[clang-format] Fix FormatToken::isSimpleTypeSpecifier()

clang/lib/Format/Format.cpp

@@ -3858,8 +3858,7 @@ LangOptions getFormattingLangOpts(const FormatStyle &Style) {
   LangOpts.Digraphs = LexingStd >= FormatStyle::LS_Cpp11;
 
   LangOpts.LineComment = 1;
-  bool AlternativeOperators = Style.isCpp();
-  LangOpts.CXXOperatorNames = AlternativeOperators ? 1 : 0;
+  LangOpts.CXXOperatorNames = Style.isCpp();
   LangOpts.Bool = 1;
   LangOpts.ObjC = 1;
   LangOpts.MicrosoftExt = 1;    // To get kw___try, kw___finally.

clang/lib/Format/FormatToken.cpp

@@ -34,59 +34,23 @@ const char *getTokenTypeName(TokenType Type) {
   return nullptr;
 }
 
-// FIXME: This is copy&pasted from Sema. Put it in a common place and remove
-// duplication.
-bool FormatToken::isSimpleTypeSpecifier() const {
-  switch (Tok.getKind()) {
-  case tok::kw_short:
-  case tok::kw_long:
-  case tok::kw___int64:
-  case tok::kw___int128:
-  case tok::kw_signed:
-  case tok::kw_unsigned:
-  case tok::kw_void:
-  case tok::kw_char:
-  case tok::kw_int:
-  case tok::kw_half:
-  case tok::kw_float:
-  case tok::kw_double:
-  case tok::kw___bf16:
-  case tok::kw__Float16:
-  case tok::kw___float128:
-  case tok::kw___ibm128:
-  case tok::kw_wchar_t:
-  case tok::kw_bool:
-#define TRANSFORM_TYPE_TRAIT_DEF(_, Trait) case tok::kw___##Trait:
-#include "clang/Basic/TransformTypeTraits.def"
-  case tok::annot_typename:
-  case tok::kw_char8_t:
-  case tok::kw_char16_t:
-  case tok::kw_char32_t:
-  case tok::kw_typeof:
-  case tok::kw_decltype:
-  case tok::kw__Atomic:
-    return true;
-  default:
-    return false;
-  }
-}
-
 // Sorted common C++ non-keyword types.
 static SmallVector<StringRef> CppNonKeywordTypes = {
     "clock_t",  "int16_t",   "int32_t", "int64_t",   "int8_t",
     "intptr_t", "ptrdiff_t", "size_t",  "time_t",    "uint16_t",
     "uint32_t", "uint64_t",  "uint8_t", "uintptr_t",
 };
 
-bool FormatToken::isTypeName(bool IsCpp) const {
-  return is(TT_TypeName) || isSimpleTypeSpecifier() ||
+bool FormatToken::isTypeName(const LangOptions &LangOpts) const {
+  const bool IsCpp = LangOpts.CXXOperatorNames;
+  return is(TT_TypeName) || Tok.isSimpleTypeSpecifier(LangOpts) ||
          (IsCpp && is(tok::identifier) &&
           std::binary_search(CppNonKeywordTypes.begin(),
                              CppNonKeywordTypes.end(), TokenText));
 }
 
-bool FormatToken::isTypeOrIdentifier(bool IsCpp) const {
-  return isTypeName(IsCpp) || isOneOf(tok::kw_auto, tok::identifier);
+bool FormatToken::isTypeOrIdentifier(const LangOptions &LangOpts) const {
+  return isTypeName(LangOpts) || isOneOf(tok::kw_auto, tok::identifier);
 }
 
 bool FormatToken::isBlockIndentedInitRBrace(const FormatStyle &Style) const {

clang/lib/Format/FormatToken.h

@@ -684,12 +684,8 @@ struct FormatToken {
            isAttribute();
   }
 
-  /// Determine whether the token is a simple-type-specifier.
-  [[nodiscard]] bool isSimpleTypeSpecifier() const;
-
-  [[nodiscard]] bool isTypeName(bool IsCpp) const;
-
-  [[nodiscard]] bool isTypeOrIdentifier(bool IsCpp) const;
+  [[nodiscard]] bool isTypeName(const LangOptions &LangOpts) const;
+  [[nodiscard]] bool isTypeOrIdentifier(const LangOptions &LangOpts) const;
 
