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 <unordered_set>
23
24namespace clang {
25namespace format {
26
27#define LIST_TOKEN_TYPES \
28 TYPE(AfterPPDirective) \
29 TYPE(ArrayInitializerLSquare) \
30 TYPE(ArraySubscriptLSquare) \
31 TYPE(AttributeColon) \
32 TYPE(AttributeLParen) \
33 TYPE(AttributeLSquare) \
34 TYPE(AttributeMacro) \
35 TYPE(AttributeRParen) \
36 TYPE(AttributeRSquare) \
37 TYPE(BinaryOperator) \
38 TYPE(BitFieldColon) \
39 TYPE(BlockComment) \
40 /* l_brace of a block that is not the body of a (e.g. loop) statement. */ \
41 TYPE(BlockLBrace) \
42 TYPE(BracedListLBrace) \
43 TYPE(CaseLabelArrow) \
44 /* The colon at the end of a case label. */ \
45 TYPE(CaseLabelColon) \
46 TYPE(CastRParen) \
47 TYPE(ClassLBrace) \
48 /* Name of class/struct/union/interface definition. */ \
49 TYPE(ClassHeadName) \
50 TYPE(ClassRBrace) \
51 TYPE(CompoundRequirementLBrace) \
52 /* ternary ?: expression */ \
53 TYPE(ConditionalExpr) \
54 /* the condition in an if statement */ \
55 TYPE(ConditionLParen) \
56 TYPE(ConflictAlternative) \
57 TYPE(ConflictEnd) \
58 TYPE(ConflictStart) \
59 /* l_brace of if/for/while/switch/catch */ \
60 TYPE(ControlStatementLBrace) \
61 TYPE(ControlStatementRBrace) \
62 TYPE(CppCastLParen) \
63 TYPE(CSharpGenericTypeConstraint) \
64 TYPE(CSharpGenericTypeConstraintColon) \
65 TYPE(CSharpGenericTypeConstraintComma) \
66 TYPE(CSharpNamedArgumentColon) \
67 TYPE(CSharpNullable) \
68 TYPE(CSharpNullConditionalLSquare) \
69 TYPE(CSharpStringLiteral) \
70 TYPE(CtorInitializerColon) \
71 TYPE(CtorInitializerComma) \
72 TYPE(CtorDtorDeclName) \
73 TYPE(DesignatedInitializerLSquare) \
74 TYPE(DesignatedInitializerPeriod) \
75 TYPE(DictLiteral) \
76 TYPE(DoWhile) \
77 TYPE(ElseLBrace) \
78 TYPE(ElseRBrace) \
79 TYPE(EnumLBrace) \
80 TYPE(EnumRBrace) \
81 TYPE(FatArrow) \
82 TYPE(ForEachMacro) \
83 TYPE(FunctionAnnotationRParen) \
84 TYPE(FunctionDeclarationName) \
85 TYPE(FunctionDeclarationLParen) \
86 TYPE(FunctionLBrace) \
87 TYPE(FunctionLikeMacro) \
88 TYPE(FunctionLikeOrFreestandingMacro) \
89 TYPE(FunctionTypeLParen) \
90 /* The colons as part of a C11 _Generic selection */ \
91 TYPE(GenericSelectionColon) \
92 /* The colon at the end of a goto label. */ \
93 TYPE(GotoLabelColon) \
94 TYPE(IfMacro) \
95 TYPE(ImplicitStringLiteral) \
96 TYPE(InheritanceColon) \
97 TYPE(InheritanceComma) \
98 TYPE(InlineASMBrace) \
99 TYPE(InlineASMColon) \
100 TYPE(InlineASMSymbolicNameLSquare) \
101 TYPE(JavaAnnotation) \
102 TYPE(JsAndAndEqual) \
103 TYPE(JsComputedPropertyName) \
104 TYPE(JsExponentiation) \
105 TYPE(JsExponentiationEqual) \
106 TYPE(JsPipePipeEqual) \
107 TYPE(JsPrivateIdentifier) \
108 TYPE(JsTypeColon) \
109 TYPE(JsTypeOperator) \
110 TYPE(JsTypeOptionalQuestion) \
111 TYPE(LambdaArrow) \
112 TYPE(LambdaDefinitionLParen) \
113 TYPE(LambdaLBrace) \
114 TYPE(LambdaLSquare) \
115 TYPE(LeadingJavaAnnotation) \
116 TYPE(LineComment) \
117 TYPE(MacroBlockBegin) \
118 TYPE(MacroBlockEnd) \
119 TYPE(ModulePartitionColon) \
120 TYPE(NamespaceLBrace) \
121 TYPE(NamespaceMacro) \
122 TYPE(NamespaceRBrace) \
123 TYPE(NonNullAssertion) \
124 TYPE(NullCoalescingEqual) \
125 TYPE(NullCoalescingOperator) \
126 TYPE(NullPropagatingOperator) \
127 TYPE(ObjCBlockLBrace) \
128 TYPE(ObjCBlockLParen) \
129 TYPE(ObjCDecl) \
130 TYPE(ObjCForIn) \
131 /* The square brackets surrounding a method call, the colon separating the \
132 * method or parameter name and the argument inside the square brackets, and \
133 * the colon separating the method or parameter name and the type inside the \
134 * method declaration. */ \
135 TYPE(ObjCMethodExpr) \
136 /* The '+' or '-' at the start of the line. */ \
137 TYPE(ObjCMethodSpecifier) \
138 TYPE(ObjCProperty) \
139 /* The parentheses following '@selector' and the colon following the method \
140 * or parameter name inside the parentheses. */ \
141 TYPE(ObjCSelector) \
142 TYPE(ObjCStringLiteral) \
143 TYPE(OverloadedOperator) \
144 TYPE(OverloadedOperatorLParen) \
145 TYPE(PointerOrReference) \
146 TYPE(ProtoExtensionLSquare) \
147 TYPE(PureVirtualSpecifier) \
148 TYPE(QtProperty) \
149 TYPE(RangeBasedForLoopColon) \
150 TYPE(RecordLBrace) \
151 TYPE(RecordRBrace) \
152 TYPE(RegexLiteral) \
153 TYPE(RequiresClause) \
154 TYPE(RequiresClauseInARequiresExpression) \
155 TYPE(RequiresExpression) \
156 TYPE(RequiresExpressionLBrace) \
157 TYPE(RequiresExpressionLParen) \
158 /* The hash key in languages that have hash literals, not including the \
159 * field name in the C++ struct literal. Also the method or parameter name \
160 * in the Objective-C method declaration or call. */ \
161 TYPE(SelectorName) \
162 TYPE(StartOfName) \
163 TYPE(StatementAttributeLikeMacro) \
164 TYPE(StatementMacro) \
165 /* A string that is part of a string concatenation. For C#, JavaScript, and \
166 * Java, it is used for marking whether a string needs parentheses around it \
167 * if it is to be split into parts joined by `+`. For Verilog, whether \
168 * braces need to be added to split it. Not used for other languages. */ \
169 TYPE(StringInConcatenation) \
170 TYPE(StructLBrace) \
171 TYPE(StructRBrace) \
172 TYPE(StructuredBindingLSquare) \
173 TYPE(SwitchExpressionLabel) \
174 TYPE(SwitchExpressionLBrace) \
175 TYPE(TableGenBangOperator) \
176 TYPE(TableGenCondOperator) \
177 TYPE(TableGenCondOperatorColon) \
178 TYPE(TableGenCondOperatorComma) \
179 TYPE(TableGenDAGArgCloser) \
180 TYPE(TableGenDAGArgListColon) \
181 TYPE(TableGenDAGArgListColonToAlign) \
182 TYPE(TableGenDAGArgListComma) \
183 TYPE(TableGenDAGArgListCommaToBreak) \
184 TYPE(TableGenDAGArgOpener) \
185 TYPE(TableGenDAGArgOpenerToBreak) \
186 TYPE(TableGenDAGArgOperatorID) \
187 TYPE(TableGenDAGArgOperatorToBreak) \
188 TYPE(TableGenListCloser) \
189 TYPE(TableGenListOpener) \
190 TYPE(TableGenMultiLineString) \
191 TYPE(TableGenTrailingPasteOperator) \
192 TYPE(TableGenValueSuffix) \
193 TYPE(TemplateCloser) \
194 TYPE(TemplateOpener) \
195 TYPE(TemplateString) \
196 TYPE(TrailingAnnotation) \
197 TYPE(TrailingReturnArrow) \
198 TYPE(TrailingUnaryOperator) \
199 TYPE(TypeDeclarationParen) \
200 TYPE(TemplateName) \
201 TYPE(TypeName) \
202 TYPE(TypenameMacro) \
203 TYPE(UnaryOperator) \
204 TYPE(UnionLBrace) \
205 TYPE(UnionRBrace) \
206 TYPE(UntouchableMacroFunc) \
207 TYPE(VariableTemplate) \
208 /* Like in 'assign x = 0, y = 1;' . */ \
209 TYPE(VerilogAssignComma) \
210 /* like in begin : block */ \
211 TYPE(VerilogBlockLabelColon) \
212 /* The square bracket for the dimension part of the type name. \
213 * In 'logic [1:0] x[1:0]', only the first '['. This way we can have space \
214 * before the first bracket but not the second. */ \
215 TYPE(VerilogDimensionedTypeName) \
216 /* list of port connections or parameters in a module instantiation */ \
217 TYPE(VerilogInstancePortComma) \
218 TYPE(VerilogInstancePortLParen) \
219 /* A parenthesized list within which line breaks are inserted by the \
220 * formatter, for example the list of ports in a module header. */ \
221 TYPE(VerilogMultiLineListLParen) \
222 /* for the base in a number literal, not including the quote */ \
223 TYPE(VerilogNumberBase) \
224 /* like `(strong1, pull0)` */ \
225 TYPE(VerilogStrength) \
226 /* Things inside the table in user-defined primitives. */ \
227 TYPE(VerilogTableItem) \
228 /* those that separate ports of different types */ \
229 TYPE(VerilogTypeComma) \
230 TYPE(Unknown)
231
232/// Determines the semantic type of a syntactic token, e.g. whether "<" is a
233/// template opener or binary operator.
