1 //===--- RangeSelector.cpp - RangeSelector implementations ------*- 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/RangeSelector.h"
10 #include "clang/AST/Expr.h"
11 #include "clang/ASTMatchers/ASTMatchFinder.h"
12 #include "clang/Basic/SourceLocation.h"
13 #include "clang/Lex/Lexer.h"
14 #include "clang/Tooling/Transformer/SourceCode.h"
15 #include "llvm/ADT/StringRef.h"
16 #include "llvm/Support/Errc.h"
17 #include "llvm/Support/Error.h"
18 #include <string>
19 #include <utility>
20 #include <vector>
21 
22 using namespace clang;
23 using namespace transformer;
24 
25 using ast_matchers::MatchFinder;
26 using llvm::Error;
27 using llvm::StringError;
28 
29 using MatchResult = MatchFinder::MatchResult;
30 
31 static Error invalidArgumentError(Twine Message) {
32   return llvm::make_error<StringError>(llvm::errc::invalid_argument, Message);
33 }
34 
35 static Error typeError(StringRef ID, const ASTNodeKind &Kind) {
36   return invalidArgumentError("mismatched type (node id=" + ID +
37                               " kind=" + Kind.asStringRef() + ")");
38 }
39 
40 static Error typeError(StringRef ID, const ASTNodeKind &Kind,
41                        Twine ExpectedType) {
42   return invalidArgumentError("mismatched type: expected one of " +
43                               ExpectedType + " (node id=" + ID +
44                               " kind=" + Kind.asStringRef() + ")");
45 }
46 
47 static Error missingPropertyError(StringRef ID, Twine Description,
48                                   StringRef Property) {
49   return invalidArgumentError(Description + " requires property '" + Property +
50                               "' (node id=" + ID + ")");
51 }
52 
53 static Expected<DynTypedNode> getNode(const ast_matchers::BoundNodes &Nodes,
54                                       StringRef ID) {
55   auto &NodesMap = Nodes.getMap();
56   auto It = NodesMap.find(ID);
57   if (It == NodesMap.end())
58     return invalidArgumentError("ID not bound: " + ID);
59   return It->second;
60 }
61 
62 // FIXME: handling of macros should be configurable.
63 static SourceLocation findPreviousTokenStart(SourceLocation Start,
64                                              const SourceManager &SM,
65                                              const LangOptions &LangOpts) {
66   if (Start.isInvalid() || Start.isMacroID())
67     return SourceLocation();
68 
69   SourceLocation BeforeStart = Start.getLocWithOffset(-1);
70   if (BeforeStart.isInvalid() || BeforeStart.isMacroID())
71     return SourceLocation();
72 
73   return Lexer::GetBeginningOfToken(BeforeStart, SM, LangOpts);
74 }
75 
76 // Finds the start location of the previous token of kind \p TK.
77 // FIXME: handling of macros should be configurable.
78 static SourceLocation findPreviousTokenKind(SourceLocation Start,
79                                             const SourceManager &SM,
80                                             const LangOptions &LangOpts,
81                                             tok::TokenKind TK) {
82   while (true) {
83     SourceLocation L = findPreviousTokenStart(Start, SM, LangOpts);
84     if (L.isInvalid() || L.isMacroID())
85       return SourceLocation();
86 
87     Token T;
88     if (Lexer::getRawToken(L, T, SM, LangOpts, /*IgnoreWhiteSpace=*/true))
89       return SourceLocation();
90 
91     if (T.is(TK))
92       return T.getLocation();
93 
94     Start = L;
95   }
96 }
97 
98 static SourceLocation findOpenParen(const CallExpr &E, const SourceManager &SM,
99                                     const LangOptions &LangOpts) {
100   SourceLocation EndLoc =
101       E.getNumArgs() == 0 ? E.getRParenLoc() : E.getArg(0)->getBeginLoc();
102   return findPreviousTokenKind(EndLoc, SM, LangOpts, tok::TokenKind::l_paren);
103 }
104 
105 RangeSelector transformer::before(RangeSelector Selector) {
106   return [Selector](const MatchResult &Result) -> Expected<CharSourceRange> {
107     Expected<CharSourceRange> SelectedRange = Selector(Result);
108     if (!SelectedRange)
109       return SelectedRange.takeError();
