clang/lib/Format/FormatToken.h

   1 //===--- FormatToken.h - Format C++ code ------------------------*- 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 /// \file
  10 /// This file contains the declaration of the FormatToken, a wrapper
  11 /// around Token with additional information related to formatting.
  12 ///
  13 //===----------------------------------------------------------------------===//
  14
  15 #ifndef LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H
  16 #define LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H
  17
  18 #include "clang/Basic/IdentifierTable.h"
  19 #include "clang/Basic/OperatorPrecedence.h"
  20 #include "clang/Format/Format.h"
  21 #include "clang/Lex/Lexer.h"
  22 #include <memory>
  23 #include <optional>
  24 #include <unordered_set>
  25
  26 namespace clang {
  27 namespace format {
  28
  29 #define LIST_TOKEN_TYPES                                                       \
  30   TYPE(ArrayInitializerLSquare)                                                \
  31   TYPE(ArraySubscriptLSquare)                                                  \
  32   TYPE(AttributeColon)                                                         \
  33   TYPE(AttributeLParen)                                                        \
  34   TYPE(AttributeMacro)                                                         \
  35   TYPE(AttributeRParen)                                                        \
  36   TYPE(AttributeSquare)                                                        \
  37   TYPE(BinaryOperator)                                                         \
  38   TYPE(BitFieldColon)                                                          \
  39   TYPE(BlockComment)                                                           \
  40   TYPE(BracedListLBrace)                                                       \
  41   /* The colon at the end of a case label. */                                  \
  42   TYPE(CaseLabelColon)                                                         \
  43   TYPE(CastRParen)                                                             \
  44   TYPE(ClassLBrace)                                                            \
  45   TYPE(ClassRBrace)                                                            \
  46   /* ternary ?: expression */                                                  \
  47   TYPE(ConditionalExpr)                                                        \
  48   /* the condition in an if statement */                                       \
  49   TYPE(ConditionLParen)                                                        \
  50   TYPE(ConflictAlternative)                                                    \
  51   TYPE(ConflictEnd)                                                            \
  52   TYPE(ConflictStart)                                                          \
  53   /* l_brace of if/for/while */                                                \
  54   TYPE(ControlStatementLBrace)                                                 \
  55   TYPE(ControlStatementRBrace)                                                 \
  56   TYPE(CppCastLParen)                                                          \
  57   TYPE(CSharpGenericTypeConstraint)                                            \
  58   TYPE(CSharpGenericTypeConstraintColon)                                       \
  59   TYPE(CSharpGenericTypeConstraintComma)                                       \
  60   TYPE(CSharpNamedArgumentColon)                                               \
  61   TYPE(CSharpNullable)                                                         \
  62   TYPE(CSharpNullConditionalLSquare)                                           \
  63   TYPE(CSharpStringLiteral)                                                    \
  64   TYPE(CtorInitializerColon)                                                   \
  65   TYPE(CtorInitializerComma)                                                   \
  66   TYPE(CtorDtorDeclName)                                                       \
  67   TYPE(DesignatedInitializerLSquare)                                           \
  68   TYPE(DesignatedInitializerPeriod)                                            \
  69   TYPE(DictLiteral)                                                            \
  70   TYPE(DoWhile)                                                                \
  71   TYPE(ElseLBrace)                                                             \
  72   TYPE(ElseRBrace)                                                             \
  73   TYPE(EnumLBrace)                                                             \
  74   TYPE(EnumRBrace)                                                             \
  75   TYPE(FatArrow)                                                               \
  76   TYPE(ForEachMacro)                                                           \
  77   TYPE(FunctionAnnotationRParen)                                               \
  78   TYPE(FunctionDeclarationName)                                                \
  79   TYPE(FunctionLBrace)                                                         \
  80   TYPE(FunctionLikeOrFreestandingMacro)                                        \
  81   TYPE(FunctionTypeLParen)                                                     \
  82   /* The colons as part of a C11 _Generic selection */                         \
  83   TYPE(GenericSelectionColon)                                                  \
  84   /* The colon at the end of a goto label. */                                  \
  85   TYPE(GotoLabelColon)                                                         \
  86   TYPE(IfMacro)                                                                \
  87   TYPE(ImplicitStringLiteral)                                                  \
  88   TYPE(InheritanceColon)                                                       \
  89   TYPE(InheritanceComma)                                                       \
  90   TYPE(InlineASMBrace)                                                         \
  91   TYPE(InlineASMColon)                                                         \
  92   TYPE(InlineASMSymbolicNameLSquare)                                           \
  93   TYPE(JavaAnnotation)                                                         \
  94   TYPE(JsAndAndEqual)                                                          \
  95   TYPE(JsComputedPropertyName)                                                 \
  96   TYPE(JsExponentiation)                                                       \
  97   TYPE(JsExponentiationEqual)                                                  \
  98   TYPE(JsPipePipeEqual)                                                        \
  99   TYPE(JsPrivateIdentifier)                                                    \
 100   TYPE(JsTypeColon)                                                            \
 101   TYPE(JsTypeOperator)                                                         \
 102   TYPE(JsTypeOptionalQuestion)                                                 \
 103   TYPE(LambdaLBrace)                                                           \
 104   TYPE(LambdaLSquare)                                                          \
 105   TYPE(LeadingJavaAnnotation)                                                  \
 106   TYPE(LineComment)                                                            \
 107   TYPE(MacroBlockBegin)                                                        \
 108   TYPE(MacroBlockEnd)                                                          \
 109   TYPE(ModulePartitionColon)                                                   \
 110   TYPE(NamespaceLBrace)                                                        \
 111   TYPE(NamespaceMacro)                                                         \
 112   TYPE(NamespaceRBrace)                                                        \
 113   TYPE(NonNullAssertion)                                                       \
 114   TYPE(NullCoalescingEqual)                                                    \
 115   TYPE(NullCoalescingOperator)                                                 \
 116   TYPE(NullPropagatingOperator)                                                \
 117   TYPE(ObjCBlockLBrace)                                                        \
 118   TYPE(ObjCBlockLParen)                                                        \
 119   TYPE(ObjCDecl)                                                               \
 120   TYPE(ObjCForIn)                                                              \
 121   TYPE(ObjCMethodExpr)                                                         \
 122   TYPE(ObjCMethodSpecifier)                                                    \
 123   TYPE(ObjCProperty)                                                           \
 124   TYPE(ObjCStringLiteral)                                                      \
 125   TYPE(OverloadedOperator)                                                     \
 126   TYPE(OverloadedOperatorLParen)                                               \
 127   TYPE(PointerOrReference)                                                     \
 128   TYPE(ProtoExtensionLSquare)                                                  \
 129   TYPE(PureVirtualSpecifier)                                                   \
 130   TYPE(RangeBasedForLoopColon)                                                 \
 131   TYPE(RecordLBrace)                                                           \
 132   TYPE(RecordRBrace)                                                           \
 133   TYPE(RegexLiteral)                                                           \
 134   TYPE(RequiresClause)                                                         \
 135   TYPE(RequiresClauseInARequiresExpression)                                    \
 136   TYPE(RequiresExpression)                                                     \
 137   TYPE(RequiresExpressionLBrace)                                               \
 138   TYPE(RequiresExpressionLParen)                                               \
 139   TYPE(SelectorName)                                                           \
 140   TYPE(StartOfName)                                                            \
 141   TYPE(StatementAttributeLikeMacro)                                            \
 142   TYPE(StatementMacro)                                                         \
 143   /* A string that is part of a string concatenation. For C#, JavaScript, and  \
 144    * Java, it is used for marking whether a string needs parentheses around it \
 145    * if it is to be split into parts joined by `+`. For Verilog, whether       \
 146    * braces need to be added to split it. Not used for other languages. */     \
 147   TYPE(StringInConcatenation)                                                  \
 148   TYPE(StructLBrace)                                                           \
 149   TYPE(StructRBrace)                                                           \
 150   TYPE(StructuredBindingLSquare)                                               \
 151   TYPE(TemplateCloser)                                                         \
 152   TYPE(TemplateOpener)                                                         \
 153   TYPE(TemplateString)                                                         \
 154   TYPE(TrailingAnnotation)                                                     \
 155   TYPE(TrailingReturnArrow)                                                    \
 156   TYPE(TrailingUnaryOperator)                                                  \
 157   TYPE(TypeDeclarationParen)                                                   \
 158   TYPE(TypeName)                                                               \
 159   TYPE(TypenameMacro)                                                          \
 160   TYPE(UnaryOperator)                                                          \
 161   TYPE(UnionLBrace)                                                            \
 162   TYPE(UnionRBrace)                                                            \
 163   TYPE(UntouchableMacroFunc)                                                   \
 164   /* Like in 'assign x = 0, y = 1;' . */                                       \
 165   TYPE(VerilogAssignComma)                                                     \
 166   /* like in begin : block */                                                  \
 167   TYPE(VerilogBlockLabelColon)                                                 \
 168   /* The square bracket for the dimension part of the type name.               \
 169    * In 'logic [1:0] x[1:0]', only the first '['. This way we can have space   \
 170    * before the first bracket but not the second. */                           \
 171   TYPE(VerilogDimensionedTypeName)                                             \
 172   /* list of port connections or parameters in a module instantiation */       \
 173   TYPE(VerilogInstancePortComma)                                               \
 174   TYPE(VerilogInstancePortLParen)                                              \
 175   /* A parenthesized list within which line breaks are inserted by the         \
 176    * formatter, for example the list of ports in a module header. */           \
 177   TYPE(VerilogMultiLineListLParen)                                             \
 178   /* for the base in a number literal, not including the quote */              \
 179   TYPE(VerilogNumberBase)                                                      \
 180   /* like `(strong1, pull0)` */                                                \
 181   TYPE(VerilogStrength)                                                        \
 182   /* Things inside the table in user-defined primitives. */                    \
 183   TYPE(VerilogTableItem)                                                       \
 184   /* those that separate ports of different types */                           \
 185   TYPE(VerilogTypeComma)                                                       \
 186   TYPE(Unknown)
 187
 188 /// Determines the semantic type of a syntactic token, e.g. whether "<" is a
 189 /// template opener or binary operator.
 190 enum TokenType : uint8_t {
 191 #define TYPE(X) TT_##X,
 192   LIST_TOKEN_TYPES
 193 #undef TYPE
 194       NUM_TOKEN_TYPES
 195 };
 196
 197 /// Determines the name of a token type.
 198 const char *getTokenTypeName(TokenType Type);
 199
 200 // Represents what type of block a set of braces open.
