeref:
236 if (E->getType()->isAnyPointerType() ||
237 isSmartPointerType(E->getType(), *Match.Context)) {
238 // Strip off any operator->. This can only occur inside an actual arrow
239 // member access, so we treat it as equivalent to an actual object
240 // expression.
241 if (const auto *OpCall = dyn_cast<clang::CXXOperatorCallExpr>(E)) {
242 if (OpCall->getOperator() == clang::OO_Arrow &&
243 OpCall->getNumArgs() == 1) {
244 E = OpCall->getArg(0);
245 }
246 }
247 Source = tooling::buildDereference(*E, *Match.Context);
248 break;
249 }
250 *Result += tooling::getText(*E, *Match.Context);
251 return Error::success();
252 case UnaryNodeOperator::AddressOf:
253 Source = tooling::buildAddressOf(*E, *Match.Context);
254 break;
255 case UnaryNodeOperator::MaybeAddressOf:
256 if (E->getType()->isAnyPointerType() ||
257 isSmartPointerType(E->getType(), *Match.Context)) {
258 // Strip off any operator->. This can only occur inside an actual arrow
259 // member access, so we treat it as equivalent to an actual object
260 // expression.
261 if (const auto *OpCall = dyn_cast<clang::CXXOperatorCallExpr>(E)) {
262 if (OpCall->getOperator() == clang::OO_Arrow &&
263 OpCall->getNumArgs() == 1) {
264 E = OpCall->getArg(0);
265 }
266 }
267 *Result += tooling::getText(*E, *Match.Context);
268 return Error::success();
269 }
270 Source = tooling::buildAddressOf(*E, *Match.Context);
271 break;
272 case UnaryNodeOperator::Describe:
273 llvm_unreachable("This case is handled at the start of the function");
274 }
275 if (!Source)
276 return llvm::make_error<StringError>(
277 errc::invalid_argument,
278 "Could not construct expression source from ID: " + Data.Id);
279 *Result += *Source;
280 return Error::success();
281 }
282
evalData(const SelectorData & Data,const MatchFinder::MatchResult & Match,std::string * Result)283 Error evalData(const SelectorData &Data, const MatchFinder::MatchResult &Match,
284 std::string *Result) {
285 auto RawRange = Data.Selector(Match);
286 if (!RawRange)
287 return RawRange.takeError();
288 CharSourceRange Range = Lexer::makeFileCharRange(
289 *RawRange, *Match.SourceManager, Match.Context->getLangOpts());
290 if (Range.isInvalid()) {
291 // Validate the original range to attempt to get a meaningful error message.
292 // If it's valid, then something else is the cause and we just return the
293 // generic failure message.
294 if (auto Err = tooling::validateEditRange(*RawRange, *Match.SourceManager))
295 return handleErrors(std::move(Err), [](std::unique_ptr<StringError> E) {
296 assert(E->convertToErrorCode() ==
297 llvm::make_error_code(errc::invalid_argument) &&
298 "Validation errors must carry the invalid_argument code");
299 return llvm::createStringError(
300 errc::invalid_argument,
301 "selected range could not be resolved to a valid source range; " +
302 E->getMessage());
303 });
304 return llvm::createStringError(
305 errc::invalid_argument,
306 "selected range could not be resolved to a valid source range");
307 }
308 // Validate `Range`, because `makeFileCharRange` accepts some ranges that
309 // `validateEditRange` rejects.
310 if (auto Err = tooling::validateEditRange(Range, *Match.SourceManager))
311 return joinErrors(
312 llvm::createStringError(errc::invalid_argument,
313 "selected range is not valid for editing"),
314 std::move(Err));
315 *Result += tooling::getText(Range, *Match.Context);
316 return Error::success();
317 }
318
evalData(const AccessData & Data,const MatchFinder::MatchResult & Match,std::string * Result)319 Error evalData(const AccessData &Data, const MatchFinder::MatchResult &Match,
320 std::string *Result) {
321 const auto *E = Match.Nodes.getNodeAs<Expr>(Data.BaseId);
322 if (E == nullptr)
323 return llvm::make_error<StringError>(errc::invalid_argument,
324 "Id not bound: " + Data.BaseId);
325 if (!E->isImplicitCXXThis()) {
326 llvm::Optional<std::string> S;
327 if (E->getType()->isAnyPointerType() ||
328 isSmartPointerType(E->getType(), *Match.Context)) {
329 // Strip off any operator->. This can only occur inside an actual arrow
330 // member access, so we treat it as equivalent to an actual object
331 // expression.
