clang-tools-extra/clang-tidy/misc/MisleadingIdentifier.cpp

   1 //===--- MisleadingIdentifier.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 "MisleadingIdentifier.h"
  10
  11 #include "clang/Frontend/CompilerInstance.h"
  12 #include "llvm/Support/ConvertUTF.h"
  13
  14 namespace clang::tidy::misc {
  15
  16 // See https://www.unicode.org/Public/14.0.0/ucd/extracted/DerivedBidiClass.txt
  17 static bool isUnassignedAL(llvm::UTF32 CP) {
  18   return (0x0600 <= CP && CP <= 0x07BF) || (0x0860 <= CP && CP <= 0x08FF) ||
  19          (0xFB50 <= CP && CP <= 0xFDCF) || (0xFDF0 <= CP && CP <= 0xFDFF) ||
  20          (0xFE70 <= CP && CP <= 0xFEFF) ||
  21          (0x00010D00 <= CP && CP <= 0x00010D3F) ||
  22          (0x00010F30 <= CP && CP <= 0x00010F6F) ||
  23          (0x0001EC70 <= CP && CP <= 0x0001ECBF) ||
  24          (0x0001ED00 <= CP && CP <= 0x0001ED4F) ||
  25          (0x0001EE00 <= CP && CP <= 0x0001EEFF);
  26 }
  27
  28 // See https://www.unicode.org/Public/14.0.0/ucd/extracted/DerivedBidiClass.txt
  29 static bool isUnassignedR(llvm::UTF32 CP) {
  30   return (0x0590 <= CP && CP <= 0x05FF) || (0x07C0 <= CP && CP <= 0x085F) ||
  31          (0xFB1D <= CP && CP <= 0xFB4F) ||
  32          (0x00010800 <= CP && CP <= 0x00010CFF) ||
  33          (0x00010D40 <= CP && CP <= 0x00010F2F) ||
  34          (0x00010F70 <= CP && CP <= 0x00010FFF) ||
  35          (0x0001E800 <= CP && CP <= 0x0001EC6F) ||
  36          (0x0001ECC0 <= CP && CP <= 0x0001ECFF) ||
  37          (0x0001ED50 <= CP && CP <= 0x0001EDFF) ||
  38          (0x0001EF00 <= CP && CP <= 0x0001EFFF);
  39 }
  40
  41 // See https://www.unicode.org/Public/14.0.0/ucd/extracted/DerivedBidiClass.txt
  42 static bool isR(llvm::UTF32 CP) {
  43   return (CP == 0x0590) || (CP == 0x05BE) || (CP == 0x05C0) || (CP == 0x05C3) ||
  44          (CP == 0x05C6) || (0x05C8 <= CP && CP <= 0x05CF) ||
  45          (0x05D0 <= CP && CP <= 0x05EA) || (0x05EB <= CP && CP <= 0x05EE) ||
  46          (0x05EF <= CP && CP <= 0x05F2) || (0x05F3 <= CP && CP <= 0x05F4) ||
  47          (0x05F5 <= CP && CP <= 0x05FF) || (0x07C0 <= CP && CP <= 0x07C9) ||
  48          (0x07CA <= CP && CP <= 0x07EA) || (0x07F4 <= CP && CP <= 0x07F5) ||
  49          (CP == 0x07FA) || (0x07FB <= CP && CP <= 0x07FC) ||
  50          (0x07FE <= CP && CP <= 0x07FF) || (0x0800 <= CP && CP <= 0x0815) ||
  51          (CP == 0x081A) || (CP == 0x0824) || (CP == 0x0828) ||
  52          (0x082E <= CP && CP <= 0x082F) || (0x0830 <= CP && CP <= 0x083E) ||
  53          (CP == 0x083F) || (0x0840 <= CP && CP <= 0x0858) ||
  54          (0x085C <= CP && CP <= 0x085D) || (CP == 0x085E) || (CP == 0x085F) ||
  55          (CP == 0x200F) || (CP == 0xFB1D) || (0xFB1F <= CP && CP <= 0xFB28) ||
  56          (0xFB2A <= CP && CP <= 0xFB36) || (CP == 0xFB37) ||
  57          (0xFB38 <= CP && CP <= 0xFB3C) || (CP == 0xFB3D) || (CP == 0xFB3E) ||
  58          (CP == 0xFB3F) || (0xFB40 <= CP && CP <= 0xFB41) || (CP == 0xFB42) ||
  59          (0xFB43 <= CP && CP <= 0xFB44) || (CP == 0xFB45) ||
  60          (0xFB46 <= CP && CP <= 0xFB4F) || (0x10800 <= CP && CP <= 0x10805) ||
  61          (0x10806 <= CP && CP <= 0x10807) || (CP == 0x10808) ||
