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