110     return CharSourceRange::getCharRange(SelectedRange->getBegin());
111   };
112 }
113 
114 RangeSelector transformer::after(RangeSelector Selector) {
115   return [Selector](const MatchResult &Result) -> Expected<CharSourceRange> {
116     Expected<CharSourceRange> SelectedRange = Selector(Result);
117     if (!SelectedRange)
118       return SelectedRange.takeError();
119     if (SelectedRange->isCharRange())
120       return CharSourceRange::getCharRange(SelectedRange->getEnd());
121     return CharSourceRange::getCharRange(Lexer::getLocForEndOfToken(
122         SelectedRange->getEnd(), 0, Result.Context->getSourceManager(),
123         Result.Context->getLangOpts()));
124   };
125 }
126 
127 RangeSelector transformer::node(std::string ID) {
128   return [ID](const MatchResult &Result) -> Expected<CharSourceRange> {
129     Expected<DynTypedNode> Node = getNode(Result.Nodes, ID);
130     if (!Node)
131       return Node.takeError();
132     return Node->get<Stmt>() != nullptr && Node->get<Expr>() == nullptr
133                ? tooling::getExtendedRange(*Node, tok::TokenKind::semi,
134                                            *Result.Context)
135                : CharSourceRange::getTokenRange(Node->getSourceRange());
136   };
137 }
138 
139 RangeSelector transformer::statement(std::string ID) {
140   return [ID](const MatchResult &Result) -> Expected<CharSourceRange> {
141     Expected<DynTypedNode> Node = getNode(Result.Nodes, ID);
142     if (!Node)
143       return Node.takeError();
144     return tooling::getExtendedRange(*Node, tok::TokenKind::semi,
145                                      *Result.Context);
146   };
147 }
148 
149 RangeSelector transformer::range(RangeSelector Begin, RangeSelector End) {
150   return [Begin, End](const MatchResult &Result) -> Expected<CharSourceRange> {
151     Expected<CharSourceRange> BeginRange = Begin(Result);
152     if (!BeginRange)
153       return BeginRange.takeError();
154     Expected<CharSourceRange> EndRange = End(Result);
155     if (!EndRange)
156       return EndRange.takeError();
157     SourceLocation B = BeginRange->getBegin();
158     SourceLocation E = EndRange->getEnd();
159     // Note: we are precluding the possibility of sub-token ranges in the case
160     // that EndRange is a token range.
161     if (Result.SourceManager->isBeforeInTranslationUnit(E, B)) {
162       return invalidArgumentError("Bad range: out of order");
163     }
164     return CharSourceRange(SourceRange(B, E), EndRange->isTokenRange());
165   };
166 }
167 
168 RangeSelector transformer::range(std::string BeginID, std::string EndID) {
169   return transformer::range(node(std::move(BeginID)), node(std::move(EndID)));
170 }
171 
172 RangeSelector transformer::member(std::string ID) {
173   return [ID](const MatchResult &Result) -> Expected<CharSourceRange> {
174     Expected<DynTypedNode> Node = getNode(Result.Nodes, ID);
175     if (!Node)
176       return Node.takeError();
177     if (auto *M = Node->get<clang::MemberExpr>())
178       return CharSourceRange::getTokenRange(
179           M->getMemberNameInfo().getSourceRange());
180     return typeError(ID, Node->getNodeKind(), "MemberExpr");
181   };
182 }
183 
184 RangeSelector transformer::name(std::string ID) {
185   return [ID](const MatchResult &Result) -> Expected<CharSourceRange> {
186     Expected<DynTypedNode> N = getNode(Result.Nodes, ID);
187     if (!N)
188       return N.takeError();
189     auto &Node = *N;
190     if (const auto *D = Node.get<NamedDecl>()) {
191       if (!D->getDeclName().isIdentifier())
192         return missingPropertyError(ID, "name", "identifier");
193       SourceLocation L = D->getLocation();
194       auto R = CharSourceRange::getTokenRange(L, L);
195       // Verify that the range covers exactly the name.
196       // FIXME: extend this code to support cases like `operator +` or
197       // `foo<int>` for which this range will be too short.  Doing so will
198       // require subcasing `NamedDecl`, because it doesn't provide virtual
199       // access to the \c DeclarationNameInfo.