 201 enum BraceBlockKind { BK_Unknown, BK_Block, BK_BracedInit };
 202
 203 // The packing kind of a function's parameters.
 204 enum ParameterPackingKind { PPK_BinPacked, PPK_OnePerLine, PPK_Inconclusive };
 205
 206 enum FormatDecision { FD_Unformatted, FD_Continue, FD_Break };
 207
 208 /// Roles a token can take in a configured macro expansion.
 209 enum MacroRole {
 210   /// The token was expanded from a macro argument when formatting the expanded
 211   /// token sequence.
 212   MR_ExpandedArg,
 213   /// The token is part of a macro argument that was previously formatted as
 214   /// expansion when formatting the unexpanded macro call.
 215   MR_UnexpandedArg,
 216   /// The token was expanded from a macro definition, and is not visible as part
 217   /// of the macro call.
 218   MR_Hidden,
 219 };
 220
 221 struct FormatToken;
 222
 223 /// Contains information on the token's role in a macro expansion.
 224 ///
 225 /// Given the following definitions:
 226 /// A(X) = [ X ]
 227 /// B(X) = < X >
 228 /// C(X) = X
 229 ///
 230 /// Consider the macro call:
 231 /// A({B(C(C(x)))}) -> [{<x>}]
 232 ///
 233 /// In this case, the tokens of the unexpanded macro call will have the
 234 /// following relevant entries in their macro context (note that formatting
 235 /// the unexpanded macro call happens *after* formatting the expanded macro
 236 /// call):
 237 ///                   A( { B( C( C(x) ) ) } )
 238 /// Role:             NN U NN NN NNUN N N U N  (N=None, U=UnexpandedArg)
 239 ///
 240 ///                   [  { <       x    > } ]
 241 /// Role:             H  E H       E    H E H  (H=Hidden, E=ExpandedArg)
 242 /// ExpandedFrom[0]:  A  A A       A    A A A
 243 /// ExpandedFrom[1]:       B       B    B
 244 /// ExpandedFrom[2]:               C
 245 /// ExpandedFrom[3]:               C
 246 /// StartOfExpansion: 1  0 1       2    0 0 0
 247 /// EndOfExpansion:   0  0 0       2    1 0 1
 248 struct MacroExpansion {
 249   MacroExpansion(MacroRole Role) : Role(Role) {}
 250
 251   /// The token's role in the macro expansion.
 252   /// When formatting an expanded macro, all tokens that are part of macro
 253   /// arguments will be MR_ExpandedArg, while all tokens that are not visible in
 254   /// the macro call will be MR_Hidden.
 255   /// When formatting an unexpanded macro call, all tokens that are part of
 256   /// macro arguments will be MR_UnexpandedArg.
 257   MacroRole Role;
 258
 259   /// The stack of macro call identifier tokens this token was expanded from.
 260   llvm::SmallVector<FormatToken *, 1> ExpandedFrom;
 261
 262   /// The number of expansions of which this macro is the first entry.
 263   unsigned StartOfExpansion = 0;
 264
 265   /// The number of currently open expansions in \c ExpandedFrom this macro is
 266   /// the last token in.
 267   unsigned EndOfExpansion = 0;
 268 };
 269
 270 class TokenRole;
 271 class AnnotatedLine;
 272
 273 /// A wrapper around a \c Token storing information about the
 274 /// whitespace characters preceding it.
 275 struct FormatToken {
 276   FormatToken()
 277       : HasUnescapedNewline(false), IsMultiline(false), IsFirst(false),
 278         MustBreakBefore(false), IsUnterminatedLiteral(false),
 279         CanBreakBefore(false), ClosesTemplateDeclaration(false),
 280         StartsBinaryExpression(false), EndsBinaryExpression(false),
 281         PartOfMultiVariableDeclStmt(false), ContinuesLineCommentSection(false),
 282         Finalized(false), ClosesRequiresClause(false),
 283         EndsCppAttributeGroup(false), BlockKind(BK_Unknown),
 284         Decision(FD_Unformatted), PackingKind(PPK_Inconclusive),
 285         TypeIsFinalized(false), Type(TT_Unknown) {}
 286
 287   /// The \c Token.
 288   Token Tok;
 289
 290   /// The raw text of the token.
 291   ///
 292   /// Contains the raw token text without leading whitespace and without leading
 293   /// escaped newlines.
 294   StringRef TokenText;
 295
 296   /// A token can have a special role that can carry extra information
 297   /// about the token's formatting.
 298   /// FIXME: Make FormatToken for parsing and AnnotatedToken two different
 299   /// classes and make this a unique_ptr in the AnnotatedToken class.
 300   std::shared_ptr<TokenRole> Role;
 301
 302   /// The range of the whitespace immediately preceding the \c Token.
 303   SourceRange WhitespaceRange;
 304
 305   /// Whether there is at least one unescaped newline before the \c
 306   /// Token.
 307   unsigned HasUnescapedNewline : 1;
 308
 309   /// Whether the token text contains newlines (escaped or not).
 310   unsigned IsMultiline : 1;
 311
 312   /// Indicates that this is the first token of the file.
 313   unsigned IsFirst : 1;
 314
 315   /// Whether there must be a line break before this token.
 316   ///
 317   /// This happens for example when a preprocessor directive ended directly
 318   /// before the token.
 319   unsigned MustBreakBefore : 1;
 320
 321   /// Set to \c true if this token is an unterminated literal.
 322   unsigned IsUnterminatedLiteral : 1;
 323
 324   /// \c true if it is allowed to break before this token.
 325   unsigned CanBreakBefore : 1;
 326
 327   /// \c true if this is the ">" of "template<..>".
 328   unsigned ClosesTemplateDeclaration : 1;
 329
 330   /// \c true if this token starts a binary expression, i.e. has at least
 331   /// one fake l_paren with a precedence greater than prec::Unknown.
 332   unsigned StartsBinaryExpression : 1;
 333   /// \c true if this token ends a binary expression.
 334   unsigned EndsBinaryExpression : 1;
 335
 336   /// Is this token part of a \c DeclStmt defining multiple variables?
 337   ///
 338   /// Only set if \c Type == \c TT_StartOfName.
 339   unsigned PartOfMultiVariableDeclStmt : 1;
 340
 341   /// Does this line comment continue a line comment section?
 342   ///
 343   /// Only set to true if \c Type == \c TT_LineComment.
 344   unsigned ContinuesLineCommentSection : 1;
 345
 346   /// If \c true, this token has been fully formatted (indented and
 347   /// potentially re-formatted inside), and we do not allow further formatting
 348   /// changes.
 349   unsigned Finalized : 1;
 350
 351   /// \c true if this is the last token within requires clause.
 352   unsigned ClosesRequiresClause : 1;
 353
 354   /// \c true if this token ends a group of C++ attributes.
 355   unsigned EndsCppAttributeGroup : 1;
 356
 357 private:
 358   /// Contains the kind of block if this token is a brace.
 359   unsigned BlockKind : 2;
 360
 361 public:
 362   BraceBlockKind getBlockKind() const {
 363     return static_cast<BraceBlockKind>(BlockKind);
 364   }
 365   void setBlockKind(BraceBlockKind BBK) {
 366     BlockKind = BBK;
 367     assert(getBlockKind() == BBK && "BraceBlockKind overflow!");
 368   }
 369
 370 private:
 371   /// Stores the formatting decision for the token once it was made.
 372   unsigned Decision : 2;
 373
 374 public:
 375   FormatDecision getDecision() const {
 376     return static_cast<FormatDecision>(Decision);
 377   }
 378   void setDecision(FormatDecision D) {
 379     Decision = D;
 380     assert(getDecision() == D && "FormatDecision overflow!");
 381   }
 382
 383 private:
 384   /// If this is an opening parenthesis, how are the parameters packed?
 385   unsigned PackingKind : 2;
 386
 387 public:
 388   ParameterPackingKind getPackingKind() const {
 389     return static_cast<ParameterPackingKind>(PackingKind);
 390   }
 391   void setPackingKind(ParameterPackingKind K) {
 392     PackingKind = K;
 393     assert(getPackingKind() == K && "ParameterPackingKind overflow!");
 394   }
 395
 396 private:
 397   unsigned TypeIsFinalized : 1;
 398   TokenType Type;
 399
 400 public:
 401   /// Returns the token's type, e.g. whether "<" is a template opener or
 402   /// binary operator.
 403   TokenType getType() const { return Type; }
 404   void setType(TokenType T) {
 405     // If this token is a macro argument while formatting an unexpanded macro
 406     // call, we do not change its type any more - the type was deduced from
 407     // formatting the expanded macro stream already.
 408     if (MacroCtx && MacroCtx->Role == MR_UnexpandedArg)
 409       return;
 410     assert((!TypeIsFinalized || T == Type) &&
 411            "Please use overwriteFixedType to change a fixed type.");
 412     Type = T;
 413   }
 414   /// Sets the type and also the finalized flag. This prevents the type to be
 415   /// reset in TokenAnnotator::resetTokenMetadata(). If the type needs to be set
 416   /// to another one please use overwriteFixedType, or even better remove the
 417   /// need to reassign the type.
 418   void setFinalizedType(TokenType T) {
 419     Type = T;
 420     TypeIsFinalized = true;
 421   }
 422   void overwriteFixedType(TokenType T) {
 423     TypeIsFinalized = false;
 424     setType(T);
 425   }
 426   bool isTypeFinalized() const { return TypeIsFinalized; }
 427
 428   /// Used to set an operator precedence explicitly.
 429   prec::Level ForcedPrecedence = prec::Unknown;
 430
 431   /// The number of newlines immediately before the \c Token.
 432   ///
 433   /// This can be used to determine what the user wrote in the original code
 434   /// and thereby e.g. leave an empty line between two function definitions.
 435   unsigned NewlinesBefore = 0;
 436
 437   /// The number of newlines immediately before the \c Token after formatting.
 438   ///
 439   /// This is used to avoid overlapping whitespace replacements when \c Newlines
 440   /// is recomputed for a finalized preprocessor branching directive.
 441   int Newlines = -1;
 442
 443   /// The offset just past the last '\n' in this token's leading
 444   /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'.
 445   unsigned LastNewlineOffset = 0;
 446
 447   /// The width of the non-whitespace parts of the token (or its first
 448   /// line for multi-line tokens) in columns.
 449   /// We need this to correctly measure number of columns a token spans.
 450   unsigned ColumnWidth = 0;
 451
 452   /// Contains the width in columns of the last line of a multi-line
 453   /// token.
 454   unsigned LastLineColumnWidth = 0;
 455
 456   /// The number of spaces that should be inserted before this token.