234enum TokenType : uint8_t {
235#define TYPE(X) TT_##X,
236 LIST_TOKEN_TYPES
237#undef TYPE
238 NUM_TOKEN_TYPES
239};
240
241/// Determines the name of a token type.
242const char *getTokenTypeName(TokenType Type);
243
244// Represents what type of block a set of braces open.
245enum BraceBlockKind { BK_Unknown, BK_Block, BK_BracedInit };
246
247// The packing kind of a function's parameters.
248enum ParameterPackingKind { PPK_BinPacked, PPK_OnePerLine, PPK_Inconclusive };
249
250enum FormatDecision { FD_Unformatted, FD_Continue, FD_Break };
251
252/// Roles a token can take in a configured macro expansion.
253enum MacroRole {
254 /// The token was expanded from a macro argument when formatting the expanded
255 /// token sequence.
256 MR_ExpandedArg,
257 /// The token is part of a macro argument that was previously formatted as
258 /// expansion when formatting the unexpanded macro call.
259 MR_UnexpandedArg,
260 /// The token was expanded from a macro definition, and is not visible as part
261 /// of the macro call.
262 MR_Hidden,
263};
264
265struct FormatToken;
266
267/// Contains information on the token's role in a macro expansion.
268///
269/// Given the following definitions:
270/// A(X) = [ X ]
271/// B(X) = < X >
272/// C(X) = X
273///
274/// Consider the macro call:
275/// A({B(C(C(x)))}) -> [{<x>}]
276///
277/// In this case, the tokens of the unexpanded macro call will have the
278/// following relevant entries in their macro context (note that formatting
279/// the unexpanded macro call happens *after* formatting the expanded macro
280/// call):
281/// A( { B( C( C(x) ) ) } )
282/// Role: NN U NN NN NNUN N N U N (N=None, U=UnexpandedArg)
283///
284/// [ { < x > } ]
285/// Role: H E H E H E H (H=Hidden, E=ExpandedArg)
286/// ExpandedFrom[0]: A A A A A A A
287/// ExpandedFrom[1]: B B B
288/// ExpandedFrom[2]: C
289/// ExpandedFrom[3]: C
290/// StartOfExpansion: 1 0 1 2 0 0 0
291/// EndOfExpansion: 0 0 0 2 1 0 1
292struct MacroExpansion {
293 MacroExpansion(MacroRole Role) : Role(Role) {}
294
295 /// The token's role in the macro expansion.
296 /// When formatting an expanded macro, all tokens that are part of macro
297 /// arguments will be MR_ExpandedArg, while all tokens that are not visible in
298 /// the macro call will be MR_Hidden.
299 /// When formatting an unexpanded macro call, all tokens that are part of
300 /// macro arguments will be MR_UnexpandedArg.
301 MacroRole Role;
302
303 /// The stack of macro call identifier tokens this token was expanded from.
304 llvm::SmallVector<FormatToken *, 1> ExpandedFrom;
305
306 /// The number of expansions of which this macro is the first entry.
307 unsigned StartOfExpansion = 0;
308
309 /// The number of currently open expansions in \c ExpandedFrom this macro is
310 /// the last token in.
311 unsigned EndOfExpansion = 0;
312};
313
314class TokenRole;
315class AnnotatedLine;
316
317/// A wrapper around a \c Token storing information about the
318/// whitespace characters preceding it.
319struct FormatToken {
320 FormatToken()
321 : HasUnescapedNewline(false), IsMultiline(false), IsFirst(false),
322 MustBreakBefore(false), MustBreakBeforeFinalized(false),
323 IsUnterminatedLiteral(false), CanBreakBefore(false),
324 ClosesTemplateDeclaration(false), StartsBinaryExpression(false),
325 EndsBinaryExpression(false), PartOfMultiVariableDeclStmt(false),
326 ContinuesLineCommentSection(false), Finalized(false),
327 ClosesRequiresClause(false), EndsCppAttributeGroup(false),
328 BlockKind(BK_Unknown), Decision(FD_Unformatted),
329 PackingKind(PPK_Inconclusive), TypeIsFinalized(false),
330 Type(TT_Unknown) {}
331
332 /// The \c Token.
333 Token Tok;
334
335 /// The raw text of the token.
336 ///
337 /// Contains the raw token text without leading whitespace and without leading
338 /// escaped newlines.
339 StringRef TokenText;
340
341 /// A token can have a special role that can carry extra information
342 /// about the token's formatting.
343 /// FIXME: Make FormatToken for parsing and AnnotatedToken two different
344 /// classes and make this a unique_ptr in the AnnotatedToken class.
345 std::shared_ptr<TokenRole> Role;
346
347 /// The range of the whitespace immediately preceding the \c Token.
348 SourceRange WhitespaceRange;
349
350 /// Whether there is at least one unescaped newline before the \c
351 /// Token.
352 unsigned HasUnescapedNewline : 1;
353
354 /// Whether the token text contains newlines (escaped or not).
355 unsigned IsMultiline : 1;
356
357 /// Indicates that this is the first token of the file.
358 unsigned IsFirst : 1;
359
360 /// Whether there must be a line break before this token.
361 ///
362 /// This happens for example when a preprocessor directive ended directly
363 /// before the token.
364 unsigned MustBreakBefore : 1;
365
366 /// Whether MustBreakBefore is finalized during parsing and must not
367 /// be reset between runs.
368 unsigned MustBreakBeforeFinalized : 1;
369
370 /// Set to \c true if this token is an unterminated literal.
371 unsigned IsUnterminatedLiteral : 1;
372
373 /// \c true if it is allowed to break before this token.
374 unsigned CanBreakBefore : 1;
375
376 /// \c true if this is the ">" of "template<..>".
377 unsigned ClosesTemplateDeclaration : 1;
378
379 /// \c true if this token starts a binary expression, i.e. has at least
380 /// one fake l_paren with a precedence greater than prec::Unknown.
381 unsigned StartsBinaryExpression : 1;
382 /// \c true if this token ends a binary expression.
383 unsigned EndsBinaryExpression : 1;
384
385 /// Is this token part of a \c DeclStmt defining multiple variables?
386 ///
387 /// Only set if \c Type == \c TT_StartOfName.
388 unsigned PartOfMultiVariableDeclStmt : 1;
389
390 /// Does this line comment continue a line comment section?
391 ///
392 /// Only set to true if \c Type == \c TT_LineComment.
393 unsigned ContinuesLineCommentSection : 1;
394
395 /// If \c true, this token has been fully formatted (indented and
396 /// potentially re-formatted inside), and we do not allow further formatting
397 /// changes.
398 unsigned Finalized : 1;
399
400 /// \c true if this is the last token within requires clause.
401 unsigned ClosesRequiresClause : 1;
402
403 /// \c true if this token ends a group of C++ attributes.
404 unsigned EndsCppAttributeGroup : 1;
405
406private:
407 /// Contains the kind of block if this token is a brace.
408 unsigned BlockKind : 2;
409
410public:
411 BraceBlockKind getBlockKind() const {
412 return static_cast<BraceBlockKind>(BlockKind);
413 }
414 void setBlockKind(BraceBlockKind BBK) {
415 BlockKind = BBK;
416 assert(getBlockKind() == BBK && "BraceBlockKind overflow!");
417 }
418
419private:
420 /// Stores the formatting decision for the token once it was made.
421 unsigned Decision : 2;
422
423public:
424 FormatDecision getDecision() const {
425 return static_cast<FormatDecision>(Decision);
426 }
427 void setDecision(FormatDecision D) {
428 Decision = D;
429 assert(getDecision() == D && "FormatDecision overflow!");
430 }
431
432private:
433 /// If this is an opening parenthesis, how are the parameters packed?
434 unsigned PackingKind : 2;
435
436public:
437 ParameterPackingKind getPackingKind() const {
438 return static_cast<ParameterPackingKind>(PackingKind);
439 }
440 void setPackingKind(ParameterPackingKind K) {
441 PackingKind = K;
442 assert(getPackingKind() == K && "ParameterPackingKind overflow!");
443 }
444
445private:
446 unsigned TypeIsFinalized : 1;
447 TokenType Type;
448
449public:
450 /// Returns the token's type, e.g. whether "<" is a template opener or
451 /// binary operator.
452 TokenType getType() const { return Type; }
453 void setType(TokenType T) {
454 // If this token is a macro argument while formatting an unexpanded macro
455 // call, we do not change its type any more - the type was deduced from
456 // formatting the expanded macro stream already.
