1//===--- SourceExtraction.cpp - Clang refactoring library -----------------===//
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/Refactoring/Extract/SourceExtraction.h"
10#include "clang/AST/Stmt.h"
11#include "clang/AST/StmtCXX.h"
12#include "clang/AST/StmtObjC.h"
13#include "clang/Basic/SourceManager.h"
14#include "clang/Lex/Lexer.h"
15
16using namespace clang;
17
18namespace {
19
20/// Returns true if the token at the given location is a semicolon.
21bool isSemicolonAtLocation(SourceLocation TokenLoc, const SourceManager &SM,
22                           const LangOptions &LangOpts) {
23  return Lexer::getSourceText(
24             CharSourceRange::getTokenRange(TokenLoc, TokenLoc), SM,
25             LangOpts) == ";";
26}
27
28/// Returns true if there should be a semicolon after the given statement.
29bool isSemicolonRequiredAfter(const Stmt *S) {
30  if (isa<CompoundStmt>(S))
31    return false;
32  if (const auto *If = dyn_cast<IfStmt>(S))
33    return isSemicolonRequiredAfter(If->getElse() ? If->getElse()
34                                                  : If->getThen());
35  if (const auto *While = dyn_cast<WhileStmt>(S))
36    return isSemicolonRequiredAfter(While->getBody());
37  if (const auto *For = dyn_cast<ForStmt>(S))
38    return isSemicolonRequiredAfter(For->getBody());
39  if (const auto *CXXFor = dyn_cast<CXXForRangeStmt>(S))
40    return isSemicolonRequiredAfter(CXXFor->getBody());
41  if (const auto *ObjCFor = dyn_cast<ObjCForCollectionStmt>(S))
42    return isSemicolonRequiredAfter(ObjCFor->getBody());
43  if(const auto *Switch = dyn_cast<SwitchStmt>(S))
44    return isSemicolonRequiredAfter(Switch->getBody());
45  if(const auto *Case = dyn_cast<SwitchCase>(S))
46    return isSemicolonRequiredAfter(Case->getSubStmt());
47  switch (S->getStmtClass()) {
48  case Stmt::DeclStmtClass:
49  case Stmt::CXXTryStmtClass:
50  case Stmt::ObjCAtSynchronizedStmtClass:
51  case Stmt::ObjCAutoreleasePoolStmtClass:
52  case Stmt::ObjCAtTryStmtClass:
53    return false;
54  default:
55    return true;
56  }
57}
58
59/// Returns true if the two source locations are on the same line.
60bool areOnSameLine(SourceLocation Loc1, SourceLocation Loc2,
61                   const SourceManager &SM) {
62  return !Loc1.isMacroID() && !Loc2.isMacroID() &&
63         SM.getSpellingLineNumber(Loc1) == SM.getSpellingLineNumber(Loc2);
64}
65
66} // end anonymous namespace
67
68namespace clang {
69namespace tooling {
70
71ExtractionSemicolonPolicy
72ExtractionSemicolonPolicy::compute(const Stmt *S, SourceRange &ExtractedRange,
73                                   const SourceManager &SM,
74                                   const LangOptions &LangOpts) {
75  auto neededInExtractedFunction = []() {
76    return ExtractionSemicolonPolicy(true, false);
77  };
78  auto neededInOriginalFunction = []() {
79    return ExtractionSemicolonPolicy(false, true);
80  };
81
82  /// The extracted expression should be terminated with a ';'. The call to
83  /// the extracted function will replace this expression, so it won't need
84  /// a terminating ';'.
85  if (isa<Expr>(S))
86    return neededInExtractedFunction();
87
88  /// Some statements don't need to be terminated with ';'. The call to the
89  /// extracted function will be a standalone statement, so it should be
90  /// terminated with a ';'.
91  bool NeedsSemi = isSemicolonRequiredAfter(S);
92  if (!NeedsSemi)
93    return neededInOriginalFunction();
94
95  /// Some statements might end at ';'. The extraction will move that ';', so
96  /// the call to the extracted function should be terminated with a ';'.
97  SourceLocation End = ExtractedRange.getEnd();
98  if (isSemicolonAtLocation(End, SM, LangOpts))
99    return neededInOriginalFunction();
100
101  /// Other statements should generally have a trailing ';'. We can try to find
102  /// it and move it together it with the extracted code.
103  Optional<Token> NextToken = Lexer::findNextToken(End, SM, LangOpts);
104  if (NextToken && NextToken->is(tok::semi) &&
105      areOnSameLine(NextToken->getLocation(), End, SM)) {
106    ExtractedRange.setEnd(NextToken->getLocation());
107    return neededInOriginalFunction();
108  }
109
110  /// Otherwise insert semicolons in both places.
111  return ExtractionSemicolonPolicy(true, true);
112}
113
114} // end namespace tooling
115} // end namespace clang
116