 457   unsigned SpacesRequiredBefore = 0;
 458
 459   /// Number of parameters, if this is "(", "[" or "<".
 460   unsigned ParameterCount = 0;
 461
 462   /// Number of parameters that are nested blocks,
 463   /// if this is "(", "[" or "<".
 464   unsigned BlockParameterCount = 0;
 465
 466   /// If this is a bracket ("<", "(", "[" or "{"), contains the kind of
 467   /// the surrounding bracket.
 468   tok::TokenKind ParentBracket = tok::unknown;
 469
 470   /// The total length of the unwrapped line up to and including this
 471   /// token.
 472   unsigned TotalLength = 0;
 473
 474   /// The original 0-based column of this token, including expanded tabs.
 475   /// The configured TabWidth is used as tab width.
 476   unsigned OriginalColumn = 0;
 477
 478   /// The length of following tokens until the next natural split point,
 479   /// or the next token that can be broken.
 480   unsigned UnbreakableTailLength = 0;
 481
 482   // FIXME: Come up with a 'cleaner' concept.
 483   /// The binding strength of a token. This is a combined value of
 484   /// operator precedence, parenthesis nesting, etc.
 485   unsigned BindingStrength = 0;
 486
 487   /// The nesting level of this token, i.e. the number of surrounding (),
 488   /// [], {} or <>.
 489   unsigned NestingLevel = 0;
 490
 491   /// The indent level of this token. Copied from the surrounding line.
 492   unsigned IndentLevel = 0;
 493
 494   /// Penalty for inserting a line break before this token.
 495   unsigned SplitPenalty = 0;
 496
 497   /// If this is the first ObjC selector name in an ObjC method
 498   /// definition or call, this contains the length of the longest name.
 499   ///
 500   /// This being set to 0 means that the selectors should not be colon-aligned,
 501   /// e.g. because several of them are block-type.
 502   unsigned LongestObjCSelectorName = 0;
 503
 504   /// If this is the first ObjC selector name in an ObjC method
 505   /// definition or call, this contains the number of parts that the whole
 506   /// selector consist of.
 507   unsigned ObjCSelectorNameParts = 0;
 508
 509   /// The 0-based index of the parameter/argument. For ObjC it is set
 510   /// for the selector name token.
 511   /// For now calculated only for ObjC.
 512   unsigned ParameterIndex = 0;
 513
 514   /// Stores the number of required fake parentheses and the
 515   /// corresponding operator precedence.
 516   ///
 517   /// If multiple fake parentheses start at a token, this vector stores them in
 518   /// reverse order, i.e. inner fake parenthesis first.
 519   SmallVector<prec::Level, 4> FakeLParens;
 520   /// Insert this many fake ) after this token for correct indentation.
 521   unsigned FakeRParens = 0;
 522
 523   /// If this is an operator (or "."/"->") in a sequence of operators
 524   /// with the same precedence, contains the 0-based operator index.
 525   unsigned OperatorIndex = 0;
 526
 527   /// If this is an operator (or "."/"->") in a sequence of operators
 528   /// with the same precedence, points to the next operator.
 529   FormatToken *NextOperator = nullptr;
 530
 531   /// If this is a bracket, this points to the matching one.
 532   FormatToken *MatchingParen = nullptr;
 533
 534   /// The previous token in the unwrapped line.
 535   FormatToken *Previous = nullptr;
 536
 537   /// The next token in the unwrapped line.
 538   FormatToken *Next = nullptr;
 539
 540   /// The first token in set of column elements.
 541   bool StartsColumn = false;
 542
 543   /// This notes the start of the line of an array initializer.
 544   bool ArrayInitializerLineStart = false;
 545
 546   /// This starts an array initializer.
 547   bool IsArrayInitializer = false;
 548
 549   /// Is optional and can be removed.
 550   bool Optional = false;
 551
 552   /// Number of optional braces to be inserted after this token:
 553   ///   -1: a single left brace
 554   ///    0: no braces
 555   ///   >0: number of right braces
 556   int8_t BraceCount = 0;
 557
 558   /// If this token starts a block, this contains all the unwrapped lines
 559   /// in it.
 560   SmallVector<AnnotatedLine *, 1> Children;
 561
 562   // Contains all attributes related to how this token takes part
 563   // in a configured macro expansion.
 564   std::optional<MacroExpansion> MacroCtx;
 565
 566   /// When macro expansion introduces nodes with children, those are marked as
 567   /// \c MacroParent.
 568   /// FIXME: The formatting code currently hard-codes the assumption that
 569   /// child nodes are introduced by blocks following an opening brace.
 570   /// This is deeply baked into the code and disentangling this will require
 571   /// signficant refactorings. \c MacroParent allows us to special-case the
 572   /// cases in which we treat parents as block-openers for now.
 573   bool MacroParent = false;
 574
 575   bool is(tok::TokenKind Kind) const { return Tok.is(Kind); }
 576   bool is(TokenType TT) const { return getType() == TT; }
 577   bool is(const IdentifierInfo *II) const {
 578     return II && II == Tok.getIdentifierInfo();
 579   }
 580   bool is(tok::PPKeywordKind Kind) const {
 581     return Tok.getIdentifierInfo() &&
 582            Tok.getIdentifierInfo()->getPPKeywordID() == Kind;
 583   }
 584   bool is(BraceBlockKind BBK) const { return getBlockKind() == BBK; }
 585   bool is(ParameterPackingKind PPK) const { return getPackingKind() == PPK; }
 586
 587   template <typename A, typename B> bool isOneOf(A K1, B K2) const {
 588     return is(K1) || is(K2);
 589   }
 590   template <typename A, typename B, typename... Ts>
 591   bool isOneOf(A K1, B K2, Ts... Ks) const {
 592     return is(K1) || isOneOf(K2, Ks...);
 593   }
 594   template <typename T> bool isNot(T Kind) const { return !is(Kind); }
 595
 596   bool isIf(bool AllowConstexprMacro = true) const {
 597     return is(tok::kw_if) || endsSequence(tok::kw_constexpr, tok::kw_if) ||
 598            (endsSequence(tok::identifier, tok::kw_if) && AllowConstexprMacro);
 599   }
 600
 601   bool closesScopeAfterBlock() const {
 602     if (getBlockKind() == BK_Block)
 603       return true;
 604     if (closesScope())
 605       return Previous->closesScopeAfterBlock();
 606     return false;
 607   }
 608
 609   /// \c true if this token starts a sequence with the given tokens in order,
 610   /// following the ``Next`` pointers, ignoring comments.
 611   template <typename A, typename... Ts>
 612   bool startsSequence(A K1, Ts... Tokens) const {
 613     return startsSequenceInternal(K1, Tokens...);
 614   }
 615
 616   /// \c true if this token ends a sequence with the given tokens in order,
 617   /// following the ``Previous`` pointers, ignoring comments.
 618   /// For example, given tokens [T1, T2, T3], the function returns true if
 619   /// 3 tokens ending at this (ignoring comments) are [T3, T2, T1]. In other
 620   /// words, the tokens passed to this function need to the reverse of the
 621   /// order the tokens appear in code.
 622   template <typename A, typename... Ts>
 623   bool endsSequence(A K1, Ts... Tokens) const {
 624     return endsSequenceInternal(K1, Tokens...);
 625   }
 626
 627   bool isStringLiteral() const { return tok::isStringLiteral(Tok.getKind()); }
 628
 629   bool isAttribute() const {
 630     return isOneOf(tok::kw___attribute, tok::kw___declspec, TT_AttributeMacro);
 631   }
 632
 633   bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const {
 634     return Tok.isObjCAtKeyword(Kind);
 635   }
 636
 637   bool isAccessSpecifier(bool ColonRequired = true) const {
 638     if (!isOneOf(tok::kw_public, tok::kw_protected, tok::kw_private))
 639       return false;
 640     if (!ColonRequired)
 641       return true;
 642     const auto NextNonComment = getNextNonComment();
 643     return NextNonComment && NextNonComment->is(tok::colon);
 644   }
 645
 646   bool canBePointerOrReferenceQualifier() const {
 647     return isOneOf(tok::kw_const, tok::kw_restrict, tok::kw_volatile,
 648                    tok::kw__Nonnull, tok::kw__Nullable,
 649                    tok::kw__Null_unspecified, tok::kw___ptr32, tok::kw___ptr64,
 650                    tok::kw___funcref) ||
 651            isAttribute();
 652   }
 653
 654   /// Determine whether the token is a simple-type-specifier.
 655   [[nodiscard]] bool isSimpleTypeSpecifier() const;
 656
 657   [[nodiscard]] bool isTypeOrIdentifier() const;
 658
 659   bool isObjCAccessSpecifier() const {
 660     return is(tok::at) && Next &&
 661            (Next->isObjCAtKeyword(tok::objc_public) ||
 662             Next->isObjCAtKeyword(tok::objc_protected) ||
 663             Next->isObjCAtKeyword(tok::objc_package) ||
 664             Next->isObjCAtKeyword(tok::objc_private));
 665   }
 666
 667   /// Returns whether \p Tok is ([{ or an opening < of a template or in
 668   /// protos.
 669   bool opensScope() const {
 670     if (is(TT_TemplateString) && TokenText.endswith("${"))
 671       return true;
 672     if (is(TT_DictLiteral) && is(tok::less))
 673       return true;
 674     return isOneOf(tok::l_paren, tok::l_brace, tok::l_square,
 675                    TT_TemplateOpener);
 676   }
 677   /// Returns whether \p Tok is )]} or a closing > of a template or in
 678   /// protos.
 679   bool closesScope() const {
 680     if (is(TT_TemplateString) && TokenText.startswith("}"))
 681       return true;
 682     if (is(TT_DictLiteral) && is(tok::greater))
 683       return true;
 684     return isOneOf(tok::r_paren, tok::r_brace, tok::r_square,
 685                    TT_TemplateCloser);
 686   }
 687
 688   /// Returns \c true if this is a "." or "->" accessing a member.
 689   bool isMemberAccess() const {
 690     return isOneOf(tok::arrow, tok::period, tok::arrowstar) &&
 691            !isOneOf(TT_DesignatedInitializerPeriod, TT_TrailingReturnArrow,
 692                     TT_LeadingJavaAnnotation);
 693   }
 694
 695   bool isUnaryOperator() const {
 696     switch (Tok.getKind()) {
 697     case tok::plus:
 698     case tok::plusplus:
 699     case tok::minus:
 700     case tok::minusminus:
 701     case tok::exclaim:
 702     case tok::tilde:
 703     case tok::kw_sizeof:
 704     case tok::kw_alignof:
 705       return true;
 706     default:
 707       return false;
 708     }
 709   }
 710
 711   bool isBinaryOperator() const {
 712     // Comma is a binary operator, but does not behave as such wrt. formatting.