457 if (MacroCtx && MacroCtx->Role == MR_UnexpandedArg)
458 return;
459 assert((!TypeIsFinalized || T == Type) &&
460 "Please use overwriteFixedType to change a fixed type.");
461 Type = T;
462 }
463 /// Sets the type and also the finalized flag. This prevents the type to be
464 /// reset in TokenAnnotator::resetTokenMetadata(). If the type needs to be set
465 /// to another one please use overwriteFixedType, or even better remove the
466 /// need to reassign the type.
467 void setFinalizedType(TokenType T) {
468 if (MacroCtx && MacroCtx->Role == MR_UnexpandedArg)
469 return;
470 Type = T;
471 TypeIsFinalized = true;
472 }
473 void overwriteFixedType(TokenType T) {
474 if (MacroCtx && MacroCtx->Role == MR_UnexpandedArg)
475 return;
476 TypeIsFinalized = false;
477 setType(T);
478 }
479 bool isTypeFinalized() const { return TypeIsFinalized; }
480
481 /// Used to set an operator precedence explicitly.
482 prec::Level ForcedPrecedence = prec::Unknown;
483
484 /// The number of newlines immediately before the \c Token.
485 ///
486 /// This can be used to determine what the user wrote in the original code
487 /// and thereby e.g. leave an empty line between two function definitions.
488 unsigned NewlinesBefore = 0;
489
490 /// The number of newlines immediately before the \c Token after formatting.
491 ///
492 /// This is used to avoid overlapping whitespace replacements when \c Newlines
493 /// is recomputed for a finalized preprocessor branching directive.
494 int Newlines = -1;
495
496 /// The offset just past the last '\n' in this token's leading
497 /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'.
498 unsigned LastNewlineOffset = 0;
499
500 /// The width of the non-whitespace parts of the token (or its first
501 /// line for multi-line tokens) in columns.
502 /// We need this to correctly measure number of columns a token spans.
503 unsigned ColumnWidth = 0;
504
505 /// Contains the width in columns of the last line of a multi-line
506 /// token.
507 unsigned LastLineColumnWidth = 0;
508
509 /// The number of spaces that should be inserted before this token.
510 unsigned SpacesRequiredBefore = 0;
511
512 /// Number of parameters, if this is "(", "[" or "<".
513 unsigned ParameterCount = 0;
514
515 /// Number of parameters that are nested blocks,
516 /// if this is "(", "[" or "<".
517 unsigned BlockParameterCount = 0;
518
519 /// If this is a bracket ("<", "(", "[" or "{"), contains the kind of
520 /// the surrounding bracket.
521 tok::TokenKind ParentBracket = tok::unknown;
522
523 /// The total length of the unwrapped line up to and including this
524 /// token.
525 unsigned TotalLength = 0;
526
527 /// The original 0-based column of this token, including expanded tabs.
528 /// The configured TabWidth is used as tab width.
529 unsigned OriginalColumn = 0;
530
531 /// The length of following tokens until the next natural split point,
532 /// or the next token that can be broken.
533 unsigned UnbreakableTailLength = 0;
534
535 // FIXME: Come up with a 'cleaner' concept.
536 /// The binding strength of a token. This is a combined value of
537 /// operator precedence, parenthesis nesting, etc.
538 unsigned BindingStrength = 0;
539
540 /// The nesting level of this token, i.e. the number of surrounding (),
541 /// [], {} or <>.
542 unsigned NestingLevel = 0;
543
544 /// The indent level of this token. Copied from the surrounding line.
545 unsigned IndentLevel = 0;
546
547 /// Penalty for inserting a line break before this token.
548 unsigned SplitPenalty = 0;
549
550 /// If this is the first ObjC selector name in an ObjC method
551 /// definition or call, this contains the length of the longest name.
552 ///
553 /// This being set to 0 means that the selectors should not be colon-aligned,
554 /// e.g. because several of them are block-type.
555 unsigned LongestObjCSelectorName = 0;
556
557 /// If this is the first ObjC selector name in an ObjC method
558 /// definition or call, this contains the number of parts that the whole
559 /// selector consist of.
560 unsigned ObjCSelectorNameParts = 0;
561
562 /// The 0-based index of the parameter/argument. For ObjC it is set
563 /// for the selector name token.
564 /// For now calculated only for ObjC.
565 unsigned ParameterIndex = 0;
566
567 /// Stores the number of required fake parentheses and the
568 /// corresponding operator precedence.
569 ///
570 /// If multiple fake parentheses start at a token, this vector stores them in
571 /// reverse order, i.e. inner fake parenthesis first.
572 SmallVector<prec::Level, 4> FakeLParens;
573 /// Insert this many fake ) after this token for correct indentation.
574 unsigned FakeRParens = 0;
575
576 /// If this is an operator (or "."/"->") in a sequence of operators
577 /// with the same precedence, contains the 0-based operator index.
578 unsigned OperatorIndex = 0;
579
580 /// If this is an operator (or "."/"->") in a sequence of operators
581 /// with the same precedence, points to the next operator.
582 FormatToken *NextOperator = nullptr;
583
584 /// If this is a bracket, this points to the matching one.
585 FormatToken *MatchingParen = nullptr;
586
587 /// The previous token in the unwrapped line.
588 FormatToken *Previous = nullptr;
589
590 /// The next token in the unwrapped line.
591 FormatToken *Next = nullptr;
592
593 /// The first token in set of column elements.
594 bool StartsColumn = false;
595
596 /// This notes the start of the line of an array initializer.
597 bool ArrayInitializerLineStart = false;
598
599 /// This starts an array initializer.
600 bool IsArrayInitializer = false;
601
602 /// Is optional and can be removed.
603 bool Optional = false;
604
605 /// Might be function declaration open/closing paren.
606 bool MightBeFunctionDeclParen = false;
607
608 /// Has "\n\f\n" or "\n\f\r\n" before TokenText.
609 bool HasFormFeedBefore = false;
610
611 /// Is the first token after a preprocessor line.
612 bool FirstAfterPPLine = false;
613
614 /// Number of optional braces to be inserted after this token:
615 /// -1: a single left brace
616 /// 0: no braces
617 /// >0: number of right braces
618 int8_t BraceCount = 0;
619
620 /// If this token starts a block, this contains all the unwrapped lines
621 /// in it.
622 SmallVector<AnnotatedLine *, 1> Children;
623
624 // Contains all attributes related to how this token takes part
625 // in a configured macro expansion.
626 std::optional<MacroExpansion> MacroCtx;
627
628 /// When macro expansion introduces nodes with children, those are marked as
629 /// \c MacroParent.
630 /// FIXME: The formatting code currently hard-codes the assumption that
631 /// child nodes are introduced by blocks following an opening brace.
632 /// This is deeply baked into the code and disentangling this will require
633 /// signficant refactorings. \c MacroParent allows us to special-case the
634 /// cases in which we treat parents as block-openers for now.
635 bool MacroParent = false;
636
637 bool is(tok::TokenKind Kind) const { return Tok.is(K: Kind); }
638 bool is(tok::ObjCKeywordKind Kind) const {
639 return Tok.getObjCKeywordID() == Kind;
640 }
641 bool is(TokenType TT) const { return getType() == TT; }
642 bool is(const IdentifierInfo *II) const {
643 return II && II == Tok.getIdentifierInfo();
644 }
645 bool is(tok::PPKeywordKind Kind) const {
646 return Tok.getIdentifierInfo() &&
647 Tok.getIdentifierInfo()->getPPKeywordID() == Kind;
648 }
649 bool is(BraceBlockKind BBK) const { return getBlockKind() == BBK; }
650 bool is(ParameterPackingKind PPK) const { return getPackingKind() == PPK; }
651
652 template <typename A, typename B> bool isOneOf(A K1, B K2) const {
653 return is(K1) || is(K2);
654 }
655 template <typename A, typename B, typename... Ts>
656 bool isOneOf(A K1, B K2, Ts... Ks) const {
657 return is(K1) || isOneOf(K2, Ks...);
658 }
659 template <typename T> bool isNot(T Kind) const { return !is(Kind); }
660 template <typename... Ts> bool isNoneOf(Ts... Ks) const {
661 return !isOneOf(Ks...);
662 }
663
664 bool isIf(bool AllowConstexprMacro = true) const {
665 return is(Kind: tok::kw_if) || endsSequence(K1: tok::kw_constexpr, Tokens: tok::kw_if) ||
666 (endsSequence(K1: tok::identifier, Tokens: tok::kw_if) && AllowConstexprMacro);
667 }
668
669 bool isLoop(const FormatStyle &Style) const {
670 return isOneOf(K1: tok::kw_for, K2: tok::kw_while) ||
671 (Style.isJavaScript() && isNot(Kind: tok::l_paren) && Previous &&
672 Previous->is(Kind: tok::kw_for));
673 }
674
675 bool closesScopeAfterBlock() const {
676 if (getBlockKind() == BK_Block)
677 return true;
678 if (closesScope())
679 return Previous->closesScopeAfterBlock();
680 return false;
681 }
682
683 /// \c true if this token starts a sequence with the given tokens in order,
684 /// following the ``Next`` pointers, ignoring comments.