  62          (CP == 0x10809) || (0x1080A <= CP && CP <= 0x10835) ||
  63          (CP == 0x10836) || (0x10837 <= CP && CP <= 0x10838) ||
  64          (0x10839 <= CP && CP <= 0x1083B) || (CP == 0x1083C) ||
  65          (0x1083D <= CP && CP <= 0x1083E) || (0x1083F <= CP && CP <= 0x10855) ||
  66          (CP == 0x10856) || (CP == 0x10857) ||
  67          (0x10858 <= CP && CP <= 0x1085F) || (0x10860 <= CP && CP <= 0x10876) ||
  68          (0x10877 <= CP && CP <= 0x10878) || (0x10879 <= CP && CP <= 0x1087F) ||
  69          (0x10880 <= CP && CP <= 0x1089E) || (0x1089F <= CP && CP <= 0x108A6) ||
  70          (0x108A7 <= CP && CP <= 0x108AF) || (0x108B0 <= CP && CP <= 0x108DF) ||
  71          (0x108E0 <= CP && CP <= 0x108F2) || (CP == 0x108F3) ||
  72          (0x108F4 <= CP && CP <= 0x108F5) || (0x108F6 <= CP && CP <= 0x108FA) ||
  73          (0x108FB <= CP && CP <= 0x108FF) || (0x10900 <= CP && CP <= 0x10915) ||
  74          (0x10916 <= CP && CP <= 0x1091B) || (0x1091C <= CP && CP <= 0x1091E) ||
  75          (0x10920 <= CP && CP <= 0x10939) || (0x1093A <= CP && CP <= 0x1093E) ||
  76          (CP == 0x1093F) || (0x10940 <= CP && CP <= 0x1097F) ||
  77          (0x10980 <= CP && CP <= 0x109B7) || (0x109B8 <= CP && CP <= 0x109BB) ||
  78          (0x109BC <= CP && CP <= 0x109BD) || (0x109BE <= CP && CP <= 0x109BF) ||
  79          (0x109C0 <= CP && CP <= 0x109CF) || (0x109D0 <= CP && CP <= 0x109D1) ||
  80          (0x109D2 <= CP && CP <= 0x109FF) || (CP == 0x10A00) ||
  81          (CP == 0x10A04) || (0x10A07 <= CP && CP <= 0x10A0B) ||
  82          (0x10A10 <= CP && CP <= 0x10A13) || (CP == 0x10A14) ||
  83          (0x10A15 <= CP && CP <= 0x10A17) || (CP == 0x10A18) ||
  84          (0x10A19 <= CP && CP <= 0x10A35) || (0x10A36 <= CP && CP <= 0x10A37) ||
  85          (0x10A3B <= CP && CP <= 0x10A3E) || (0x10A40 <= CP && CP <= 0x10A48) ||
  86          (0x10A49 <= CP && CP <= 0x10A4F) || (0x10A50 <= CP && CP <= 0x10A58) ||
  87          (0x10A59 <= CP && CP <= 0x10A5F) || (0x10A60 <= CP && CP <= 0x10A7C) ||
  88          (0x10A7D <= CP && CP <= 0x10A7E) || (CP == 0x10A7F) ||
  89          (0x10A80 <= CP && CP <= 0x10A9C) || (0x10A9D <= CP && CP <= 0x10A9F) ||
  90          (0x10AA0 <= CP && CP <= 0x10ABF) || (0x10AC0 <= CP && CP <= 0x10AC7) ||
  91          (CP == 0x10AC8) || (0x10AC9 <= CP && CP <= 0x10AE4) ||
  92          (0x10AE7 <= CP && CP <= 0x10AEA) || (0x10AEB <= CP && CP <= 0x10AEF) ||
  93          (0x10AF0 <= CP && CP <= 0x10AF6) || (0x10AF7 <= CP && CP <= 0x10AFF) ||
  94          (0x10B00 <= CP && CP <= 0x10B35) || (0x10B36 <= CP && CP <= 0x10B38) ||
  95          (0x10B40 <= CP && CP <= 0x10B55) || (0x10B56 <= CP && CP <= 0x10B57) ||
  96          (0x10B58 <= CP && CP <= 0x10B5F) || (0x10B60 <= CP && CP <= 0x10B72) ||
  97          (0x10B73 <= CP && CP <= 0x10B77) || (0x10B78 <= CP && CP <= 0x10B7F) ||
  98          (0x10B80 <= CP && CP <= 0x10B91) || (0x10B92 <= CP && CP <= 0x10B98) ||
  99          (0x10B99 <= CP && CP <= 0x10B9C) || (0x10B9D <= CP && CP <= 0x10BA8) ||
 100          (0x10BA9 <= CP && CP <= 0x10BAF) || (0x10BB0 <= CP && CP <= 0x10BFF) ||