 713     return getPrecedence() > prec::Comma;
 714   }
 715
 716   bool isTrailingComment() const {
 717     return is(tok::comment) &&
 718            (is(TT_LineComment) || !Next || Next->NewlinesBefore > 0);
 719   }
 720
 721   /// Returns \c true if this is a keyword that can be used
 722   /// like a function call (e.g. sizeof, typeid, ...).
 723   bool isFunctionLikeKeyword() const {
 724     if (isAttribute())
 725       return true;
 726
 727     return isOneOf(tok::kw_throw, tok::kw_typeid, tok::kw_return,
 728                    tok::kw_sizeof, tok::kw_alignof, tok::kw_alignas,
 729                    tok::kw_decltype, tok::kw_noexcept, tok::kw_static_assert,
 730                    tok::kw__Atomic,
 731 #define TRANSFORM_TYPE_TRAIT_DEF(_, Trait) tok::kw___##Trait,
 732 #include "clang/Basic/TransformTypeTraits.def"
 733                    tok::kw_requires);
 734   }
 735
 736   /// Returns \c true if this is a string literal that's like a label,
 737   /// e.g. ends with "=" or ":".
 738   bool isLabelString() const {
 739     if (isNot(tok::string_literal))
 740       return false;
 741     StringRef Content = TokenText;
 742     if (Content.startswith("\"") || Content.startswith("'"))
 743       Content = Content.drop_front(1);
 744     if (Content.endswith("\"") || Content.endswith("'"))
 745       Content = Content.drop_back(1);
 746     Content = Content.trim();
 747     return Content.size() > 1 &&
 748            (Content.back() == ':' || Content.back() == '=');
 749   }
 750
 751   /// Returns actual token start location without leading escaped
 752   /// newlines and whitespace.
 753   ///
 754   /// This can be different to Tok.getLocation(), which includes leading escaped
 755   /// newlines.
 756   SourceLocation getStartOfNonWhitespace() const {
 757     return WhitespaceRange.getEnd();
 758   }
 759
 760   /// Returns \c true if the range of whitespace immediately preceding the \c
 761   /// Token is not empty.
 762   bool hasWhitespaceBefore() const {
 763     return WhitespaceRange.getBegin() != WhitespaceRange.getEnd();
 764   }
 765
 766   prec::Level getPrecedence() const {
 767     if (ForcedPrecedence != prec::Unknown)
 768       return ForcedPrecedence;
 769     return getBinOpPrecedence(Tok.getKind(), /*GreaterThanIsOperator=*/true,
 770                               /*CPlusPlus11=*/true);
 771   }
 772
 773   /// Returns the previous token ignoring comments.
 774   [[nodiscard]] FormatToken *getPreviousNonComment() const {
 775     FormatToken *Tok = Previous;
 776     while (Tok && Tok->is(tok::comment))
 777       Tok = Tok->Previous;
 778     return Tok;
 779   }
 780
 781   /// Returns the next token ignoring comments.
 782   [[nodiscard]] FormatToken *getNextNonComment() const {
 783     FormatToken *Tok = Next;
 784     while (Tok && Tok->is(tok::comment))
 785       Tok = Tok->Next;
 786     return Tok;
 787   }
 788
 789   /// Returns \c true if this token ends a block indented initializer list.
 790   [[nodiscard]] bool isBlockIndentedInitRBrace(const FormatStyle &Style) const;
 791
 792   /// Returns \c true if this tokens starts a block-type list, i.e. a
 793   /// list that should be indented with a block indent.
 794   [[nodiscard]] bool opensBlockOrBlockTypeList(const FormatStyle &Style) const;
 795
 796   /// Returns whether the token is the left square bracket of a C++
 797   /// structured binding declaration.
 798   bool isCppStructuredBinding(const FormatStyle &Style) const {
 799     if (!Style.isCpp() || isNot(tok::l_square))
 800       return false;
 801     const FormatToken *T = this;
 802     do {
 803       T = T->getPreviousNonComment();
 804     } while (T && T->isOneOf(tok::kw_const, tok::kw_volatile, tok::amp,
 805                              tok::ampamp));
 806     return T && T->is(tok::kw_auto);
 807   }
 808
 809   /// Same as opensBlockOrBlockTypeList, but for the closing token.
 810   bool closesBlockOrBlockTypeList(const FormatStyle &Style) const {
 811     if (is(TT_TemplateString) && closesScope())
 812       return true;
 813     return MatchingParen && MatchingParen->opensBlockOrBlockTypeList(Style);
 814   }
 815
 816   /// Return the actual namespace token, if this token starts a namespace
 817   /// block.
 818   const FormatToken *getNamespaceToken() const {
 819     const FormatToken *NamespaceTok = this;
 820     if (is(tok::comment))
 821       NamespaceTok = NamespaceTok->getNextNonComment();
 822     // Detect "(inline|export)? namespace" in the beginning of a line.
 823     if (NamespaceTok && NamespaceTok->isOneOf(tok::kw_inline, tok::kw_export))
 824       NamespaceTok = NamespaceTok->getNextNonComment();
 825     return NamespaceTok &&
 826                    NamespaceTok->isOneOf(tok::kw_namespace, TT_NamespaceMacro)
 827                ? NamespaceTok
 828                : nullptr;
 829   }
 830
 831   void copyFrom(const FormatToken &Tok) { *this = Tok; }
 832
 833 private:
 834   // Only allow copying via the explicit copyFrom method.
 835   FormatToken(const FormatToken &) = delete;
 836   FormatToken &operator=(const FormatToken &) = default;
 837
 838   template <typename A, typename... Ts>
 839   bool startsSequenceInternal(A K1, Ts... Tokens) const {
 840     if (is(tok::comment) && Next)
 841       return Next->startsSequenceInternal(K1, Tokens...);
 842     return is(K1) && Next && Next->startsSequenceInternal(Tokens...);
 843   }
 844
 845   template <typename A> bool startsSequenceInternal(A K1) const {
 846     if (is(tok::comment) && Next)
 847       return Next->startsSequenceInternal(K1);
 848     return is(K1);
 849   }
 850
 851   template <typename A, typename... Ts> bool endsSequenceInternal(A K1) const {
 852     if (is(tok::comment) && Previous)
 853       return Previous->endsSequenceInternal(K1);
 854     return is(K1);
 855   }
 856
 857   template <typename A, typename... Ts>
 858   bool endsSequenceInternal(A K1, Ts... Tokens) const {
 859     if (is(tok::comment) && Previous)
 860       return Previous->endsSequenceInternal(K1, Tokens...);
 861     return is(K1) && Previous && Previous->endsSequenceInternal(Tokens...);
 862   }
 863 };
 864
 865 class ContinuationIndenter;
 866 struct LineState;
 867
 868 class TokenRole {
 869 public:
 870   TokenRole(const FormatStyle &Style) : Style(Style) {}
 871   virtual ~TokenRole();
 872
 873   /// After the \c TokenAnnotator has finished annotating all the tokens,
 874   /// this function precomputes required information for formatting.
 875   virtual void precomputeFormattingInfos(const FormatToken *Token);
 876
 877   /// Apply the special formatting that the given role demands.
 878   ///
 879   /// Assumes that the token having this role is already formatted.
 880   ///
 881   /// Continues formatting from \p State leaving indentation to \p Indenter and
 882   /// returns the total penalty that this formatting incurs.
 883   virtual unsigned formatFromToken(LineState &State,
 884                                    ContinuationIndenter *Indenter,
 885                                    bool DryRun) {
 886     return 0;
 887   }
 888
 889   /// Same as \c formatFromToken, but assumes that the first token has
 890   /// already been set thereby deciding on the first line break.
 891   virtual unsigned formatAfterToken(LineState &State,
 892                                     ContinuationIndenter *Indenter,
 893                                     bool DryRun) {
 894     return 0;
 895   }
 896
 897   /// Notifies the \c Role that a comma was found.
 898   virtual void CommaFound(const FormatToken *Token) {}
 899
 900   virtual const FormatToken *lastComma() { return nullptr; }
 901
 902 protected:
 903   const FormatStyle &Style;
 904 };
 905
 906 class CommaSeparatedList : public TokenRole {
 907 public:
 908   CommaSeparatedList(const FormatStyle &Style)
 909       : TokenRole(Style), HasNestedBracedList(false) {}
 910
 911   void precomputeFormattingInfos(const FormatToken *Token) override;
 912
 913   unsigned formatAfterToken(LineState &State, ContinuationIndenter *Indenter,
 914                             bool DryRun) override;
 915
 916   unsigned formatFromToken(LineState &State, ContinuationIndenter *Indenter,
 917                            bool DryRun) override;
 918
 919   /// Adds \p Token as the next comma to the \c CommaSeparated list.
 920   void CommaFound(const FormatToken *Token) override {
 921     Commas.push_back(Token);
 922   }
 923
 924   const FormatToken *lastComma() override {
 925     if (Commas.empty())
 926       return nullptr;
 927     return Commas.back();
 928   }
 929
 930 private:
 931   /// A struct that holds information on how to format a given list with
 932   /// a specific number of columns.
 933   struct ColumnFormat {
 934     /// The number of columns to use.
 935     unsigned Columns;
 936
 937     /// The total width in characters.
 938     unsigned TotalWidth;
 939
 940     /// The number of lines required for this format.
 941     unsigned LineCount;
 942
 943     /// The size of each column in characters.
 944     SmallVector<unsigned, 8> ColumnSizes;
 945   };
 946
 947   /// Calculate which \c ColumnFormat fits best into
 948   /// \p RemainingCharacters.
 949   const ColumnFormat *getColumnFormat(unsigned RemainingCharacters) const;
 950
 951   /// The ordered \c FormatTokens making up the commas of this list.
 952   SmallVector<const FormatToken *, 8> Commas;
 953
 954   /// The length of each of the list's items in characters including the
 955   /// trailing comma.
 956   SmallVector<unsigned, 8> ItemLengths;
 957
 958   /// Precomputed formats that can be used for this list.
 959   SmallVector<ColumnFormat, 4> Formats;
 960
 961   bool HasNestedBracedList;
 962 };
 963
 964 /// Encapsulates keywords that are context sensitive or for languages not
 965 /// properly supported by Clang's lexer.