   bool isObjCAccessSpecifier() const {
     return is(tok::at) && Next &&

clang/lib/Format/FormatTokenLexer.cpp

@@ -1442,7 +1442,6 @@ void FormatTokenLexer::readRawToken(FormatToken &Tok) {
 
 void FormatTokenLexer::resetLexer(unsigned Offset) {
   StringRef Buffer = SourceMgr.getBufferData(ID);
-  LangOpts = getFormattingLangOpts(Style);
   Lex.reset(new Lexer(SourceMgr.getLocForStartOfFile(ID), LangOpts,
                       Buffer.begin(), Buffer.begin() + Offset, Buffer.end()));
   Lex->SetKeepWhitespaceMode(true);

clang/lib/Format/QualifierAlignmentFixer.cpp

@@ -268,13 +268,11 @@ const FormatToken *LeftRightQualifierAlignmentFixer::analyzeRight(
   if (isPossibleMacro(TypeToken))
     return Tok;
 
-  const bool IsCpp = Style.isCpp();
-
   // The case `const long long int volatile` -> `long long int const volatile`
   // The case `long const long int volatile` -> `long long int const volatile`
   // The case `long long volatile int const` -> `long long int const volatile`
   // The case `const long long volatile int` -> `long long int const volatile`
-  if (TypeToken->isTypeName(IsCpp)) {
+  if (TypeToken->isTypeName(LangOpts)) {
     // The case `const decltype(foo)` -> `const decltype(foo)`
     // The case `const typeof(foo)` -> `const typeof(foo)`
     // The case `const _Atomic(foo)` -> `const _Atomic(foo)`
@@ -283,7 +281,7 @@ const FormatToken *LeftRightQualifierAlignmentFixer::analyzeRight(
 
     const FormatToken *LastSimpleTypeSpecifier = TypeToken;
     while (isQualifierOrType(LastSimpleTypeSpecifier->getNextNonComment(),
-                             IsCpp)) {
+                             LangOpts)) {
       LastSimpleTypeSpecifier = LastSimpleTypeSpecifier->getNextNonComment();
     }
 
@@ -295,7 +293,7 @@ const FormatToken *LeftRightQualifierAlignmentFixer::analyzeRight(
   // The case  `unsigned short const` -> `unsigned short const`
   // The case:
   // `unsigned short volatile const` -> `unsigned short const volatile`
-  if (PreviousCheck && PreviousCheck->isTypeName(IsCpp)) {
+  if (PreviousCheck && PreviousCheck->isTypeName(LangOpts)) {
     if (LastQual != Tok)
       rotateTokens(SourceMgr, Fixes, Tok, LastQual, /*Left=*/false);
     return Tok;
@@ -412,11 +410,11 @@ const FormatToken *LeftRightQualifierAlignmentFixer::analyzeLeft(
   // The case `volatile long long const int` -> `const volatile long long int`
   // The case `const long long volatile int` -> `const volatile long long int`
   // The case `long volatile long int const` -> `const volatile long long int`
-  if (const bool IsCpp = Style.isCpp(); TypeToken->isTypeName(IsCpp)) {
+  if (TypeToken->isTypeName(LangOpts)) {
     const FormatToken *LastSimpleTypeSpecifier = TypeToken;
     while (isConfiguredQualifierOrType(
         LastSimpleTypeSpecifier->getPreviousNonComment(),
-        ConfiguredQualifierTokens, IsCpp)) {
+        ConfiguredQualifierTokens, LangOpts)) {
       LastSimpleTypeSpecifier =
           LastSimpleTypeSpecifier->getPreviousNonComment();
     }
@@ -614,16 +612,16 @@ void prepareLeftRightOrderingForQualifierAlignmentFixer(
   }
 }
 