685 template <typename A, typename... Ts>
686 bool startsSequence(A K1, Ts... Tokens) const {
687 return startsSequenceInternal(K1, Tokens...);
688 }
689
690 /// \c true if this token ends a sequence with the given tokens in order,
691 /// following the ``Previous`` pointers, ignoring comments.
692 /// For example, given tokens [T1, T2, T3], the function returns true if
693 /// 3 tokens ending at this (ignoring comments) are [T3, T2, T1]. In other
694 /// words, the tokens passed to this function need to the reverse of the
695 /// order the tokens appear in code.
696 template <typename A, typename... Ts>
697 bool endsSequence(A K1, Ts... Tokens) const {
698 return endsSequenceInternal(K1, Tokens...);
699 }
700
701 bool isStringLiteral() const { return tok::isStringLiteral(K: Tok.getKind()); }
702
703 bool isAttribute() const {
704 return isOneOf(K1: tok::kw___attribute, K2: tok::kw___declspec, Ks: TT_AttributeMacro);
705 }
706
707 bool isAccessSpecifierKeyword() const {
708 return isOneOf(K1: tok::kw_public, K2: tok::kw_protected, Ks: tok::kw_private);
709 }
710
711 bool isAccessSpecifier(bool ColonRequired = true) const {
712 if (!isAccessSpecifierKeyword())
713 return false;
714 if (!ColonRequired)
715 return true;
716 const auto *NextNonComment = getNextNonComment();
717 return NextNonComment && NextNonComment->is(Kind: tok::colon);
718 }
719
720 bool canBePointerOrReferenceQualifier() const {
721 return isOneOf(K1: tok::kw_const, K2: tok::kw_restrict, Ks: tok::kw_volatile,
722 Ks: tok::kw__Nonnull, Ks: tok::kw__Nullable,
723 Ks: tok::kw__Null_unspecified, Ks: tok::kw___ptr32, Ks: tok::kw___ptr64,
724 Ks: tok::kw___funcref) ||
725 isAttribute();
726 }
727
728 [[nodiscard]] bool isQtProperty() const;
729 [[nodiscard]] bool isTypeName(const LangOptions &LangOpts) const;
730 [[nodiscard]] bool isTypeOrIdentifier(const LangOptions &LangOpts) const;
731
732 bool isObjCAccessSpecifier() const {
733 return is(Kind: tok::at) && Next &&
734 Next->isOneOf(K1: tok::objc_public, K2: tok::objc_protected,
735 Ks: tok::objc_package, Ks: tok::objc_private);
736 }
737
738 bool isObjCLifetimeQualifier(const FormatStyle &Style) const {
739 if (Style.Language != FormatStyle::LK_ObjC || isNot(Kind: tok::identifier) ||
740 !TokenText.starts_with(Prefix: "__")) {
741 return false;
742 }
743 const auto Qualifier = TokenText.substr(Start: 2);
744 return Qualifier == "autoreleasing" || Qualifier == "strong" ||
745 Qualifier == "weak" || Qualifier == "unsafe_unretained";
746 }
747
748 /// Returns whether \p Tok is ([{ or an opening < of a template or in
749 /// protos.
750 bool opensScope() const {
751 if (is(TT: TT_TemplateString) && TokenText.ends_with(Suffix: "${"))
752 return true;
753 if (is(TT: TT_DictLiteral) && is(Kind: tok::less))
754 return true;
755 return isOneOf(K1: tok::l_paren, K2: tok::l_brace, Ks: tok::l_square,
756 Ks: TT_TemplateOpener);
757 }
758 /// Returns whether \p Tok is )]} or a closing > of a template or in
759 /// protos.
760 bool closesScope() const {
761 if (is(TT: TT_TemplateString) && TokenText.starts_with(Prefix: "}"))
762 return true;
763 if (is(TT: TT_DictLiteral) && is(Kind: tok::greater))
764 return true;
765 return isOneOf(K1: tok::r_paren, K2: tok::r_brace, Ks: tok::r_square,
766 Ks: TT_TemplateCloser);
767 }
768
769 /// Returns \c true if this is a "." or "->" accessing a member.
770 bool isMemberAccess() const {
771 return isOneOf(K1: tok::arrow, K2: tok::period, Ks: tok::arrowstar) &&
772 isNoneOf(Ks: TT_DesignatedInitializerPeriod, Ks: TT_TrailingReturnArrow,
773 Ks: TT_LambdaArrow, Ks: TT_LeadingJavaAnnotation);
774 }
775
776 bool isPointerOrReference() const {
777 return isOneOf(K1: tok::star, K2: tok::amp, Ks: tok::ampamp);
778 }
779
780 bool isPlacementOperator() const {
781 return isOneOf(K1: tok::kw_new, K2: tok::kw_delete);
782 }
783
784 bool isUnaryOperator() const {
785 switch (Tok.getKind()) {
786 case tok::plus:
787 case tok::plusplus:
788 case tok::minus:
789 case tok::minusminus:
790 case tok::exclaim:
791 case tok::tilde:
792 case tok::kw_sizeof:
793 case tok::kw_alignof:
794 return true;
795 default:
796 return false;
797 }
798 }
799
800 bool isBinaryOperator() const {
801 // Comma is a binary operator, but does not behave as such wrt. formatting.
802 return getPrecedence() > prec::Comma;
803 }
804
805 bool isTrailingComment() const {
806 return is(Kind: tok::comment) &&
807 (is(TT: TT_LineComment) || !Next || Next->NewlinesBefore > 0);
808 }
809
810 /// Returns \c true if this is a keyword that can be used
811 /// like a function call (e.g. sizeof, typeid, ...).
812 bool isFunctionLikeKeyword() const {
813 if (isAttribute())
814 return true;
815
816 return isOneOf(K1: tok::kw_throw, K2: tok::kw_typeid, Ks: tok::kw_return,
817 Ks: tok::kw_sizeof, Ks: tok::kw_alignof, Ks: tok::kw_alignas,
818 Ks: tok::kw_decltype, Ks: tok::kw_noexcept, Ks: tok::kw_static_assert,
819 Ks: tok::kw__Atomic,
820#define TRANSFORM_TYPE_TRAIT_DEF(_, Trait) tok::kw___##Trait,
821#include "clang/Basic/TransformTypeTraits.def"
822 Ks: tok::kw_requires);
823 }
824
825 /// Returns \c true if this is a string literal that's like a label,
826 /// e.g. ends with "=" or ":".
827 bool isLabelString() const {
828 if (isNot(Kind: tok::string_literal))
829 return false;
830 StringRef Content = TokenText;
831 if (Content.starts_with(Prefix: "\"") || Content.starts_with(Prefix: "'"))
832 Content = Content.drop_front(N: 1);
833 if (Content.ends_with(Suffix: "\"") || Content.ends_with(Suffix: "'"))
834 Content = Content.drop_back(N: 1);
835 Content = Content.trim();
836 return Content.size() > 1 &&
837 (Content.back() == ':' || Content.back() == '=');
838 }
839
840 /// Returns actual token start location without leading escaped
841 /// newlines and whitespace.
842 ///
843 /// This can be different to Tok.getLocation(), which includes leading escaped
844 /// newlines.
845 SourceLocation getStartOfNonWhitespace() const {
846 return WhitespaceRange.getEnd();
847 }
848
849 /// Returns \c true if the range of whitespace immediately preceding the \c
850 /// Token is not empty.
851 bool hasWhitespaceBefore() const {
852 return WhitespaceRange.getBegin() != WhitespaceRange.getEnd();
853 }
854
855 prec::Level getPrecedence() const {
856 if (ForcedPrecedence != prec::Unknown)
857 return ForcedPrecedence;
858 return getBinOpPrecedence(Kind: Tok.getKind(), /*GreaterThanIsOperator=*/GreaterThanIsOperator: true,
859 /*CPlusPlus11=*/CPlusPlus11: true);
860 }
861
862 /// Returns the previous token ignoring comments.
863 [[nodiscard]] FormatToken *getPreviousNonComment() const {
864 FormatToken *Tok = Previous;
865 while (Tok && Tok->is(Kind: tok::comment))
866 Tok = Tok->Previous;
867 return Tok;
868 }
869
870 /// Returns the next token ignoring comments.
871 [[nodiscard]] FormatToken *getNextNonComment() const {
872 FormatToken *Tok = Next;
873 while (Tok && Tok->is(Kind: tok::comment))
874 Tok = Tok->Next;
875 return Tok;
876 }
877
878 /// Returns \c true if this token ends a block indented initializer list.
879 [[nodiscard]] bool isBlockIndentedInitRBrace(const FormatStyle &Style) const;
880
881 /// Returns \c true if this tokens starts a block-type list, i.e. a
882 /// list that should be indented with a block indent.
883 [[nodiscard]] bool opensBlockOrBlockTypeList(const FormatStyle &Style) const;
884
885 /// Returns whether the token is the left square bracket of a C++
886 /// structured binding declaration.
887 bool isCppStructuredBinding(bool IsCpp) const {
888 if (!IsCpp || isNot(Kind: tok::l_square))
889 return false;
890 const FormatToken *T = this;
891 do {
892 T = T->getPreviousNonComment();
893 } while (T && T->isOneOf(K1: tok::kw_const, K2: tok::kw_volatile, Ks: tok::amp,
894 Ks: tok::ampamp));
895 return T && T->is(Kind: tok::kw_auto);
896 }
897
898 /// Same as opensBlockOrBlockTypeList, but for the closing token.