 966 struct AdditionalKeywords {
 967   AdditionalKeywords(IdentifierTable &IdentTable) {
 968     kw_final = &IdentTable.get("final");
 969     kw_override = &IdentTable.get("override");
 970     kw_in = &IdentTable.get("in");
 971     kw_of = &IdentTable.get("of");
 972     kw_CF_CLOSED_ENUM = &IdentTable.get("CF_CLOSED_ENUM");
 973     kw_CF_ENUM = &IdentTable.get("CF_ENUM");
 974     kw_CF_OPTIONS = &IdentTable.get("CF_OPTIONS");
 975     kw_NS_CLOSED_ENUM = &IdentTable.get("NS_CLOSED_ENUM");
 976     kw_NS_ENUM = &IdentTable.get("NS_ENUM");
 977     kw_NS_ERROR_ENUM = &IdentTable.get("NS_ERROR_ENUM");
 978     kw_NS_OPTIONS = &IdentTable.get("NS_OPTIONS");
 979
 980     kw_as = &IdentTable.get("as");
 981     kw_async = &IdentTable.get("async");
 982     kw_await = &IdentTable.get("await");
 983     kw_declare = &IdentTable.get("declare");
 984     kw_finally = &IdentTable.get("finally");
 985     kw_from = &IdentTable.get("from");
 986     kw_function = &IdentTable.get("function");
 987     kw_get = &IdentTable.get("get");
 988     kw_import = &IdentTable.get("import");
 989     kw_infer = &IdentTable.get("infer");
 990     kw_is = &IdentTable.get("is");
 991     kw_let = &IdentTable.get("let");
 992     kw_module = &IdentTable.get("module");
 993     kw_readonly = &IdentTable.get("readonly");
 994     kw_set = &IdentTable.get("set");
 995     kw_type = &IdentTable.get("type");
 996     kw_typeof = &IdentTable.get("typeof");
 997     kw_var = &IdentTable.get("var");
 998     kw_yield = &IdentTable.get("yield");
 999
1000     kw_abstract = &IdentTable.get("abstract");
1001     kw_assert = &IdentTable.get("assert");
1002     kw_extends = &IdentTable.get("extends");
1003     kw_implements = &IdentTable.get("implements");
1004     kw_instanceof = &IdentTable.get("instanceof");
1005     kw_interface = &IdentTable.get("interface");
1006     kw_native = &IdentTable.get("native");
1007     kw_package = &IdentTable.get("package");
1008     kw_synchronized = &IdentTable.get("synchronized");
1009     kw_throws = &IdentTable.get("throws");
1010     kw___except = &IdentTable.get("__except");
1011     kw___has_include = &IdentTable.get("__has_include");
1012     kw___has_include_next = &IdentTable.get("__has_include_next");
1013
1014     kw_mark = &IdentTable.get("mark");
1015     kw_region = &IdentTable.get("region");
1016
1017     kw_extend = &IdentTable.get("extend");
1018     kw_option = &IdentTable.get("option");
1019     kw_optional = &IdentTable.get("optional");
1020     kw_repeated = &IdentTable.get("repeated");
1021     kw_required = &IdentTable.get("required");
1022     kw_returns = &IdentTable.get("returns");
1023
1024     kw_signals = &IdentTable.get("signals");
1025     kw_qsignals = &IdentTable.get("Q_SIGNALS");
1026     kw_slots = &IdentTable.get("slots");
1027     kw_qslots = &IdentTable.get("Q_SLOTS");
1028
1029     // For internal clang-format use.
1030     kw_internal_ident_after_define =
1031         &IdentTable.get("__CLANG_FORMAT_INTERNAL_IDENT_AFTER_DEFINE__");
1032
1033     // C# keywords
1034     kw_dollar = &IdentTable.get("dollar");
1035     kw_base = &IdentTable.get("base");
1036     kw_byte = &IdentTable.get("byte");
1037     kw_checked = &IdentTable.get("checked");
1038     kw_decimal = &IdentTable.get("decimal");
1039     kw_delegate = &IdentTable.get("delegate");
1040     kw_event = &IdentTable.get("event");
1041     kw_fixed = &IdentTable.get("fixed");
1042     kw_foreach = &IdentTable.get("foreach");
1043     kw_init = &IdentTable.get("init");
1044     kw_implicit = &IdentTable.get("implicit");
1045     kw_internal = &IdentTable.get("internal");
1046     kw_lock = &IdentTable.get("lock");
1047     kw_null = &IdentTable.get("null");
1048     kw_object = &IdentTable.get("object");
1049     kw_out = &IdentTable.get("out");
1050     kw_params = &IdentTable.get("params");
1051     kw_ref = &IdentTable.get("ref");
1052     kw_string = &IdentTable.get("string");
1053     kw_stackalloc = &IdentTable.get("stackalloc");
1054     kw_sbyte = &IdentTable.get("sbyte");
1055     kw_sealed = &IdentTable.get("sealed");
1056     kw_uint = &IdentTable.get("uint");
1057     kw_ulong = &IdentTable.get("ulong");
1058     kw_unchecked = &IdentTable.get("unchecked");
1059     kw_unsafe = &IdentTable.get("unsafe");
1060     kw_ushort = &IdentTable.get("ushort");
1061     kw_when = &IdentTable.get("when");
1062     kw_where = &IdentTable.get("where");
1063
1064     // Verilog keywords
1065     kw_always = &IdentTable.get("always");
1066     kw_always_comb = &IdentTable.get("always_comb");
1067     kw_always_ff = &IdentTable.get("always_ff");
1068     kw_always_latch = &IdentTable.get("always_latch");
1069     kw_assign = &IdentTable.get("assign");
1070     kw_assume = &IdentTable.get("assume");
1071     kw_automatic = &IdentTable.get("automatic");
1072     kw_before = &IdentTable.get("before");
1073     kw_begin = &IdentTable.get("begin");
1074     kw_begin_keywords = &IdentTable.get("begin_keywords");
1075     kw_bins = &IdentTable.get("bins");
1076     kw_binsof = &IdentTable.get("binsof");
1077     kw_casex = &IdentTable.get("casex");
1078     kw_casez = &IdentTable.get("casez");
1079     kw_celldefine = &IdentTable.get("celldefine");
1080     kw_checker = &IdentTable.get("checker");
1081     kw_clocking = &IdentTable.get("clocking");
1082     kw_constraint = &IdentTable.get("constraint");
1083     kw_cover = &IdentTable.get("cover");
1084     kw_covergroup = &IdentTable.get("covergroup");
1085     kw_coverpoint = &IdentTable.get("coverpoint");
1086     kw_default_decay_time = &IdentTable.get("default_decay_time");
1087     kw_default_nettype = &IdentTable.get("default_nettype");
1088     kw_default_trireg_strength = &IdentTable.get("default_trireg_strength");
1089     kw_delay_mode_distributed = &IdentTable.get("delay_mode_distributed");
1090     kw_delay_mode_path = &IdentTable.get("delay_mode_path");
1091     kw_delay_mode_unit = &IdentTable.get("delay_mode_unit");
1092     kw_delay_mode_zero = &IdentTable.get("delay_mode_zero");
1093     kw_disable = &IdentTable.get("disable");
1094     kw_dist = &IdentTable.get("dist");
1095     kw_edge = &IdentTable.get("edge");
1096     kw_elsif = &IdentTable.get("elsif");
1097     kw_end = &IdentTable.get("end");
1098     kw_end_keywords = &IdentTable.get("end_keywords");
1099     kw_endcase = &IdentTable.get("endcase");
1100     kw_endcelldefine = &IdentTable.get("endcelldefine");
1101     kw_endchecker = &IdentTable.get("endchecker");
1102     kw_endclass = &IdentTable.get("endclass");
1103     kw_endclocking = &IdentTable.get("endclocking");
1104     kw_endfunction = &IdentTable.get("endfunction");
1105     kw_endgenerate = &IdentTable.get("endgenerate");
1106     kw_endgroup = &IdentTable.get("endgroup");
1107     kw_endinterface = &IdentTable.get("endinterface");
1108     kw_endmodule = &IdentTable.get("endmodule");
1109     kw_endpackage = &IdentTable.get("endpackage");
1110     kw_endprimitive = &IdentTable.get("endprimitive");
1111     kw_endprogram = &IdentTable.get("endprogram");
1112     kw_endproperty = &IdentTable.get("endproperty");
1113     kw_endsequence = &IdentTable.get("endsequence");
1114     kw_endspecify = &IdentTable.get("endspecify");
1115     kw_endtable = &IdentTable.get("endtable");
1116     kw_endtask = &IdentTable.get("endtask");
1117     kw_forever = &IdentTable.get("forever");
1118     kw_fork = &IdentTable.get("fork");
1119     kw_generate = &IdentTable.get("generate");
1120     kw_highz0 = &IdentTable.get("highz0");
1121     kw_highz1 = &IdentTable.get("highz1");
1122     kw_iff = &IdentTable.get("iff");
1123     kw_ifnone = &IdentTable.get("ifnone");
1124     kw_ignore_bins = &IdentTable.get("ignore_bins");
1125     kw_illegal_bins = &IdentTable.get("illegal_bins");
1126     kw_initial = &IdentTable.get("initial");
1127     kw_inout = &IdentTable.get("inout");
1128     kw_input = &IdentTable.get("input");
1129     kw_inside = &IdentTable.get("inside");
1130     kw_interconnect = &IdentTable.get("interconnect");
1131     kw_intersect = &IdentTable.get("intersect");
1132     kw_join = &IdentTable.get("join");
1133     kw_join_any = &IdentTable.get("join_any");
1134     kw_join_none = &IdentTable.get("join_none");
1135     kw_large = &IdentTable.get("large");
1136     kw_local = &IdentTable.get("local");
1137     kw_localparam = &IdentTable.get("localparam");
1138     kw_macromodule = &IdentTable.get("macromodule");
1139     kw_matches = &IdentTable.get("matches");
1140     kw_medium = &IdentTable.get("medium");
1141     kw_negedge = &IdentTable.get("negedge");
1142     kw_nounconnected_drive = &IdentTable.get("nounconnected_drive");
1143     kw_output = &IdentTable.get("output");
1144     kw_packed = &IdentTable.get("packed");
1145     kw_parameter = &IdentTable.get("parameter");
1146     kw_posedge = &IdentTable.get("posedge");
1147     kw_primitive = &IdentTable.get("primitive");
1148     kw_priority = &IdentTable.get("priority");
1149     kw_program = &IdentTable.get("program");
1150     kw_property = &IdentTable.get("property");
1151     kw_pull0 = &IdentTable.get("pull0");
1152     kw_pull1 = &IdentTable.get("pull1");
1153     kw_pure = &IdentTable.get("pure");
1154     kw_rand = &IdentTable.get("rand");
1155     kw_randc = &IdentTable.get("randc");
1156     kw_randcase = &IdentTable.get("randcase");
1157     kw_randsequence = &IdentTable.get("randsequence");
1158     kw_repeat = &IdentTable.get("repeat");
1159     kw_resetall = &IdentTable.get("resetall");
1160     kw_sample = &IdentTable.get("sample");
1161     kw_scalared = &IdentTable.get("scalared");
1162     kw_sequence = &IdentTable.get("sequence");
1163     kw_small = &IdentTable.get("small");
1164     kw_soft = &IdentTable.get("soft");
1165     kw_solve = &IdentTable.get("solve");
1166     kw_specify = &IdentTable.get("specify");
1167     kw_specparam = &IdentTable.get("specparam");
1168     kw_strong0 = &IdentTable.get("strong0");
1169     kw_strong1 = &IdentTable.get("strong1");
1170     kw_supply0 = &IdentTable.get("supply0");
1171     kw_supply1 = &IdentTable.get("supply1");
1172     kw_table = &IdentTable.get("table");
1173     kw_tagged = &IdentTable.get("tagged");
1174     kw_task = &IdentTable.get("task");
1175     kw_timescale = &IdentTable.get("timescale");
1176     kw_tri = &IdentTable.get("tri");
1177     kw_tri0 = &IdentTable.get("tri0");
1178     kw_tri1 = &IdentTable.get("tri1");
1179     kw_triand = &IdentTable.get("triand");
1180     kw_trior = &IdentTable.get("trior");
1181     kw_trireg = &IdentTable.get("trireg");
1182     kw_unconnected_drive = &IdentTable.get("unconnected_drive");
1183     kw_undefineall = &IdentTable.get("undefineall");
1184     kw_unique = &IdentTable.get("unique");
1185     kw_unique0 = &IdentTable.get("unique0");
1186     kw_uwire = &IdentTable.get("uwire");
1187     kw_vectored = &IdentTable.get("vectored");
1188     kw_wand = &IdentTable.get("wand");
1189     kw_weak0 = &IdentTable.get("weak0");
1190     kw_weak1 = &IdentTable.get("weak1");
1191     kw_wildcard = &IdentTable.get("wildcard");
1192     kw_wire = &IdentTable.get("wire");
1193     kw_with = &IdentTable.get("with");
1194     kw_wor = &IdentTable.get("wor");
1195
1196     // Symbols that are treated as keywords.