 101          (0x10C00 <= CP && CP <= 0x10C48) || (0x10C49 <= CP && CP <= 0x10C7F) ||
 102          (0x10C80 <= CP && CP <= 0x10CB2) || (0x10CB3 <= CP && CP <= 0x10CBF) ||
 103          (0x10CC0 <= CP && CP <= 0x10CF2) || (0x10CF3 <= CP && CP <= 0x10CF9) ||
 104          (0x10CFA <= CP && CP <= 0x10CFF) || (0x10D40 <= CP && CP <= 0x10E5F) ||
 105          (CP == 0x10E7F) || (0x10E80 <= CP && CP <= 0x10EA9) ||
 106          (CP == 0x10EAA) || (CP == 0x10EAD) ||
 107          (0x10EAE <= CP && CP <= 0x10EAF) || (0x10EB0 <= CP && CP <= 0x10EB1) ||
 108          (0x10EB2 <= CP && CP <= 0x10EFF) || (0x10F00 <= CP && CP <= 0x10F1C) ||
 109          (0x10F1D <= CP && CP <= 0x10F26) || (CP == 0x10F27) ||
 110          (0x10F28 <= CP && CP <= 0x10F2F) || (0x10F70 <= CP && CP <= 0x10F81) ||
 111          (0x10F86 <= CP && CP <= 0x10F89) || (0x10F8A <= CP && CP <= 0x10FAF) ||
 112          (0x10FB0 <= CP && CP <= 0x10FC4) || (0x10FC5 <= CP && CP <= 0x10FCB) ||
 113          (0x10FCC <= CP && CP <= 0x10FDF) || (0x10FE0 <= CP && CP <= 0x10FF6) ||
 114          (0x10FF7 <= CP && CP <= 0x10FFF) || (0x1E800 <= CP && CP <= 0x1E8C4) ||
 115          (0x1E8C5 <= CP && CP <= 0x1E8C6) || (0x1E8C7 <= CP && CP <= 0x1E8CF) ||
 116          (0x1E8D7 <= CP && CP <= 0x1E8FF) || (0x1E900 <= CP && CP <= 0x1E943) ||
 117          (CP == 0x1E94B) || (0x1E94C <= CP && CP <= 0x1E94F) ||
 118          (0x1E950 <= CP && CP <= 0x1E959) || (0x1E95A <= CP && CP <= 0x1E95D) ||
 119          (0x1E95E <= CP && CP <= 0x1E95F) || (0x1E960 <= CP && CP <= 0x1EC6F) ||
 120          (0x1ECC0 <= CP && CP <= 0x1ECFF) || (0x1ED50 <= CP && CP <= 0x1EDFF);
 121 }
 122
 123 static bool hasRTLCharacters(StringRef Buffer) {
 124   const char *CurPtr = Buffer.begin();
 125   const char *EndPtr = Buffer.end();
 126   while (CurPtr < EndPtr) {
 127     llvm::UTF32 CodePoint;
 128     llvm::ConversionResult Result = llvm::convertUTF8Sequence(
 129         (const llvm::UTF8 **)&CurPtr, (const llvm::UTF8 *)EndPtr, &CodePoint,
 130         llvm::strictConversion);
 131     if (Result != llvm::conversionOK)
 132       break;
 133     if (isUnassignedAL(CodePoint) || isUnassignedR(CodePoint) || isR(CodePoint))
 134       return true;
 135   }
 136   return false;
 137 }
 138
 139 MisleadingIdentifierCheck::MisleadingIdentifierCheck(StringRef Name,
 140                                                      ClangTidyContext *Context)
 141     : ClangTidyCheck(Name, Context) {}
 142
 143 MisleadingIdentifierCheck::~MisleadingIdentifierCheck() = default;
 144
 145 void MisleadingIdentifierCheck::check(
 146     const ast_matchers::MatchFinder::MatchResult &Result) {
 147   if (const auto *ND = Result.Nodes.getNodeAs<NamedDecl>("nameddecl")) {
 148     IdentifierInfo *II = ND->getIdentifier();
 149     if (II) {
 150       StringRef NDName = II->getName();
 151       if (hasRTLCharacters(NDName))
 152         diag(ND->getBeginLoc(), "identifier has right-to-left codepoints");
 153     }
 154   }
 155 }
 156
 157 void MisleadingIdentifierCheck::registerMatchers(
 158     ast_matchers::MatchFinder *Finder) {
 159   Finder->addMatcher(ast_matchers::namedDecl().bind("nameddecl"), this);
 160 }
 161
 162 } // namespace clang::tidy::misc