-bool LeftRightQualifierAlignmentFixer::isQualifierOrType(const FormatToken *Tok,
-                                                         bool IsCpp) {
-  return Tok &&
-         (Tok->isTypeName(IsCpp) || Tok->is(tok::kw_auto) || isQualifier(Tok));
+bool LeftRightQualifierAlignmentFixer::isQualifierOrType(
+    const FormatToken *Tok, const LangOptions &LangOpts) {
+  return Tok && (Tok->isTypeName(LangOpts) || Tok->is(tok::kw_auto) ||
+                 isQualifier(Tok));
 }
 
 bool LeftRightQualifierAlignmentFixer::isConfiguredQualifierOrType(
     const FormatToken *Tok, const std::vector<tok::TokenKind> &Qualifiers,
-    bool IsCpp) {
-  return Tok && (Tok->isTypeName(IsCpp) || Tok->is(tok::kw_auto) ||
+    const LangOptions &LangOpts) {
+  return Tok && (Tok->isTypeName(LangOpts) || Tok->is(tok::kw_auto) ||
                  isConfiguredQualifier(Tok, Qualifiers));
 }
 

clang/lib/Format/QualifierAlignmentFixer.h

@@ -71,11 +71,12 @@ class LeftRightQualifierAlignmentFixer : public TokenAnalyzer {
                                  tok::TokenKind QualifierType);
 
   // Is the Token a simple or qualifier type
-  static bool isQualifierOrType(const FormatToken *Tok, bool IsCpp = true);
+  static bool isQualifierOrType(const FormatToken *Tok,
+                                const LangOptions &LangOpts);
   static bool
   isConfiguredQualifierOrType(const FormatToken *Tok,
                               const std::vector<tok::TokenKind> &Qualifiers,
-                              bool IsCpp = true);
+                              const LangOptions &LangOpts);
 
   // Is the Token likely a Macro
   static bool isPossibleMacro(const FormatToken *Tok);

clang/lib/Format/TokenAnalyzer.cpp

@@ -84,7 +84,7 @@ Environment::Environment(StringRef Code, StringRef FileName,
       NextStartColumn(NextStartColumn), LastStartColumn(LastStartColumn) {}
 
 TokenAnalyzer::TokenAnalyzer(const Environment &Env, const FormatStyle &Style)
-    : Style(Style), Env(Env),
+    : Style(Style), LangOpts(getFormattingLangOpts(Style)), Env(Env),
       AffectedRangeMgr(Env.getSourceManager(), Env.getCharRanges()),
       UnwrappedLines(1),
       Encoding(encoding::detectEncoding(
@@ -101,7 +101,7 @@ std::pair<tooling::Replacements, unsigned>
 TokenAnalyzer::process(bool SkipAnnotation) {
   tooling::Replacements Result;
   llvm::SpecificBumpPtrAllocator<FormatToken> Allocator;
-  IdentifierTable IdentTable(getFormattingLangOpts(Style));
+  IdentifierTable IdentTable(LangOpts);
   FormatTokenLexer Lex(Env.getSourceManager(), Env.getFileID(),
                        Env.getFirstStartColumn(), Style, Encoding, Allocator,
                        IdentTable);

clang/lib/Format/TokenAnalyzer.h

@@ -92,6 +92,7 @@ class TokenAnalyzer : public UnwrappedLineConsumer {
   void finishRun() override;
 
   FormatStyle Style;
+  LangOptions LangOpts;
   // Stores Style, FileID and SourceManager etc.
   const Environment &Env;
   // AffectedRangeMgr stores ranges to be fixed.

clang/lib/Format/TokenAnnotator.cpp

@@ -126,7 +126,9 @@ class AnnotatingParser {
                    const AdditionalKeywords &Keywords,
                    SmallVector<ScopeType> &Scopes)
       : Style(Style), Line(Line), CurrentToken(Line.First), AutoFound(false),
-        IsCpp(Style.isCpp()), Keywords(Keywords), Scopes(Scopes) {
+        IsCpp(Style.isCpp()), LangOpts(getFormattingLangOpts(Style)),
+        Keywords(Keywords), Scopes(Scopes) {
+    assert(IsCpp == LangOpts.CXXOperatorNames);
     Contexts.push_back(Context(tok::unknown, 1, /*IsExpression=*/false));
     resetTokenMetadata();
   }
@@ -562,7 +564,7 @@ class AnnotatingParser {
           (CurrentToken->is(tok::l_paren) && CurrentToken->Next &&
            CurrentToken->Next->isOneOf(tok::star, tok::amp, tok::caret));
       if ((CurrentToken->Previous->isOneOf(tok::kw_const, tok::kw_auto) ||
-           CurrentToken->Previous->isTypeName(IsCpp)) &&
+           CurrentToken->Previous->isTypeName(LangOpts)) &&
           !(CurrentToken->is(tok::l_brace) ||
             (CurrentToken->is(tok::l_paren) && !ProbablyFunctionTypeLParen))) {
         Contexts.back().IsExpression = false;
@@ -2624,7 +2626,7 @@ class AnnotatingParser {
       return true;
 