1197     kw_verilogHash = &IdentTable.get("#");
1198     kw_verilogHashHash = &IdentTable.get("##");
1199     kw_apostrophe = &IdentTable.get("\'");
1200
1201     // Keep this at the end of the constructor to make sure everything here
1202     // is
1203     // already initialized.
1204     JsExtraKeywords = std::unordered_set<IdentifierInfo *>(
1205         {kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from,
1206          kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_override,
1207          kw_readonly, kw_set, kw_type, kw_typeof, kw_var, kw_yield,
1208          // Keywords from the Java section.
1209          kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface});
1210
1211     CSharpExtraKeywords = std::unordered_set<IdentifierInfo *>(
1212         {kw_base, kw_byte, kw_checked, kw_decimal, kw_delegate, kw_event,
1213          kw_fixed, kw_foreach, kw_implicit, kw_in, kw_init, kw_interface,
1214          kw_internal, kw_is, kw_lock, kw_null, kw_object, kw_out, kw_override,
1215          kw_params, kw_readonly, kw_ref, kw_string, kw_stackalloc, kw_sbyte,
1216          kw_sealed, kw_uint, kw_ulong, kw_unchecked, kw_unsafe, kw_ushort,
1217          kw_when, kw_where,
1218          // Keywords from the JavaScript section.
1219          kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from,
1220          kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_readonly,
1221          kw_set, kw_type, kw_typeof, kw_var, kw_yield,
1222          // Keywords from the Java section.
1223          kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface});
1224
1225     // Some keywords are not included here because they don't need special
1226     // treatment like `showcancelled` or they should be treated as identifiers
1227     // like `int` and `logic`.
1228     VerilogExtraKeywords = std::unordered_set<IdentifierInfo *>(
1229         {kw_always,       kw_always_comb,
1230          kw_always_ff,    kw_always_latch,
1231          kw_assert,       kw_assign,
1232          kw_assume,       kw_automatic,
1233          kw_before,       kw_begin,
1234          kw_bins,         kw_binsof,
1235          kw_casex,        kw_casez,
1236          kw_celldefine,   kw_checker,
1237          kw_clocking,     kw_constraint,
1238          kw_cover,        kw_covergroup,
1239          kw_coverpoint,   kw_disable,
1240          kw_dist,         kw_edge,
1241          kw_end,          kw_endcase,
1242          kw_endchecker,   kw_endclass,
1243          kw_endclocking,  kw_endfunction,
1244          kw_endgenerate,  kw_endgroup,
1245          kw_endinterface, kw_endmodule,
1246          kw_endpackage,   kw_endprimitive,
1247          kw_endprogram,   kw_endproperty,
1248          kw_endsequence,  kw_endspecify,
1249          kw_endtable,     kw_endtask,
1250          kw_extends,      kw_final,
1251          kw_foreach,      kw_forever,
1252          kw_fork,         kw_function,
1253          kw_generate,     kw_highz0,
1254          kw_highz1,       kw_iff,
1255          kw_ifnone,       kw_ignore_bins,
1256          kw_illegal_bins, kw_implements,
1257          kw_import,       kw_initial,
1258          kw_inout,        kw_input,
1259          kw_inside,       kw_interconnect,
1260          kw_interface,    kw_intersect,
1261          kw_join,         kw_join_any,
1262          kw_join_none,    kw_large,
1263          kw_let,          kw_local,
1264          kw_localparam,   kw_macromodule,
1265          kw_matches,      kw_medium,
1266          kw_negedge,      kw_output,
1267          kw_package,      kw_packed,
1268          kw_parameter,    kw_posedge,
1269          kw_primitive,    kw_priority,
1270          kw_program,      kw_property,
1271          kw_pull0,        kw_pull1,
1272          kw_pure,         kw_rand,
1273          kw_randc,        kw_randcase,
1274          kw_randsequence, kw_ref,
1275          kw_repeat,       kw_sample,
1276          kw_scalared,     kw_sequence,
1277          kw_small,        kw_soft,
1278          kw_solve,        kw_specify,
1279          kw_specparam,    kw_strong0,
1280          kw_strong1,      kw_supply0,
1281          kw_supply1,      kw_table,
1282          kw_tagged,       kw_task,
1283          kw_tri,          kw_tri0,
1284          kw_tri1,         kw_triand,
1285          kw_trior,        kw_trireg,
1286          kw_unique,       kw_unique0,
1287          kw_uwire,        kw_var,
1288          kw_vectored,     kw_wand,
1289          kw_weak0,        kw_weak1,
1290          kw_wildcard,     kw_wire,
1291          kw_with,         kw_wor,
1292          kw_verilogHash,  kw_verilogHashHash});
1293   }
1294
1295   // Context sensitive keywords.
1296   IdentifierInfo *kw_final;
1297   IdentifierInfo *kw_override;
1298   IdentifierInfo *kw_in;
1299   IdentifierInfo *kw_of;
1300   IdentifierInfo *kw_CF_CLOSED_ENUM;
1301   IdentifierInfo *kw_CF_ENUM;
1302   IdentifierInfo *kw_CF_OPTIONS;
1303   IdentifierInfo *kw_NS_CLOSED_ENUM;
1304   IdentifierInfo *kw_NS_ENUM;
1305   IdentifierInfo *kw_NS_ERROR_ENUM;
1306   IdentifierInfo *kw_NS_OPTIONS;
1307   IdentifierInfo *kw___except;
1308   IdentifierInfo *kw___has_include;
1309   IdentifierInfo *kw___has_include_next;
1310
1311   // JavaScript keywords.
1312   IdentifierInfo *kw_as;
1313   IdentifierInfo *kw_async;
1314   IdentifierInfo *kw_await;
1315   IdentifierInfo *kw_declare;
1316   IdentifierInfo *kw_finally;
1317   IdentifierInfo *kw_from;
1318   IdentifierInfo *kw_function;
1319   IdentifierInfo *kw_get;
1320   IdentifierInfo *kw_import;
1321   IdentifierInfo *kw_infer;
1322   IdentifierInfo *kw_is;
1323   IdentifierInfo *kw_let;
1324   IdentifierInfo *kw_module;
1325   IdentifierInfo *kw_readonly;
1326   IdentifierInfo *kw_set;
1327   IdentifierInfo *kw_type;
1328   IdentifierInfo *kw_typeof;
1329   IdentifierInfo *kw_var;
1330   IdentifierInfo *kw_yield;
1331
1332   // Java keywords.
1333   IdentifierInfo *kw_abstract;
1334   IdentifierInfo *kw_assert;
1335   IdentifierInfo *kw_extends;
1336   IdentifierInfo *kw_implements;
1337   IdentifierInfo *kw_instanceof;
1338   IdentifierInfo *kw_interface;
1339   IdentifierInfo *kw_native;
1340   IdentifierInfo *kw_package;
1341   IdentifierInfo *kw_synchronized;
1342   IdentifierInfo *kw_throws;
1343
1344   // Pragma keywords.
1345   IdentifierInfo *kw_mark;
1346   IdentifierInfo *kw_region;
1347
1348   // Proto keywords.
1349   IdentifierInfo *kw_extend;
1350   IdentifierInfo *kw_option;
1351   IdentifierInfo *kw_optional;
1352   IdentifierInfo *kw_repeated;
1353   IdentifierInfo *kw_required;
1354   IdentifierInfo *kw_returns;
1355
1356   // QT keywords.
1357   IdentifierInfo *kw_signals;
1358   IdentifierInfo *kw_qsignals;
1359   IdentifierInfo *kw_slots;
1360   IdentifierInfo *kw_qslots;
1361
1362   // For internal use by clang-format.