200       if (tooling::getText(R, *Result.Context) != D->getName())
201         return CharSourceRange();
202       return R;
203     }
204     if (const auto *E = Node.get<DeclRefExpr>()) {
205       if (!E->getNameInfo().getName().isIdentifier())
206         return missingPropertyError(ID, "name", "identifier");
207       SourceLocation L = E->getLocation();
208       return CharSourceRange::getTokenRange(L, L);
209     }
210     if (const auto *I = Node.get<CXXCtorInitializer>()) {
211       if (!I->isMemberInitializer() && I->isWritten())
212         return missingPropertyError(ID, "name", "explicit member initializer");
213       SourceLocation L = I->getMemberLocation();
214       return CharSourceRange::getTokenRange(L, L);
215     }
216     return typeError(ID, Node.getNodeKind(),
217                      "DeclRefExpr, NamedDecl, CXXCtorInitializer");
218   };
219 }
220 
221 namespace {
222 // FIXME: make this available in the public API for users to easily create their
223 // own selectors.
224 
225 // Creates a selector from a range-selection function \p Func, which selects a
226 // range that is relative to a bound node id.  \c T is the node type expected by
227 // \p Func.
228 template <typename T, CharSourceRange (*Func)(const MatchResult &, const T &)>
229 class RelativeSelector {
230   std::string ID;
231 
232 public:
233   RelativeSelector(std::string ID) : ID(std::move(ID)) {}
234 
235   Expected<CharSourceRange> operator()(const MatchResult &Result) {
236     Expected<DynTypedNode> N = getNode(Result.Nodes, ID);
237     if (!N)
238       return N.takeError();
239     if (const auto *Arg = N->get<T>())
240       return Func(Result, *Arg);
241     return typeError(ID, N->getNodeKind());
242   }
243 };
244 } // namespace
245 
246 // FIXME: Change the following functions from being in an anonymous namespace
247 // to static functions, after the minimum Visual C++ has _MSC_VER >= 1915
248 // (equivalent to Visual Studio 2017 v15.8 or higher). Using the anonymous
249 // namespace works around a bug in earlier versions.
250 namespace {
251 // Returns the range of the statements (all source between the braces).
252 CharSourceRange getStatementsRange(const MatchResult &,
253                                    const CompoundStmt &CS) {
254   return CharSourceRange::getCharRange(CS.getLBracLoc().getLocWithOffset(1),
255                                        CS.getRBracLoc());
256 }
257 } // namespace
258 
259 RangeSelector transformer::statements(std::string ID) {
260   return RelativeSelector<CompoundStmt, getStatementsRange>(std::move(ID));
261 }
262 
263 namespace {
264 // Returns the range of the source between the call's parentheses.
265 CharSourceRange getCallArgumentsRange(const MatchResult &Result,
266                                       const CallExpr &CE) {
267   return CharSourceRange::getCharRange(
268       findOpenParen(CE, *Result.SourceManager, Result.Context->getLangOpts())
269           .getLocWithOffset(1),
270       CE.getRParenLoc());
271 }
272 } // namespace
273 
274 RangeSelector transformer::callArgs(std::string ID) {
275   return RelativeSelector<CallExpr, getCallArgumentsRange>(std::move(ID));
276 }
277 
278 namespace {
279 // Returns the range of the elements of the initializer list. Includes all
280 // source between the braces.
281 CharSourceRange getElementsRange(const MatchResult &,
282                                  const InitListExpr &E) {
283   return CharSourceRange::getCharRange(E.getLBraceLoc().getLocWithOffset(1),
284                                        E.getRBraceLoc());
285 }
286 } // namespace
287 
288 RangeSelector transformer::initListElements(std::string ID) {
289   return RelativeSelector<InitListExpr, getElementsRange>(std::move(ID));
290 }
291 
292 namespace {
293 // Returns the range of the else branch, including the `else` keyword.
294 CharSourceRange getElseRange(const MatchResult &Result, const IfStmt &S) {
295   return tooling::maybeExtendRange(
296       CharSourceRange::getTokenRange(S.getElseLoc(), S.getEndLoc()),
297       tok::TokenKind::semi, *Result.Context);
298 }
299 } // namespace
300 
301 RangeSelector transformer::elseBranch(std::string ID) {
302   return RelativeSelector<IfStmt, getElseRange>(std::move(ID));
303 }
304 
305 RangeSelector transformer::expansion(RangeSelector S) {
306   return [S](const MatchResult &Result) -> Expected<CharSourceRange> {
307     Expected<CharSourceRange> SRange = S(Result);
308     if (!SRange)
309       return SRange.takeError();
310     return Result.SourceManager->getExpansionRange(*SRange);
311   };
312 }
313