     // MyClass a;
-    if (PreviousNotConst->isTypeName(IsCpp))
+    if (PreviousNotConst->isTypeName(LangOpts))
       return true;
 
     // type[] a in Java
@@ -2728,7 +2730,7 @@ class AnnotatingParser {
     }
 
     if (Tok.Next->is(tok::question) ||
-        (Tok.Next->is(tok::ampamp) && !Tok.Previous->isTypeName(IsCpp))) {
+        (Tok.Next->is(tok::ampamp) && !Tok.Previous->isTypeName(LangOpts))) {
       return false;
     }
 
@@ -2757,9 +2759,10 @@ class AnnotatingParser {
     }
 
     // Heuristically try to determine whether the parentheses contain a type.
-    auto IsQualifiedPointerOrReference = [](FormatToken *T, bool IsCpp) {
+    auto IsQualifiedPointerOrReference = [](FormatToken *T,
+                                            const LangOptions &LangOpts) {
       // This is used to handle cases such as x = (foo *const)&y;
-      assert(!T->isTypeName(IsCpp) && "Should have already been checked");
+      assert(!T->isTypeName(LangOpts) && "Should have already been checked");
       // Strip trailing qualifiers such as const or volatile when checking
       // whether the parens could be a cast to a pointer/reference type.
       while (T) {
@@ -2791,8 +2794,8 @@ class AnnotatingParser {
     bool ParensAreType =
         !Tok.Previous ||
         Tok.Previous->isOneOf(TT_TemplateCloser, TT_TypeDeclarationParen) ||
-        Tok.Previous->isTypeName(IsCpp) ||
-        IsQualifiedPointerOrReference(Tok.Previous, IsCpp);
+        Tok.Previous->isTypeName(LangOpts) ||
+        IsQualifiedPointerOrReference(Tok.Previous, LangOpts);
     bool ParensCouldEndDecl =
         Tok.Next->isOneOf(tok::equal, tok::semi, tok::l_brace, tok::greater);
     if (ParensAreType && !ParensCouldEndDecl)
@@ -3065,6 +3068,7 @@ class AnnotatingParser {
   FormatToken *CurrentToken;
   bool AutoFound;
   bool IsCpp;
+  LangOptions LangOpts;
   const AdditionalKeywords &Keywords;
 
   SmallVector<ScopeType> &Scopes;
@@ -3639,7 +3643,8 @@ void TokenAnnotator::annotate(AnnotatedLine &Line) {
 
 // This function heuristically determines whether 'Current' starts the name of a
 // function declaration.
-static bool isFunctionDeclarationName(bool IsCpp, const FormatToken &Current,
+static bool isFunctionDeclarationName(const LangOptions &LangOpts,
+                                      const FormatToken &Current,
                                       const AnnotatedLine &Line,
                                       FormatToken *&ClosingParen) {
   assert(Current.Previous);
@@ -3658,7 +3663,7 @@ static bool isFunctionDeclarationName(bool IsCpp, const FormatToken &Current,
   }
 
   auto skipOperatorName =
-      [IsCpp](const FormatToken *Next) -> const FormatToken * {
+      [&LangOpts](const FormatToken *Next) -> const FormatToken * {
     for (; Next; Next = Next->Next) {
       if (Next->is(TT_OverloadedOperatorLParen))
         return Next;
@@ -3677,7 +3682,7 @@ static bool isFunctionDeclarationName(bool IsCpp, const FormatToken &Current,
         Next = Next->Next;
         continue;
       }
-      if ((Next->isTypeName(IsCpp) || Next->is(tok::identifier)) &&
+      if ((Next->isTypeName(LangOpts) || Next->is(tok::identifier)) &&
           Next->Next && Next->Next->isPointerOrReference()) {
         // For operator void*(), operator char*(), operator Foo*().
         Next = Next->Next;
@@ -3693,8 +3698,10 @@ static bool isFunctionDeclarationName(bool IsCpp, const FormatToken &Current,
     return nullptr;
   };
 