1363   IdentifierInfo *kw_internal_ident_after_define;
1364
1365   // C# keywords
1366   IdentifierInfo *kw_dollar;
1367   IdentifierInfo *kw_base;
1368   IdentifierInfo *kw_byte;
1369   IdentifierInfo *kw_checked;
1370   IdentifierInfo *kw_decimal;
1371   IdentifierInfo *kw_delegate;
1372   IdentifierInfo *kw_event;
1373   IdentifierInfo *kw_fixed;
1374   IdentifierInfo *kw_foreach;
1375   IdentifierInfo *kw_implicit;
1376   IdentifierInfo *kw_init;
1377   IdentifierInfo *kw_internal;
1378
1379   IdentifierInfo *kw_lock;
1380   IdentifierInfo *kw_null;
1381   IdentifierInfo *kw_object;
1382   IdentifierInfo *kw_out;
1383
1384   IdentifierInfo *kw_params;
1385
1386   IdentifierInfo *kw_ref;
1387   IdentifierInfo *kw_string;
1388   IdentifierInfo *kw_stackalloc;
1389   IdentifierInfo *kw_sbyte;
1390   IdentifierInfo *kw_sealed;
1391   IdentifierInfo *kw_uint;
1392   IdentifierInfo *kw_ulong;
1393   IdentifierInfo *kw_unchecked;
1394   IdentifierInfo *kw_unsafe;
1395   IdentifierInfo *kw_ushort;
1396   IdentifierInfo *kw_when;
1397   IdentifierInfo *kw_where;
1398
1399   // Verilog keywords
1400   IdentifierInfo *kw_always;
1401   IdentifierInfo *kw_always_comb;
1402   IdentifierInfo *kw_always_ff;
1403   IdentifierInfo *kw_always_latch;
1404   IdentifierInfo *kw_assign;
1405   IdentifierInfo *kw_assume;
1406   IdentifierInfo *kw_automatic;
1407   IdentifierInfo *kw_before;
1408   IdentifierInfo *kw_begin;
1409   IdentifierInfo *kw_begin_keywords;
1410   IdentifierInfo *kw_bins;
1411   IdentifierInfo *kw_binsof;
1412   IdentifierInfo *kw_casex;
1413   IdentifierInfo *kw_casez;
1414   IdentifierInfo *kw_celldefine;
1415   IdentifierInfo *kw_checker;
1416   IdentifierInfo *kw_clocking;
1417   IdentifierInfo *kw_constraint;
1418   IdentifierInfo *kw_cover;
1419   IdentifierInfo *kw_covergroup;
1420   IdentifierInfo *kw_coverpoint;
1421   IdentifierInfo *kw_default_decay_time;
1422   IdentifierInfo *kw_default_nettype;
1423   IdentifierInfo *kw_default_trireg_strength;
1424   IdentifierInfo *kw_delay_mode_distributed;
1425   IdentifierInfo *kw_delay_mode_path;
1426   IdentifierInfo *kw_delay_mode_unit;
1427   IdentifierInfo *kw_delay_mode_zero;
1428   IdentifierInfo *kw_disable;
1429   IdentifierInfo *kw_dist;
1430   IdentifierInfo *kw_elsif;
1431   IdentifierInfo *kw_edge;
1432   IdentifierInfo *kw_end;
1433   IdentifierInfo *kw_end_keywords;
1434   IdentifierInfo *kw_endcase;
1435   IdentifierInfo *kw_endcelldefine;
1436   IdentifierInfo *kw_endchecker;
1437   IdentifierInfo *kw_endclass;
1438   IdentifierInfo *kw_endclocking;
1439   IdentifierInfo *kw_endfunction;
1440   IdentifierInfo *kw_endgenerate;
1441   IdentifierInfo *kw_endgroup;
1442   IdentifierInfo *kw_endinterface;
1443   IdentifierInfo *kw_endmodule;
1444   IdentifierInfo *kw_endpackage;
1445   IdentifierInfo *kw_endprimitive;
1446   IdentifierInfo *kw_endprogram;
1447   IdentifierInfo *kw_endproperty;
1448   IdentifierInfo *kw_endsequence;
1449   IdentifierInfo *kw_endspecify;
1450   IdentifierInfo *kw_endtable;
1451   IdentifierInfo *kw_endtask;
1452   IdentifierInfo *kw_forever;
1453   IdentifierInfo *kw_fork;
1454   IdentifierInfo *kw_generate;
1455   IdentifierInfo *kw_highz0;
1456   IdentifierInfo *kw_highz1;
1457   IdentifierInfo *kw_iff;
1458   IdentifierInfo *kw_ifnone;
1459   IdentifierInfo *kw_ignore_bins;
1460   IdentifierInfo *kw_illegal_bins;
1461   IdentifierInfo *kw_initial;
1462   IdentifierInfo *kw_inout;
1463   IdentifierInfo *kw_input;
1464   IdentifierInfo *kw_inside;
1465   IdentifierInfo *kw_interconnect;
1466   IdentifierInfo *kw_intersect;
1467   IdentifierInfo *kw_join;
1468   IdentifierInfo *kw_join_any;
1469   IdentifierInfo *kw_join_none;
1470   IdentifierInfo *kw_large;
1471   IdentifierInfo *kw_local;
1472   IdentifierInfo *kw_localparam;
1473   IdentifierInfo *kw_macromodule;
1474   IdentifierInfo *kw_matches;
1475   IdentifierInfo *kw_medium;
1476   IdentifierInfo *kw_negedge;
1477   IdentifierInfo *kw_nounconnected_drive;
1478   IdentifierInfo *kw_output;
1479   IdentifierInfo *kw_packed;
1480   IdentifierInfo *kw_parameter;
1481   IdentifierInfo *kw_posedge;
1482   IdentifierInfo *kw_primitive;
1483   IdentifierInfo *kw_priority;
1484   IdentifierInfo *kw_program;
1485   IdentifierInfo *kw_property;
1486   IdentifierInfo *kw_pull0;
1487   IdentifierInfo *kw_pull1;
1488   IdentifierInfo *kw_pure;
1489   IdentifierInfo *kw_rand;
1490   IdentifierInfo *kw_randc;
1491   IdentifierInfo *kw_randcase;
1492   IdentifierInfo *kw_randsequence;
1493   IdentifierInfo *kw_repeat;
1494   IdentifierInfo *kw_resetall;
1495   IdentifierInfo *kw_sample;
1496   IdentifierInfo *kw_scalared;
1497   IdentifierInfo *kw_sequence;
1498   IdentifierInfo *kw_small;
1499   IdentifierInfo *kw_soft;
1500   IdentifierInfo *kw_solve;
1501   IdentifierInfo *kw_specify;
1502   IdentifierInfo *kw_specparam;
1503   IdentifierInfo *kw_strong0;
1504   IdentifierInfo *kw_strong1;
1505   IdentifierInfo *kw_supply0;
1506   IdentifierInfo *kw_supply1;
1507   IdentifierInfo *kw_table;
1508   IdentifierInfo *kw_tagged;
1509   IdentifierInfo *kw_task;
1510   IdentifierInfo *kw_timescale;
1511   IdentifierInfo *kw_tri0;
1512   IdentifierInfo *kw_tri1;
1513   IdentifierInfo *kw_tri;
1514   IdentifierInfo *kw_triand;
1515   IdentifierInfo *kw_trior;
1516   IdentifierInfo *kw_trireg;
1517   IdentifierInfo *kw_unconnected_drive;
1518   IdentifierInfo *kw_undefineall;
1519   IdentifierInfo *kw_unique;
1520   IdentifierInfo *kw_unique0;
1521   IdentifierInfo *kw_uwire;
1522   IdentifierInfo *kw_vectored;
1523   IdentifierInfo *kw_wand;
1524   IdentifierInfo *kw_weak0;
1525   IdentifierInfo *kw_weak1;
1526   IdentifierInfo *kw_wildcard;
1527   IdentifierInfo *kw_wire;
1528   IdentifierInfo *kw_with;
1529   IdentifierInfo *kw_wor;
1530
1531   // Workaround for hashes and backticks in Verilog.
1532   IdentifierInfo *kw_verilogHash;
1533   IdentifierInfo *kw_verilogHashHash;
1534
1535   // Symbols in Verilog that don't exist in C++.
1536   IdentifierInfo *kw_apostrophe;
1537
1538   /// Returns \c true if \p Tok is a keyword or an identifier.
1539   bool isWordLike(const FormatToken &Tok) const {
1540     // getIdentifierinfo returns non-null for keywords as well as identifiers.
1541     return Tok.Tok.getIdentifierInfo() &&
1542            !Tok.isOneOf(kw_verilogHash, kw_verilogHashHash, kw_apostrophe);
1543   }
1544
1545   /// Returns \c true if \p Tok is a true JavaScript identifier, returns
1546   /// \c false if it is a keyword or a pseudo keyword.
1547   /// If \c AcceptIdentifierName is true, returns true not only for keywords,
1548   // but also for IdentifierName tokens (aka pseudo-keywords), such as
1549   // ``yield``.
1550   bool IsJavaScriptIdentifier(const FormatToken &Tok,
1551                               bool AcceptIdentifierName = true) const {
1552     // Based on the list of JavaScript & TypeScript keywords here:
1553     // https://github.com/microsoft/TypeScript/blob/main/src/compiler/scanner.ts#L74
1554     switch (Tok.Tok.getKind()) {
1555     case tok::kw_break:
1556     case tok::kw_case:
1557     case tok::kw_catch:
1558     case tok::kw_class:
1559     case tok::kw_continue:
1560     case tok::kw_const:
1561     case tok::kw_default:
1562     case tok::kw_delete:
1563     case tok::kw_do:
1564     case tok::kw_else:
1565     case tok::kw_enum:
1566     case tok::kw_export:
1567     case tok::kw_false:
1568     case tok::kw_for:
1569     case tok::kw_if:
1570     case tok::kw_import:
1571     case tok::kw_module:
1572     case tok::kw_new:
1573     case tok::kw_private:
1574     case tok::kw_protected:
1575     case tok::kw_public:
1576     case tok::kw_return:
1577     case tok::kw_static:
1578     case tok::kw_switch:
1579     case tok::kw_this:
1580     case tok::kw_throw:
1581     case tok::kw_true:
1582     case tok::kw_try:
1583     case tok::kw_typeof:
1584     case tok::kw_void:
1585     case tok::kw_while:
1586       // These are JS keywords that are lexed by LLVM/clang as keywords.
1587       return false;
1588     case tok::identifier: {
1589       // For identifiers, make sure they are true identifiers, excluding the
1590       // JavaScript pseudo-keywords (not lexed by LLVM/clang as keywords).
1591       bool IsPseudoKeyword =
1592           JsExtraKeywords.find(Tok.Tok.getIdentifierInfo()) !=
1593           JsExtraKeywords.end();
1594       return AcceptIdentifierName || !IsPseudoKeyword;
1595     }
1596     default:
1597       // Other keywords are handled in the switch below, to avoid problems due
1598       // to duplicate case labels when using the #include trick.
1599       break;
1600     }
1601
1602     switch (Tok.Tok.getKind()) {
1603       // Handle C++ keywords not included above: these are all JS identifiers.