899 bool closesBlockOrBlockTypeList(const FormatStyle &Style) const {
900 if (is(TT: TT_TemplateString) && closesScope())
901 return true;
902 return MatchingParen && MatchingParen->opensBlockOrBlockTypeList(Style);
903 }
904
905 /// Return the actual namespace token, if this token starts a namespace
906 /// block.
907 const FormatToken *getNamespaceToken() const {
908 const FormatToken *NamespaceTok = this;
909 if (is(Kind: tok::comment))
910 NamespaceTok = NamespaceTok->getNextNonComment();
911 // Detect "(inline|export)? namespace" in the beginning of a line.
912 if (NamespaceTok && NamespaceTok->isOneOf(K1: tok::kw_inline, K2: tok::kw_export))
913 NamespaceTok = NamespaceTok->getNextNonComment();
914 return NamespaceTok &&
915 NamespaceTok->isOneOf(K1: tok::kw_namespace, K2: TT_NamespaceMacro)
916 ? NamespaceTok
917 : nullptr;
918 }
919
920 void copyFrom(const FormatToken &Tok) { *this = Tok; }
921
922private:
923 // Only allow copying via the explicit copyFrom method.
924 FormatToken(const FormatToken &) = delete;
925 FormatToken &operator=(const FormatToken &) = default;
926
927 template <typename A, typename... Ts>
928 bool startsSequenceInternal(A K1, Ts... Tokens) const {
929 if (is(Kind: tok::comment) && Next)
930 return Next->startsSequenceInternal(K1, Tokens...);
931 return is(K1) && Next && Next->startsSequenceInternal(Tokens...);
932 }
933
934 template <typename A> bool startsSequenceInternal(A K1) const {
935 if (is(Kind: tok::comment) && Next)
936 return Next->startsSequenceInternal(K1);
937 return is(K1);
938 }
939
940 template <typename A, typename... Ts> bool endsSequenceInternal(A K1) const {
941 if (is(Kind: tok::comment) && Previous)
942 return Previous->endsSequenceInternal(K1);
943 return is(K1);
944 }
945
946 template <typename A, typename... Ts>
947 bool endsSequenceInternal(A K1, Ts... Tokens) const {
948 if (is(Kind: tok::comment) && Previous)
949 return Previous->endsSequenceInternal(K1, Tokens...);
950 return is(K1) && Previous && Previous->endsSequenceInternal(Tokens...);
951 }
952};
953
954class ContinuationIndenter;
955struct LineState;
956
957class TokenRole {
958public:
959 TokenRole(const FormatStyle &Style) : Style(Style) {}
960 virtual ~TokenRole();
961
962 /// After the \c TokenAnnotator has finished annotating all the tokens,
963 /// this function precomputes required information for formatting.
964 virtual void precomputeFormattingInfos(const FormatToken *Token);
965
966 /// Apply the special formatting that the given role demands.
967 ///
968 /// Assumes that the token having this role is already formatted.
969 ///
970 /// Continues formatting from \p State leaving indentation to \p Indenter and
971 /// returns the total penalty that this formatting incurs.
972 virtual unsigned formatFromToken(LineState &State,
973 ContinuationIndenter *Indenter,
974 bool DryRun) {
975 return 0;
976 }
977
978 /// Same as \c formatFromToken, but assumes that the first token has
979 /// already been set thereby deciding on the first line break.
980 virtual unsigned formatAfterToken(LineState &State,
981 ContinuationIndenter *Indenter,
982 bool DryRun) {
983 return 0;
984 }
985
986 /// Notifies the \c Role that a comma was found.
987 virtual void CommaFound(const FormatToken *Token) {}
988
989 virtual const FormatToken *lastComma() { return nullptr; }
990
991protected:
992 const FormatStyle &Style;
993};
994
995class CommaSeparatedList : public TokenRole {
996public:
997 CommaSeparatedList(const FormatStyle &Style)
998 : TokenRole(Style), HasNestedBracedList(false) {}
999
1000 void precomputeFormattingInfos(const FormatToken *Token) override;
1001
1002 unsigned formatAfterToken(LineState &State, ContinuationIndenter *Indenter,
1003 bool DryRun) override;
1004
1005 unsigned formatFromToken(LineState &State, ContinuationIndenter *Indenter,
1006 bool DryRun) override;
1007
1008 /// Adds \p Token as the next comma to the \c CommaSeparated list.
1009 void CommaFound(const FormatToken *Token) override {
1010 Commas.push_back(Elt: Token);
1011 }
1012
1013 const FormatToken *lastComma() override {
1014 if (Commas.empty())
1015 return nullptr;
1016 return Commas.back();
1017 }
1018
1019private:
1020 /// A struct that holds information on how to format a given list with
1021 /// a specific number of columns.
1022 struct ColumnFormat {
1023 /// The number of columns to use.
1024 unsigned Columns;
1025
1026 /// The total width in characters.
1027 unsigned TotalWidth;
1028
1029 /// The number of lines required for this format.
1030 unsigned LineCount;
1031
1032 /// The size of each column in characters.
1033 SmallVector<unsigned, 8> ColumnSizes;
1034 };
1035
1036 /// Calculate which \c ColumnFormat fits best into
1037 /// \p RemainingCharacters.
1038 const ColumnFormat *getColumnFormat(unsigned RemainingCharacters) const;
1039
1040 /// The ordered \c FormatTokens making up the commas of this list.
1041 SmallVector<const FormatToken *, 8> Commas;
1042
1043 /// The length of each of the list's items in characters including the
1044 /// trailing comma.
1045 SmallVector<unsigned, 8> ItemLengths;
1046
1047 /// Precomputed formats that can be used for this list.
1048 SmallVector<ColumnFormat, 4> Formats;
1049
1050 bool HasNestedBracedList;
1051};
1052
1053/// Encapsulates keywords that are context sensitive or for languages not
1054/// properly supported by Clang's lexer.
1055struct AdditionalKeywords {
1056 AdditionalKeywords(IdentifierTable &IdentTable) {
1057 kw_final = &IdentTable.get(Name: "final");
1058 kw_override = &IdentTable.get(Name: "override");
1059 kw_in = &IdentTable.get(Name: "in");
1060 kw_of = &IdentTable.get(Name: "of");
1061 kw_CF_CLOSED_ENUM = &IdentTable.get(Name: "CF_CLOSED_ENUM");
1062 kw_CF_ENUM = &IdentTable.get(Name: "CF_ENUM");
1063 kw_CF_OPTIONS = &IdentTable.get(Name: "CF_OPTIONS");
1064 kw_NS_CLOSED_ENUM = &IdentTable.get(Name: "NS_CLOSED_ENUM");
1065 kw_NS_ENUM = &IdentTable.get(Name: "NS_ENUM");
1066 kw_NS_ERROR_ENUM = &IdentTable.get(Name: "NS_ERROR_ENUM");
1067 kw_NS_OPTIONS = &IdentTable.get(Name: "NS_OPTIONS");
1068
1069 kw_as = &IdentTable.get(Name: "as");
1070 kw_async = &IdentTable.get(Name: "async");
1071 kw_await = &IdentTable.get(Name: "await");
1072 kw_declare = &IdentTable.get(Name: "declare");
1073 kw_finally = &IdentTable.get(Name: "finally");
1074 kw_from = &IdentTable.get(Name: "from");
1075 kw_function = &IdentTable.get(Name: "function");
1076 kw_get = &IdentTable.get(Name: "get");
1077 kw_import = &IdentTable.get(Name: "import");
1078 kw_infer = &IdentTable.get(Name: "infer");
1079 kw_is = &IdentTable.get(Name: "is");
1080 kw_let = &IdentTable.get(Name: "let");
1081 kw_module = &IdentTable.get(Name: "module");
1082 kw_readonly = &IdentTable.get(Name: "readonly");
1083 kw_set = &IdentTable.get(Name: "set");
1084 kw_type = &IdentTable.get(Name: "type");
1085 kw_typeof = &IdentTable.get(Name: "typeof");
1086 kw_var = &IdentTable.get(Name: "var");
1087 kw_yield = &IdentTable.get(Name: "yield");
1088
1089 kw_abstract = &IdentTable.get(Name: "abstract");
1090 kw_assert = &IdentTable.get(Name: "assert");
1091 kw_extends = &IdentTable.get(Name: "extends");
1092 kw_implements = &IdentTable.get(Name: "implements");
1093 kw_instanceof = &IdentTable.get(Name: "instanceof");
1094 kw_interface = &IdentTable.get(Name: "interface");
1095 kw_native = &IdentTable.get(Name: "native");
1096 kw_package = &IdentTable.get(Name: "package");
1097 kw_record = &IdentTable.get(Name: "record");
1098 kw_synchronized = &IdentTable.get(Name: "synchronized");
1099 kw_throws = &IdentTable.get(Name: "throws");
1100 kw___except = &IdentTable.get(Name: "__except");
1101 kw___has_include = &IdentTable.get(Name: "__has_include");
1102 kw___has_include_next = &IdentTable.get(Name: "__has_include_next");
1103
1104 kw_mark = &IdentTable.get(Name: "mark");
1105 kw_region = &IdentTable.get(Name: "region");
1106
1107 kw_extend = &IdentTable.get(Name: "extend");
1108 kw_option = &IdentTable.get(Name: "option");
1109 kw_optional = &IdentTable.get(Name: "optional");
1110 kw_repeated = &IdentTable.get(Name: "repeated");
1111 kw_required = &IdentTable.get(Name: "required");
1112 kw_returns = &IdentTable.get(Name: "returns");
1113
1114 kw_signals = &IdentTable.get(Name: "signals");
1115 kw_qsignals = &IdentTable.get(Name: "Q_SIGNALS");
1116 kw_slots = &IdentTable.get(Name: "slots");
1117 kw_qslots = &IdentTable.get(Name: "Q_SLOTS");
1118
1119 // For internal clang-format use.
