1 | //===--- MisleadingBidirectional.cpp - clang-tidy -------------------------===// |
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 "MisleadingBidirectional.h" |
10 | |
11 | #include "clang/Frontend/CompilerInstance.h" |
12 | #include "clang/Lex/Preprocessor.h" |
13 | #include "llvm/Support/ConvertUTF.h" |
14 | |
15 | using namespace clang; |
16 | using namespace clang::tidy::misc; |
17 | |
18 | static bool containsMisleadingBidi(StringRef Buffer, |
19 | bool HonorLineBreaks = true) { |
20 | const char *CurPtr = Buffer.begin(); |
21 | |
22 | enum BidiChar { |
23 | PS = 0x2029, |
24 | RLO = 0x202E, |
25 | RLE = 0x202B, |
26 | LRO = 0x202D, |
27 | LRE = 0x202A, |
28 | PDF = 0x202C, |
29 | RLI = 0x2067, |
30 | LRI = 0x2066, |
31 | FSI = 0x2068, |
32 | PDI = 0x2069 |
33 | }; |
34 | |
35 | SmallVector<BidiChar> BidiContexts; |
36 | |
37 | // Scan each character while maintaining a stack of opened bidi context. |
38 | // RLO/RLE/LRO/LRE all are closed by PDF while RLI LRI and FSI are closed by |
39 | // PDI. New lines reset the context count. Extra PDF / PDI are ignored. |
40 | // |
41 | // Warn if we end up with an unclosed context. |
42 | while (CurPtr < Buffer.end()) { |
43 | unsigned char C = *CurPtr; |
44 | if (isASCII(c: C)) { |
45 | ++CurPtr; |
46 | bool IsParagrapSep = |
47 | (C == 0xA || C == 0xD || (0x1C <= C && C <= 0x1E) || C == 0x85); |
48 | bool IsSegmentSep = (C == 0x9 || C == 0xB || C == 0x1F); |
49 | if (IsParagrapSep || IsSegmentSep) |
50 | BidiContexts.clear(); |
51 | continue; |
52 | } |
53 | llvm::UTF32 CodePoint = 0; |
54 | llvm::ConversionResult Result = llvm::convertUTF8Sequence( |
55 | source: (const llvm::UTF8 **)&CurPtr, sourceEnd: (const llvm::UTF8 *)Buffer.end(), |
56 | target: &CodePoint, flags: llvm::strictConversion); |
57 | |
58 | // If conversion fails, utf-8 is designed so that we can just try next char. |
59 | if (Result != llvm::conversionOK) { |
60 | ++CurPtr; |
61 | continue; |
62 | } |
63 | |
64 | // Open a PDF context. |
65 | if (CodePoint == RLO || CodePoint == RLE || CodePoint == LRO || |
66 | CodePoint == LRE) |
67 | BidiContexts.push_back(Elt: PDF); |
68 | // Close PDF Context. |
69 | else if (CodePoint == PDF) { |
70 | if (!BidiContexts.empty() && BidiContexts.back() == PDF) |
71 | BidiContexts.pop_back(); |
72 | } |
73 | // Open a PDI Context. |
74 | else if (CodePoint == RLI || CodePoint == LRI || CodePoint == FSI) |
75 | BidiContexts.push_back(Elt: PDI); |
76 | // Close a PDI Context. |
77 | else if (CodePoint == PDI) { |
78 | auto R = llvm::find(Range: llvm::reverse(C&: BidiContexts), Val: PDI); |
79 | if (R != BidiContexts.rend()) |
80 | BidiContexts.resize(N: BidiContexts.rend() - R - 1); |
81 | } |
82 | // Line break or equivalent |
83 | else if (CodePoint == PS) |
84 | BidiContexts.clear(); |
85 | } |
86 | return !BidiContexts.empty(); |
87 | } |
88 | |
89 | class MisleadingBidirectionalCheck::MisleadingBidirectionalHandler |
90 | : public CommentHandler { |
91 | public: |
92 | MisleadingBidirectionalHandler(MisleadingBidirectionalCheck &Check) |
93 | : Check(Check) {} |
94 | |
95 | bool HandleComment(Preprocessor &PP, SourceRange Range) override { |
96 | // FIXME: check that we are in a /* */ comment |
97 | StringRef Text = |
98 | Lexer::getSourceText(Range: CharSourceRange::getCharRange(R: Range), |
99 | SM: PP.getSourceManager(), LangOpts: PP.getLangOpts()); |
100 | |
101 | if (containsMisleadingBidi(Buffer: Text, HonorLineBreaks: true)) |
102 | Check.diag( |
103 | Loc: Range.getBegin(), |
104 | Description: "comment contains misleading bidirectional Unicode characters" ); |
105 | return false; |
106 | } |
107 | |
108 | private: |
109 | MisleadingBidirectionalCheck &Check; |
110 | }; |
111 | |
112 | MisleadingBidirectionalCheck::MisleadingBidirectionalCheck( |
113 | StringRef Name, ClangTidyContext *Context) |
114 | : ClangTidyCheck(Name, Context), |
115 | Handler(std::make_unique<MisleadingBidirectionalHandler>(args&: *this)) {} |
116 | |
117 | MisleadingBidirectionalCheck::~MisleadingBidirectionalCheck() = default; |
118 | |
119 | void MisleadingBidirectionalCheck::registerPPCallbacks( |
120 | const SourceManager &SM, Preprocessor *PP, Preprocessor *ModuleExpanderPP) { |
121 | PP->addCommentHandler(Handler: Handler.get()); |
122 | } |
123 | |
124 | void MisleadingBidirectionalCheck::check( |
125 | const ast_matchers::MatchFinder::MatchResult &Result) { |
126 | if (const auto *SL = Result.Nodes.getNodeAs<StringLiteral>(ID: "strlit" )) { |
127 | StringRef Literal = SL->getBytes(); |
128 | if (containsMisleadingBidi(Buffer: Literal, HonorLineBreaks: false)) |
129 | diag(Loc: SL->getBeginLoc(), Description: "string literal contains misleading " |
130 | "bidirectional Unicode characters" ); |
131 | } |
132 | } |
133 | |
134 | void MisleadingBidirectionalCheck::registerMatchers( |
135 | ast_matchers::MatchFinder *Finder) { |
136 | Finder->addMatcher(NodeMatch: ast_matchers::stringLiteral().bind(ID: "strlit" ), Action: this); |
137 | } |
138 | |