1604 #define KEYWORD(X, Y) case tok::kw_##X:
1605 #include "clang/Basic/TokenKinds.def"
1606       // #undef KEYWORD is not needed -- it's #undef-ed at the end of
1607       // TokenKinds.def
1608       return true;
1609     default:
1610       // All other tokens (punctuation etc) are not JS identifiers.
1611       return false;
1612     }
1613   }
1614
1615   /// Returns \c true if \p Tok is a C# keyword, returns
1616   /// \c false if it is a anything else.
1617   bool isCSharpKeyword(const FormatToken &Tok) const {
1618     switch (Tok.Tok.getKind()) {
1619     case tok::kw_bool:
1620     case tok::kw_break:
1621     case tok::kw_case:
1622     case tok::kw_catch:
1623     case tok::kw_char:
1624     case tok::kw_class:
1625     case tok::kw_const:
1626     case tok::kw_continue:
1627     case tok::kw_default:
1628     case tok::kw_do:
1629     case tok::kw_double:
1630     case tok::kw_else:
1631     case tok::kw_enum:
1632     case tok::kw_explicit:
1633     case tok::kw_extern:
1634     case tok::kw_false:
1635     case tok::kw_float:
1636     case tok::kw_for:
1637     case tok::kw_goto:
1638     case tok::kw_if:
1639     case tok::kw_int:
1640     case tok::kw_long:
1641     case tok::kw_namespace:
1642     case tok::kw_new:
1643     case tok::kw_operator:
1644     case tok::kw_private:
1645     case tok::kw_protected:
1646     case tok::kw_public:
1647     case tok::kw_return:
1648     case tok::kw_short:
1649     case tok::kw_sizeof:
1650     case tok::kw_static:
1651     case tok::kw_struct:
1652     case tok::kw_switch:
1653     case tok::kw_this:
1654     case tok::kw_throw:
1655     case tok::kw_true:
1656     case tok::kw_try:
1657     case tok::kw_typeof:
1658     case tok::kw_using:
1659     case tok::kw_virtual:
1660     case tok::kw_void:
1661     case tok::kw_volatile:
1662     case tok::kw_while:
1663       return true;
1664     default:
1665       return Tok.is(tok::identifier) &&
1666              CSharpExtraKeywords.find(Tok.Tok.getIdentifierInfo()) ==
1667                  CSharpExtraKeywords.end();
1668     }
1669   }
1670
1671   bool isVerilogWordOperator(const FormatToken &Tok) const {
1672     return Tok.isOneOf(kw_before, kw_intersect, kw_dist, kw_iff, kw_inside,
1673                        kw_with);
1674   }
1675
1676   bool isVerilogIdentifier(const FormatToken &Tok) const {
1677     switch (Tok.Tok.getKind()) {
1678     case tok::kw_case:
1679     case tok::kw_class:
1680     case tok::kw_const:
1681     case tok::kw_continue:
1682     case tok::kw_default:
1683     case tok::kw_do:
1684     case tok::kw_extern:
1685     case tok::kw_else:
1686     case tok::kw_enum:
1687     case tok::kw_for:
1688     case tok::kw_if:
1689     case tok::kw_restrict:
1690     case tok::kw_signed:
1691     case tok::kw_static:
1692     case tok::kw_struct:
1693     case tok::kw_typedef:
1694     case tok::kw_union:
1695     case tok::kw_unsigned:
1696     case tok::kw_virtual:
1697     case tok::kw_while:
1698       return false;
1699     case tok::identifier:
1700       return isWordLike(Tok) &&
1701              VerilogExtraKeywords.find(Tok.Tok.getIdentifierInfo()) ==
1702                  VerilogExtraKeywords.end();
1703     default:
1704       // getIdentifierInfo returns non-null for both identifiers and keywords.
1705       return Tok.Tok.getIdentifierInfo();
1706     }
1707   }
1708
1709   /// Returns whether \p Tok is a Verilog preprocessor directive.  This is
1710   /// needed because macro expansions start with a backtick as well and they
1711   /// need to be treated differently.
1712   bool isVerilogPPDirective(const FormatToken &Tok) const {
1713     auto Info = Tok.Tok.getIdentifierInfo();
1714     if (!Info)
1715       return false;
1716     switch (Info->getPPKeywordID()) {
1717     case tok::pp_define:
1718     case tok::pp_else:
1719     case tok::pp_endif:
1720     case tok::pp_ifdef:
1721     case tok::pp_ifndef:
1722     case tok::pp_include:
1723     case tok::pp_line:
1724     case tok::pp_pragma:
1725     case tok::pp_undef:
1726       return true;
1727     default:
1728       return Tok.isOneOf(kw_begin_keywords, kw_celldefine,
1729                          kw_default_decay_time, kw_default_nettype,
1730                          kw_default_trireg_strength, kw_delay_mode_distributed,
1731                          kw_delay_mode_path, kw_delay_mode_unit,
1732                          kw_delay_mode_zero, kw_elsif, kw_end_keywords,
1733                          kw_endcelldefine, kw_nounconnected_drive, kw_resetall,
1734                          kw_timescale, kw_unconnected_drive, kw_undefineall);
1735     }
1736   }
1737
1738   /// Returns whether \p Tok is a Verilog keyword that opens a block.
1739   bool isVerilogBegin(const FormatToken &Tok) const {
1740     // `table` is not included since it needs to be treated specially.
1741     return !Tok.endsSequence(kw_fork, kw_disable) &&
1742            Tok.isOneOf(kw_begin, kw_fork, kw_generate, kw_specify);
1743   }
1744
1745   /// Returns whether \p Tok is a Verilog keyword that closes a block.
1746   bool isVerilogEnd(const FormatToken &Tok) const {
1747     return !Tok.endsSequence(kw_join, kw_rand) &&
1748            Tok.isOneOf(TT_MacroBlockEnd, kw_end, kw_endcase, kw_endclass,
1749                        kw_endclocking, kw_endchecker, kw_endfunction,
1750                        kw_endgenerate, kw_endgroup, kw_endinterface,
1751                        kw_endmodule, kw_endpackage, kw_endprimitive,
1752                        kw_endprogram, kw_endproperty, kw_endsequence,
1753                        kw_endspecify, kw_endtable, kw_endtask, kw_join,
1754                        kw_join_any, kw_join_none);
1755   }
1756
1757   /// Returns whether \p Tok is a Verilog keyword that opens a module, etc.
1758   bool isVerilogHierarchy(const FormatToken &Tok) const {
1759     if (Tok.endsSequence(kw_function, kw_with))
1760       return false;
1761     if (Tok.is(kw_property)) {
1762       const FormatToken *Prev = Tok.getPreviousNonComment();
1763       return !(Prev &&
1764                Prev->isOneOf(tok::kw_restrict, kw_assert, kw_assume, kw_cover));
1765     }
1766     return Tok.isOneOf(tok::kw_case, tok::kw_class, kw_function, kw_module,
1767                        kw_interface, kw_package, kw_casex, kw_casez, kw_checker,
1768                        kw_clocking, kw_covergroup, kw_macromodule, kw_primitive,
1769                        kw_program, kw_property, kw_randcase, kw_randsequence,
1770                        kw_task);
1771   }
1772
1773   bool isVerilogEndOfLabel(const FormatToken &Tok) const {
1774     const FormatToken *Next = Tok.getNextNonComment();
1775     // In Verilog the colon in a default label is optional.
1776     return Tok.is(TT_CaseLabelColon) ||
1777            (Tok.is(tok::kw_default) &&
1778             !(Next && Next->isOneOf(tok::colon, tok::semi, kw_clocking, kw_iff,
1779                                     kw_input, kw_output, kw_sequence)));
1780   }
1781
1782   /// Returns whether \p Tok is a Verilog keyword that starts a
1783   /// structured procedure like 'always'.
1784   bool isVerilogStructuredProcedure(const FormatToken &Tok) const {
1785     return Tok.isOneOf(kw_always, kw_always_comb, kw_always_ff, kw_always_latch,
1786                        kw_final, kw_forever, kw_initial);
1787   }
1788
1789   bool isVerilogQualifier(const FormatToken &Tok) const {
1790     switch (Tok.Tok.getKind()) {
1791     case tok::kw_extern:
1792     case tok::kw_signed:
1793     case tok::kw_static:
1794     case tok::kw_unsigned:
1795     case tok::kw_virtual:
1796       return true;
1797     case tok::identifier:
1798       return Tok.isOneOf(
1799           kw_let, kw_var, kw_ref, kw_automatic, kw_bins, kw_coverpoint,
1800           kw_ignore_bins, kw_illegal_bins, kw_inout, kw_input, kw_interconnect,
1801           kw_local, kw_localparam, kw_output, kw_parameter, kw_pure, kw_rand,
1802           kw_randc, kw_scalared, kw_specparam, kw_tri, kw_tri0, kw_tri1,
1803           kw_triand, kw_trior, kw_trireg, kw_uwire, kw_vectored, kw_wand,
1804           kw_wildcard, kw_wire, kw_wor);
1805     default:
1806       return false;
1807     }
1808   }
1809
1810 private:
1811   /// The JavaScript keywords beyond the C++ keyword set.
1812   std::unordered_set<IdentifierInfo *> JsExtraKeywords;
1813
1814   /// The C# keywords beyond the C++ keyword set
1815   std::unordered_set<IdentifierInfo *> CSharpExtraKeywords;
1816
1817   /// The Verilog keywords beyond the C++ keyword set.
1818   std::unordered_set<IdentifierInfo *> VerilogExtraKeywords;
1819 };
1820
1821 inline bool isLineComment(const FormatToken &FormatTok) {
1822   return FormatTok.is(tok::comment) && !FormatTok.TokenText.startswith("/*");
1823 }
1824
1825 // Checks if \p FormatTok is a line comment that continues the line comment
1826 // \p Previous. The original column of \p MinColumnToken is used to determine
1827 // whether \p FormatTok is indented enough to the right to continue \p Previous.
1828 inline bool continuesLineComment(const FormatToken &FormatTok,
1829                                  const FormatToken *Previous,
1830                                  const FormatToken *MinColumnToken) {
1831   if (!Previous || !MinColumnToken)
1832     return false;
1833   unsigned MinContinueColumn =
1834       MinColumnToken->OriginalColumn + (isLineComment(*MinColumnToken) ? 0 : 1);
1835   return isLineComment(FormatTok) && FormatTok.NewlinesBefore == 1 &&
1836          isLineComment(*Previous) &&
1837          FormatTok.OriginalColumn >= MinContinueColumn;
1838 }
1839
1840 } // namespace format
1841 } // namespace clang
1842
1843 #endif