1120 kw_internal_ident_after_define =
1121 &IdentTable.get(Name: "__CLANG_FORMAT_INTERNAL_IDENT_AFTER_DEFINE__");
1122
1123 // C# keywords
1124 kw_dollar = &IdentTable.get(Name: "dollar");
1125 kw_base = &IdentTable.get(Name: "base");
1126 kw_byte = &IdentTable.get(Name: "byte");
1127 kw_checked = &IdentTable.get(Name: "checked");
1128 kw_decimal = &IdentTable.get(Name: "decimal");
1129 kw_delegate = &IdentTable.get(Name: "delegate");
1130 kw_event = &IdentTable.get(Name: "event");
1131 kw_fixed = &IdentTable.get(Name: "fixed");
1132 kw_foreach = &IdentTable.get(Name: "foreach");
1133 kw_init = &IdentTable.get(Name: "init");
1134 kw_implicit = &IdentTable.get(Name: "implicit");
1135 kw_internal = &IdentTable.get(Name: "internal");
1136 kw_lock = &IdentTable.get(Name: "lock");
1137 kw_null = &IdentTable.get(Name: "null");
1138 kw_object = &IdentTable.get(Name: "object");
1139 kw_out = &IdentTable.get(Name: "out");
1140 kw_params = &IdentTable.get(Name: "params");
1141 kw_ref = &IdentTable.get(Name: "ref");
1142 kw_string = &IdentTable.get(Name: "string");
1143 kw_stackalloc = &IdentTable.get(Name: "stackalloc");
1144 kw_sbyte = &IdentTable.get(Name: "sbyte");
1145 kw_sealed = &IdentTable.get(Name: "sealed");
1146 kw_uint = &IdentTable.get(Name: "uint");
1147 kw_ulong = &IdentTable.get(Name: "ulong");
1148 kw_unchecked = &IdentTable.get(Name: "unchecked");
1149 kw_unsafe = &IdentTable.get(Name: "unsafe");
1150 kw_ushort = &IdentTable.get(Name: "ushort");
1151 kw_when = &IdentTable.get(Name: "when");
1152 kw_where = &IdentTable.get(Name: "where");
1153
1154 // Verilog keywords
1155 kw_always = &IdentTable.get(Name: "always");
1156 kw_always_comb = &IdentTable.get(Name: "always_comb");
1157 kw_always_ff = &IdentTable.get(Name: "always_ff");
1158 kw_always_latch = &IdentTable.get(Name: "always_latch");
1159 kw_assign = &IdentTable.get(Name: "assign");
1160 kw_assume = &IdentTable.get(Name: "assume");
1161 kw_automatic = &IdentTable.get(Name: "automatic");
1162 kw_before = &IdentTable.get(Name: "before");
1163 kw_begin = &IdentTable.get(Name: "begin");
1164 kw_begin_keywords = &IdentTable.get(Name: "begin_keywords");
1165 kw_bins = &IdentTable.get(Name: "bins");
1166 kw_binsof = &IdentTable.get(Name: "binsof");
1167 kw_casex = &IdentTable.get(Name: "casex");
1168 kw_casez = &IdentTable.get(Name: "casez");
1169 kw_celldefine = &IdentTable.get(Name: "celldefine");
1170 kw_checker = &IdentTable.get(Name: "checker");
1171 kw_clocking = &IdentTable.get(Name: "clocking");
1172 kw_constraint = &IdentTable.get(Name: "constraint");
1173 kw_context = &IdentTable.get(Name: "context");
1174 kw_cover = &IdentTable.get(Name: "cover");
1175 kw_covergroup = &IdentTable.get(Name: "covergroup");
1176 kw_coverpoint = &IdentTable.get(Name: "coverpoint");
1177 kw_default_decay_time = &IdentTable.get(Name: "default_decay_time");
1178 kw_default_nettype = &IdentTable.get(Name: "default_nettype");
1179 kw_default_trireg_strength = &IdentTable.get(Name: "default_trireg_strength");
1180 kw_delay_mode_distributed = &IdentTable.get(Name: "delay_mode_distributed");
1181 kw_delay_mode_path = &IdentTable.get(Name: "delay_mode_path");
1182 kw_delay_mode_unit = &IdentTable.get(Name: "delay_mode_unit");
1183 kw_delay_mode_zero = &IdentTable.get(Name: "delay_mode_zero");
1184 kw_disable = &IdentTable.get(Name: "disable");
1185 kw_dist = &IdentTable.get(Name: "dist");
1186 kw_edge = &IdentTable.get(Name: "edge");
1187 kw_elsif = &IdentTable.get(Name: "elsif");
1188 kw_end = &IdentTable.get(Name: "end");
1189 kw_end_keywords = &IdentTable.get(Name: "end_keywords");
1190 kw_endcase = &IdentTable.get(Name: "endcase");
1191 kw_endcelldefine = &IdentTable.get(Name: "endcelldefine");
1192 kw_endchecker = &IdentTable.get(Name: "endchecker");
1193 kw_endclass = &IdentTable.get(Name: "endclass");
1194 kw_endclocking = &IdentTable.get(Name: "endclocking");
1195 kw_endfunction = &IdentTable.get(Name: "endfunction");
1196 kw_endgenerate = &IdentTable.get(Name: "endgenerate");
1197 kw_endgroup = &IdentTable.get(Name: "endgroup");
1198 kw_endinterface = &IdentTable.get(Name: "endinterface");
1199 kw_endmodule = &IdentTable.get(Name: "endmodule");
1200 kw_endpackage = &IdentTable.get(Name: "endpackage");
1201 kw_endprimitive = &IdentTable.get(Name: "endprimitive");
1202 kw_endprogram = &IdentTable.get(Name: "endprogram");
1203 kw_endproperty = &IdentTable.get(Name: "endproperty");
1204 kw_endsequence = &IdentTable.get(Name: "endsequence");
1205 kw_endspecify = &IdentTable.get(Name: "endspecify");
1206 kw_endtable = &IdentTable.get(Name: "endtable");
1207 kw_endtask = &IdentTable.get(Name: "endtask");
1208 kw_forever = &IdentTable.get(Name: "forever");
1209 kw_fork = &IdentTable.get(Name: "fork");
1210 kw_generate = &IdentTable.get(Name: "generate");
1211 kw_highz0 = &IdentTable.get(Name: "highz0");
1212 kw_highz1 = &IdentTable.get(Name: "highz1");
1213 kw_iff = &IdentTable.get(Name: "iff");
1214 kw_ifnone = &IdentTable.get(Name: "ifnone");
1215 kw_ignore_bins = &IdentTable.get(Name: "ignore_bins");
1216 kw_illegal_bins = &IdentTable.get(Name: "illegal_bins");
1217 kw_initial = &IdentTable.get(Name: "initial");
1218 kw_inout = &IdentTable.get(Name: "inout");
1219 kw_input = &IdentTable.get(Name: "input");
1220 kw_inside = &IdentTable.get(Name: "inside");
1221 kw_interconnect = &IdentTable.get(Name: "interconnect");
1222 kw_intersect = &IdentTable.get(Name: "intersect");
1223 kw_join = &IdentTable.get(Name: "join");
1224 kw_join_any = &IdentTable.get(Name: "join_any");
1225 kw_join_none = &IdentTable.get(Name: "join_none");
1226 kw_large = &IdentTable.get(Name: "large");
1227 kw_local = &IdentTable.get(Name: "local");
1228 kw_localparam = &IdentTable.get(Name: "localparam");
1229 kw_macromodule = &IdentTable.get(Name: "macromodule");
1230 kw_matches = &IdentTable.get(Name: "matches");
1231 kw_medium = &IdentTable.get(Name: "medium");
1232 kw_negedge = &IdentTable.get(Name: "negedge");
1233 kw_nounconnected_drive = &IdentTable.get(Name: "nounconnected_drive");
1234 kw_output = &IdentTable.get(Name: "output");
1235 kw_packed = &IdentTable.get(Name: "packed");
1236 kw_parameter = &IdentTable.get(Name: "parameter");
1237 kw_posedge = &IdentTable.get(Name: "posedge");
1238 kw_primitive = &IdentTable.get(Name: "primitive");
1239 kw_priority = &IdentTable.get(Name: "priority");
1240 kw_program = &IdentTable.get(Name: "program");
1241 kw_property = &IdentTable.get(Name: "property");
1242 kw_pull0 = &IdentTable.get(Name: "pull0");
1243 kw_pull1 = &IdentTable.get(Name: "pull1");
1244 kw_pure = &IdentTable.get(Name: "pure");
1245 kw_rand = &IdentTable.get(Name: "rand");
1246 kw_randc = &IdentTable.get(Name: "randc");