+  const auto *Next = Current.Next;
+  const bool IsCpp = LangOpts.CXXOperatorNames;
+
   // Find parentheses of parameter list.
-  const FormatToken *Next = Current.Next;
   if (Current.is(tok::kw_operator)) {
     if (Previous.Tok.getIdentifierInfo() &&
         !Previous.isOneOf(tok::kw_return, tok::kw_co_return)) {
@@ -3774,7 +3781,7 @@ static bool isFunctionDeclarationName(bool IsCpp, const FormatToken &Current,
       Tok = Tok->MatchingParen;
       continue;
     }
-    if (Tok->is(tok::kw_const) || Tok->isTypeName(IsCpp) ||
+    if (Tok->is(tok::kw_const) || Tok->isTypeName(LangOpts) ||
         Tok->isOneOf(TT_PointerOrReference, TT_StartOfName, tok::ellipsis)) {
       return true;
     }
@@ -3837,7 +3844,7 @@ void TokenAnnotator::calculateFormattingInformation(AnnotatedLine &Line) const {
       AfterLastAttribute = Tok;
     if (const bool IsCtorOrDtor = Tok->is(TT_CtorDtorDeclName);
         IsCtorOrDtor ||
-        isFunctionDeclarationName(IsCpp, *Tok, Line, ClosingParen)) {
+        isFunctionDeclarationName(LangOpts, *Tok, Line, ClosingParen)) {
       if (!IsCtorOrDtor)
         Tok->setFinalizedType(TT_FunctionDeclarationName);
       LineIsFunctionDeclaration = true;
@@ -4447,7 +4454,7 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
     if (Left.Tok.isLiteral())
       return true;
     // for (auto a = 0, b = 0; const auto & c : {1, 2, 3})
-    if (Left.isTypeOrIdentifier(IsCpp) && Right.Next && Right.Next->Next &&
+    if (Left.isTypeOrIdentifier(LangOpts) && Right.Next && Right.Next->Next &&
         Right.Next->Next->is(TT_RangeBasedForLoopColon)) {
       return getTokenPointerOrReferenceAlignment(Right) !=
              FormatStyle::PAS_Left;
@@ -4490,7 +4497,7 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
     if (Right.is(tok::l_brace) && Right.is(BK_Block))
       return true;
     // for (auto a = 0, b = 0; const auto& c : {1, 2, 3})
-    if (BeforeLeft && BeforeLeft->isTypeOrIdentifier(IsCpp) && Right.Next &&
+    if (BeforeLeft && BeforeLeft->isTypeOrIdentifier(LangOpts) && Right.Next &&
         Right.Next->is(TT_RangeBasedForLoopColon)) {
       return getTokenPointerOrReferenceAlignment(Left) !=
              FormatStyle::PAS_Right;
@@ -4534,7 +4541,7 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
   if (Right.isPointerOrReference()) {
     const FormatToken *Previous = &Left;
     while (Previous && Previous->isNot(tok::kw_operator)) {
-      if (Previous->is(tok::identifier) || Previous->isTypeName(IsCpp)) {
+      if (Previous->is(tok::identifier) || Previous->isTypeName(LangOpts)) {
         Previous = Previous->getPreviousNonComment();
         continue;
       }
@@ -4723,7 +4730,7 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
   if (!Style.isVerilog() &&
       (Left.isOneOf(tok::identifier, tok::greater, tok::r_square,
                     tok::r_paren) ||
-       Left.isTypeName(IsCpp)) &&
+       Left.isTypeName(LangOpts)) &&
       Right.is(tok::l_brace) && Right.getNextNonComment() &&
       Right.isNot(BK_Block)) {
     return false;

clang/lib/Format/TokenAnnotator.h

@@ -211,7 +211,10 @@ class AnnotatedLine {
 class TokenAnnotator {
 public:
   TokenAnnotator(const FormatStyle &Style, const AdditionalKeywords &Keywords)
-      : Style(Style), IsCpp(Style.isCpp()), Keywords(Keywords) {}
+      : Style(Style), IsCpp(Style.isCpp()),
+        LangOpts(getFormattingLangOpts(Style)), Keywords(Keywords) {
+    assert(IsCpp == LangOpts.CXXOperatorNames);
+  }
 
   /// Adapts the indent levels of comment lines to the indent of the
   /// subsequent line.
@@ -260,6 +263,7 @@ class TokenAnnotator {
   const FormatStyle &Style;
 
   bool IsCpp;
+  LangOptions LangOpts;
 
   const AdditionalKeywords &Keywords;