332 if (const auto *OpCall = dyn_cast<clang::CXXOperatorCallExpr>(E)) {
333 if (OpCall->getOperator() == clang::OO_Arrow &&
334 OpCall->getNumArgs() == 1) {
335 E = OpCall->getArg(0);
336 }
337 }
338 S = tooling::buildArrow(*E, *Match.Context);
339 } else {
340 S = tooling::buildDot(*E, *Match.Context);
341 }
342 if (S.hasValue())
343 *Result += *S;
344 else
345 return llvm::make_error<StringError>(
346 errc::invalid_argument,
347 "Could not construct object text from ID: " + Data.BaseId);
348 }
349 return Data.Member->eval(Match, Result);
350 }
351
evalData(const IfBoundData & Data,const MatchFinder::MatchResult & Match,std::string * Result)352 Error evalData(const IfBoundData &Data, const MatchFinder::MatchResult &Match,
353 std::string *Result) {
354 auto &M = Match.Nodes.getMap();
355 return (M.find(Data.Id) != M.end() ? Data.TrueStencil : Data.FalseStencil)
356 ->eval(Match, Result);
357 }
358
evalData(const MatchConsumer<std::string> & Fn,const MatchFinder::MatchResult & Match,std::string * Result)359 Error evalData(const MatchConsumer<std::string> &Fn,
360 const MatchFinder::MatchResult &Match, std::string *Result) {
361 Expected<std::string> Value = Fn(Match);
362 if (!Value)
363 return Value.takeError();
364 *Result += *Value;
365 return Error::success();
366 }
367
evalData(const SequenceData & Data,const MatchFinder::MatchResult & Match,std::string * Result)368 Error evalData(const SequenceData &Data, const MatchFinder::MatchResult &Match,
369 std::string *Result) {
370 for (const auto &S : Data.Stencils)
371 if (auto Err = S->eval(Match, Result))
372 return Err;
373 return Error::success();
374 }
375
376 template <typename T> class StencilImpl : public StencilInterface {
377 T Data;
378
379 public:
380 template <typename... Ps>
StencilImpl(Ps &&...Args)381 explicit StencilImpl(Ps &&... Args) : Data(std::forward<Ps>(Args)...) {}
382
eval(const MatchFinder::MatchResult & Match,std::string * Result) const383 Error eval(const MatchFinder::MatchResult &Match,
384 std::string *Result) const override {
385 return evalData(Data, Match, Result);
386 }
387
toString() const388 std::string toString() const override { return toStringData(Data); }
389 };
390 } // namespace
391
makeStencil(StringRef Text)392 Stencil transformer::detail::makeStencil(StringRef Text) {
393 return std::make_shared<StencilImpl<RawTextData>>(std::string(Text));
394 }
395
makeStencil(RangeSelector Selector)396 Stencil transformer::detail::makeStencil(RangeSelector Selector) {
397 return std::make_shared<StencilImpl<SelectorData>>(std::move(Selector));
398 }
399
dPrint(StringRef Id)400 Stencil transformer::dPrint(StringRef Id) {
401 return std::make_shared<StencilImpl<DebugPrintNodeData>>(std::string(Id));
402 }
403
expression(llvm::StringRef Id)404 Stencil transformer::expression(llvm::StringRef Id) {
405 return std::make_shared<StencilImpl<UnaryOperationData>>(
406 UnaryNodeOperator::Parens, std::string(Id));
407 }
408
deref(llvm::StringRef ExprId)409 Stencil transformer::deref(llvm::StringRef ExprId) {
410 return std::make_shared<StencilImpl<UnaryOperationData>>(
411 UnaryNodeOperator::Deref, std::string(ExprId));
412 }
413
maybeDeref(llvm::StringRef ExprId)414 Stencil transformer::maybeDeref(llvm::StringRef ExprId) {
415 return std::make_shared<StencilImpl<UnaryOperationData>>(
416 UnaryNodeOperator::MaybeDeref, std::string(ExprId));
417 }
418
addressOf(llvm::StringRef ExprId)419 Stencil transformer::addressOf(llvm::StringRef ExprId) {
420 return std::make_shared<StencilImpl<UnaryOperationData>>(
421 UnaryNodeOperator::AddressOf, std::string(ExprId));
422 }
423
maybeAddressOf(llvm::StringRef ExprId)424 Stencil transformer::maybeAddressOf(llvm::StringRef ExprId) {
425 return std::make_shared<StencilImpl<UnaryOperationData>>(
426 UnaryNodeOperator::MaybeAddressOf, std::string(ExprId));
427 }
428
describe(StringRef Id)429 Stencil transformer::describe(StringRef Id) {
430 return std::make_shared<StencilImpl<UnaryOperationData>>(
431 UnaryNodeOperator::Describe, std::string(Id));
432 }
433
access(StringRef BaseId,Stencil Member)434 Stencil transformer::access(StringRef BaseId, Stencil Member) {
435 return std::make_shared<StencilImpl<AccessData>>(BaseId, std::move(Member));
436 }
437
ifBound(StringRef Id,Stencil TrueStencil,Stencil FalseStencil)438 Stencil transformer::ifBound(StringRef Id, Stencil TrueStencil,
439 Stencil FalseStencil) {
440 return std::make_shared<StencilImpl<IfBoundData>>(Id, std::move(TrueStencil),
441 std::move(FalseStencil));
442 }
443
run(MatchConsumer<std::string> Fn)444 Stencil transformer::run(MatchConsumer<std::string> Fn) {
445 return std::make_shared<StencilImpl<MatchConsumer<std::string>>>(
446 std::move(Fn));
447 }
448
catVector(std::vector<Stencil> Parts)449 Stencil transformer::catVector(std::vector<Stencil> Parts) {
450 // Only one argument, so don't wrap in sequence.
451 if (Parts.size() == 1)
452 return std::move(Parts[0]);
453 return std::make_shared<StencilImpl<SequenceData>>(std::move(Parts));
454 }
455