1247 kw_randcase = &IdentTable.get(Name: "randcase");
1248 kw_randsequence = &IdentTable.get(Name: "randsequence");
1249 kw_repeat = &IdentTable.get(Name: "repeat");
1250 kw_resetall = &IdentTable.get(Name: "resetall");
1251 kw_sample = &IdentTable.get(Name: "sample");
1252 kw_scalared = &IdentTable.get(Name: "scalared");
1253 kw_sequence = &IdentTable.get(Name: "sequence");
1254 kw_small = &IdentTable.get(Name: "small");
1255 kw_soft = &IdentTable.get(Name: "soft");
1256 kw_solve = &IdentTable.get(Name: "solve");
1257 kw_specify = &IdentTable.get(Name: "specify");
1258 kw_specparam = &IdentTable.get(Name: "specparam");
1259 kw_strong0 = &IdentTable.get(Name: "strong0");
1260 kw_strong1 = &IdentTable.get(Name: "strong1");
1261 kw_supply0 = &IdentTable.get(Name: "supply0");
1262 kw_supply1 = &IdentTable.get(Name: "supply1");
1263 kw_table = &IdentTable.get(Name: "table");
1264 kw_tagged = &IdentTable.get(Name: "tagged");
1265 kw_task = &IdentTable.get(Name: "task");
1266 kw_timescale = &IdentTable.get(Name: "timescale");
1267 kw_tri = &IdentTable.get(Name: "tri");
1268 kw_tri0 = &IdentTable.get(Name: "tri0");
1269 kw_tri1 = &IdentTable.get(Name: "tri1");
1270 kw_triand = &IdentTable.get(Name: "triand");
1271 kw_trior = &IdentTable.get(Name: "trior");
1272 kw_trireg = &IdentTable.get(Name: "trireg");
1273 kw_unconnected_drive = &IdentTable.get(Name: "unconnected_drive");
1274 kw_undefineall = &IdentTable.get(Name: "undefineall");
1275 kw_unique = &IdentTable.get(Name: "unique");
1276 kw_unique0 = &IdentTable.get(Name: "unique0");
1277 kw_uwire = &IdentTable.get(Name: "uwire");
1278 kw_vectored = &IdentTable.get(Name: "vectored");
1279 kw_wait = &IdentTable.get(Name: "wait");
1280 kw_wand = &IdentTable.get(Name: "wand");
1281 kw_weak0 = &IdentTable.get(Name: "weak0");
1282 kw_weak1 = &IdentTable.get(Name: "weak1");
1283 kw_wildcard = &IdentTable.get(Name: "wildcard");
1284 kw_wire = &IdentTable.get(Name: "wire");
1285 kw_with = &IdentTable.get(Name: "with");
1286 kw_wor = &IdentTable.get(Name: "wor");
1287
1288 // Symbols that are treated as keywords.
1289 kw_verilogHash = &IdentTable.get(Name: "#");
1290 kw_verilogHashHash = &IdentTable.get(Name: "##");
1291 kw_apostrophe = &IdentTable.get(Name: "\'");
1292
1293 // TableGen keywords.
1294 kw_bit = &IdentTable.get(Name: "bit");
1295 kw_bits = &IdentTable.get(Name: "bits");
1296 kw_code = &IdentTable.get(Name: "code");
1297 kw_dag = &IdentTable.get(Name: "dag");
1298 kw_def = &IdentTable.get(Name: "def");
1299 kw_defm = &IdentTable.get(Name: "defm");
1300 kw_defset = &IdentTable.get(Name: "defset");
1301 kw_defvar = &IdentTable.get(Name: "defvar");
1302 kw_dump = &IdentTable.get(Name: "dump");
1303 kw_include = &IdentTable.get(Name: "include");
1304 kw_list = &IdentTable.get(Name: "list");
1305 kw_multiclass = &IdentTable.get(Name: "multiclass");
1306 kw_then = &IdentTable.get(Name: "then");
1307
1308 // Keep this at the end of the constructor to make sure everything here is
1309 // already initialized.
1310 JsExtraKeywords = std::unordered_set<IdentifierInfo *>(
1311 {kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from,
1312 kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_override,
1313 kw_readonly, kw_set, kw_type, kw_typeof, kw_var, kw_yield,
1314 // Keywords from the Java section.
1315 kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface});
1316
1317 CSharpExtraKeywords = JsExtraKeywords;
1318 CSharpExtraKeywords.insert(
1319 l: {kw_base, kw_byte, kw_checked, kw_decimal, kw_delegate,
1320 kw_event, kw_fixed, kw_foreach, kw_implicit, kw_in,
1321 kw_init, kw_internal, kw_lock, kw_null, kw_object,
1322 kw_out, kw_params, kw_ref, kw_string, kw_stackalloc,
1323 kw_sbyte, kw_sealed, kw_uint, kw_ulong, kw_unchecked,
1324 kw_unsafe, kw_ushort, kw_when, kw_where});
1325
1326 // Some keywords are not included here because they don't need special
1327 // treatment like `showcancelled` or they should be treated as identifiers
1328 // like `int` and `logic`.
1329 VerilogExtraKeywords =
1330 std::unordered_set<IdentifierInfo *>({kw_always,
1331 kw_always_comb,
1332 kw_always_ff,
1333 kw_always_latch,
1334 kw_assert,
1335 kw_assign,
1336 kw_assume,
1337 kw_automatic,
1338 kw_before,
1339 kw_begin,
1340 kw_bins,
1341 kw_binsof,
1342 kw_casex,
1343 kw_casez,
1344 kw_celldefine,
1345 kw_checker,
1346 kw_clocking,
1347 kw_constraint,
1348 kw_context,
1349 kw_cover,
1350 kw_covergroup,
1351 kw_coverpoint,
1352 kw_disable,
1353 kw_dist,
1354 kw_edge,
1355 kw_end,
1356 kw_endcase,
1357 kw_endchecker,
1358 kw_endclass,
1359 kw_endclocking,
1360 kw_endfunction,
1361 kw_endgenerate,
1362 kw_endgroup,
1363 kw_endinterface,
1364 kw_endmodule,
1365 kw_endpackage,
1366 kw_endprimitive,
1367 kw_endprogram,
1368 kw_endproperty,
1369 kw_endsequence,
1370 kw_endspecify,
1371 kw_endtable,
1372 kw_endtask,
1373 kw_extends,
1374 kw_final,
1375 kw_foreach,
1376 kw_forever,
1377 kw_fork,
1378 kw_function,
1379 kw_generate,
1380 kw_highz0,
1381 kw_highz1,
1382 kw_iff,
1383 kw_ifnone,
1384 kw_ignore_bins,
1385 kw_illegal_bins,
1386 kw_implements,
1387 kw_import,
1388 kw_initial,
1389 kw_inout,
1390 kw_input,
1391 kw_inside,
1392 kw_interconnect,
1393 kw_interface,
1394 kw_intersect,
1395 kw_join,
1396 kw_join_any,
1397 kw_join_none,
1398 kw_large,
1399 kw_let,
1400 kw_local,
1401 kw_localparam,
1402 kw_macromodule,
1403 kw_matches,
1404 kw_medium,
1405 kw_module,
1406 kw_negedge,
1407 kw_output,
1408 kw_package,
1409 kw_packed,
1410 kw_parameter,
1411 kw_posedge,
1412 kw_primitive,
1413 kw_priority,
1414 kw_program,
1415 kw_property,
1416 kw_pull0,
1417 kw_pull1,
1418 kw_pure,
1419 kw_rand,
1420 kw_randc,
1421 kw_randcase,
1422 kw_randsequence,
1423 kw_ref,
1424 kw_repeat,
1425 kw_sample,
1426 kw_scalared,
1427 kw_sequence,
1428 kw_small,
1429 kw_soft,
1430 kw_solve,
1431 kw_specify,
1432 kw_specparam,
1433 kw_strong0,
1434 kw_strong1,
1435 kw_supply0,
1436 kw_supply1,
1437 kw_table,
1438 kw_tagged,
1439 kw_task,
1440 kw_tri,
1441 kw_tri0,
1442 kw_tri1,
1443 kw_triand,
1444 kw_trior,
1445 kw_trireg,
1446 kw_unique,
1447 kw_unique0,
1448 kw_uwire,
1449 kw_var,
1450 kw_vectored,
1451 kw_wait,
1452 kw_wand,
1453 kw_weak0,
1454 kw_weak1,
1455 kw_wildcard,
1456 kw_wire,
1457 kw_with,
1458 kw_wor,
1459 kw_verilogHash,
1460 kw_verilogHashHash});
1461
1462 TableGenExtraKeywords = std::unordered_set<IdentifierInfo *>({
1463 kw_assert,
1464 kw_bit,
1465 kw_bits,
1466 kw_code,
1467 kw_dag,
1468 kw_def,
1469 kw_defm,
1470 kw_defset,
1471 kw_defvar,
1472 kw_dump,
1473 kw_foreach,
1474 kw_in,
1475 kw_include,
1476 kw_let,
1477 kw_list,
1478 kw_multiclass,
1479 kw_string,
1480 kw_then,
1481 });
1482 }
1483
1484 // Context sensitive keywords.
1485 IdentifierInfo *kw_final;
1486 IdentifierInfo *kw_override;
1487 IdentifierInfo *kw_in;
1488 IdentifierInfo *kw_of;
1489 IdentifierInfo *kw_CF_CLOSED_ENUM;
1490 IdentifierInfo *kw_CF_ENUM;
1491 IdentifierInfo *kw_CF_OPTIONS;
1492 IdentifierInfo *kw_NS_CLOSED_ENUM;
1493 IdentifierInfo *kw_NS_ENUM;
1494 IdentifierInfo *kw_NS_ERROR_ENUM;
1495 IdentifierInfo *kw_NS_OPTIONS;
1496 IdentifierInfo *kw___except;
1497 IdentifierInfo *kw___has_include;
1498 IdentifierInfo *kw___has_include_next;
1499
1500 // JavaScript keywords.
1501 IdentifierInfo *kw_as;
1502 IdentifierInfo *kw_async;
1503 IdentifierInfo *kw_await;
1504 IdentifierInfo *kw_declare;
1505 IdentifierInfo *kw_finally;
1506 IdentifierInfo *kw_from;
1507 IdentifierInfo *kw_function;
1508 IdentifierInfo *kw_get;
1509 IdentifierInfo *kw_import;
1510 IdentifierInfo *kw_infer;
1511 IdentifierInfo *kw_is;
1512 IdentifierInfo *kw_let;
1513 IdentifierInfo *kw_module;
1514 IdentifierInfo *kw_readonly;
1515 IdentifierInfo *kw_set;
1516 IdentifierInfo *kw_type;
1517 IdentifierInfo *kw_typeof;
1518 IdentifierInfo *kw_var;
1519 IdentifierInfo *kw_yield;
1520
1521 // Java keywords.
1522 IdentifierInfo *kw_abstract;
1523 IdentifierInfo *kw_assert;
1524 IdentifierInfo *kw_extends;
1525 IdentifierInfo *kw_implements;
1526 IdentifierInfo *kw_instanceof;
1527 IdentifierInfo *kw_interface;
1528 IdentifierInfo *kw_native;
1529 IdentifierInfo *kw_package;
1530 IdentifierInfo *kw_record;
1531 IdentifierInfo *kw_synchronized;
1532 IdentifierInfo *kw_throws;
1533
1534 // Pragma keywords.
1535 IdentifierInfo *kw_mark;
1536 IdentifierInfo *kw_region;
1537
1538 // Proto keywords.
1539 IdentifierInfo *kw_extend;
1540 IdentifierInfo *kw_option;
1541 IdentifierInfo *kw_optional;
1542 IdentifierInfo *kw_repeated;
1543 IdentifierInfo *kw_required;
1544 IdentifierInfo *kw_returns;
1545
1546 // QT keywords.
1547 IdentifierInfo *kw_signals;
1548 IdentifierInfo *kw_qsignals;
1549 IdentifierInfo *kw_slots;
1550 IdentifierInfo *kw_qslots;
1551
1552 // For internal use by clang-format.
1553 IdentifierInfo *kw_internal_ident_after_define;
1554
1555 // C# keywords
1556 IdentifierInfo *kw_dollar;
1557 IdentifierInfo *kw_base;
1558 IdentifierInfo *kw_byte;
1559 IdentifierInfo *kw_checked;
1560 IdentifierInfo *kw_decimal;
1561 IdentifierInfo *kw_delegate;
1562 IdentifierInfo *kw_event;
1563 IdentifierInfo *kw_fixed;
1564 IdentifierInfo *kw_foreach;
1565 IdentifierInfo *kw_implicit;
1566 IdentifierInfo *kw_init;
1567 IdentifierInfo *kw_internal;
1568
1569 IdentifierInfo *kw_lock;
1570 IdentifierInfo *kw_null;
1571 IdentifierInfo *kw_object;
1572 IdentifierInfo *kw_out;
1573
1574 IdentifierInfo *kw_params;
1575
1576 IdentifierInfo *kw_ref;
1577 IdentifierInfo *kw_string;
1578 IdentifierInfo *kw_stackalloc;
1579 IdentifierInfo *kw_sbyte;
1580 IdentifierInfo *kw_sealed;
1581 IdentifierInfo *kw_uint;
1582 IdentifierInfo *kw_ulong;
1583 IdentifierInfo *kw_unchecked;
1584 IdentifierInfo *kw_unsafe;
1585 IdentifierInfo *kw_ushort;
1586 IdentifierInfo *kw_when;
1587 IdentifierInfo *kw_where;
1588
1589 // Verilog keywords
1590 IdentifierInfo *kw_always;
1591 IdentifierInfo *kw_always_comb;
1592 IdentifierInfo *kw_always_ff;
1593 IdentifierInfo *kw_always_latch;
1594 IdentifierInfo *kw_assign;
1595 IdentifierInfo *kw_assume;
1596 IdentifierInfo *kw_automatic;
1597 IdentifierInfo *kw_before;
1598 IdentifierInfo *kw_begin;
1599 IdentifierInfo *kw_begin_keywords;
1600 IdentifierInfo *kw_bins;
1601 IdentifierInfo *kw_binsof;
1602 IdentifierInfo *kw_casex;
1603 IdentifierInfo *kw_casez;
1604 IdentifierInfo *kw_celldefine;
1605 IdentifierInfo *kw_checker;
1606 IdentifierInfo *kw_clocking;
1607 IdentifierInfo *kw_constraint;
1608 IdentifierInfo *kw_context;
1609 IdentifierInfo *kw_cover;
1610 IdentifierInfo *kw_covergroup;
1611 IdentifierInfo *kw_coverpoint;
1612 IdentifierInfo *kw_default_decay_time;
1613 IdentifierInfo *kw_default_nettype;
1614 IdentifierInfo *kw_default_trireg_strength;
1615 IdentifierInfo *kw_delay_mode_distributed;
1616 IdentifierInfo *kw_delay_mode_path;
1617 IdentifierInfo *kw_delay_mode_unit;
1618 IdentifierInfo *kw_delay_mode_zero;
1619 IdentifierInfo *kw_disable;
1620 IdentifierInfo *kw_dist;
1621 IdentifierInfo *kw_elsif;
1622 IdentifierInfo *kw_edge;
1623 IdentifierInfo *kw_end;
1624 IdentifierInfo *kw_end_keywords;
1625 IdentifierInfo *kw_endcase;
1626 IdentifierInfo *kw_endcelldefine;
1627 IdentifierInfo *kw_endchecker;
1628 IdentifierInfo *kw_endclass;
1629 IdentifierInfo *kw_endclocking;
1630 IdentifierInfo *kw_endfunction;
1631 IdentifierInfo *kw_endgenerate;
1632 IdentifierInfo *kw_endgroup;
1633 IdentifierInfo *kw_endinterface;
1634 IdentifierInfo *kw_endmodule;
1635 IdentifierInfo *kw_endpackage;
1636 IdentifierInfo *kw_endprimitive;
1637 IdentifierInfo *kw_endprogram;
1638 IdentifierInfo *kw_endproperty;
1639 IdentifierInfo *kw_endsequence;
1640 IdentifierInfo *kw_endspecify;
1641 IdentifierInfo *kw_endtable;
1642 IdentifierInfo *kw_endtask;
1643 IdentifierInfo *kw_forever;
1644 IdentifierInfo *kw_fork;
1645 IdentifierInfo *kw_generate;
1646 IdentifierInfo *kw_highz0;
1647 IdentifierInfo *kw_highz1;
1648 IdentifierInfo *kw_iff;
1649 IdentifierInfo *kw_ifnone;
1650 IdentifierInfo *kw_ignore_bins;
1651 IdentifierInfo *kw_illegal_bins;
1652 IdentifierInfo *kw_initial;
1653 IdentifierInfo *kw_inout;
1654 IdentifierInfo *kw_input;
1655 IdentifierInfo *kw_inside;
1656 IdentifierInfo *kw_interconnect;
1657 IdentifierInfo *kw_intersect;
1658 IdentifierInfo *kw_join;
1659 IdentifierInfo *kw_join_any;
1660 IdentifierInfo *kw_join_none;
1661 IdentifierInfo *kw_large;
1662 IdentifierInfo *kw_local;
1663 IdentifierInfo *kw_localparam;
1664 IdentifierInfo *kw_macromodule;
1665 IdentifierInfo *kw_matches;
1666 IdentifierInfo *kw_medium;
1667 IdentifierInfo *kw_negedge;
1668 IdentifierInfo *kw_nounconnected_drive;
1669 IdentifierInfo *kw_output;
1670 IdentifierInfo *kw_packed;
1671 IdentifierInfo *kw_parameter;
1672 IdentifierInfo *kw_posedge;
1673 IdentifierInfo *kw_primitive;
1674 IdentifierInfo *kw_priority;
1675 IdentifierInfo *kw_program;
1676 IdentifierInfo *kw_property;
1677 IdentifierInfo *kw_pull0;
1678 IdentifierInfo *kw_pull1;
1679 IdentifierInfo *kw_pure;
1680 IdentifierInfo *kw_rand;
1681 IdentifierInfo *kw_randc;
1682 IdentifierInfo *kw_randcase;
1683 IdentifierInfo *kw_randsequence;
1684 IdentifierInfo *kw_repeat;
1685 IdentifierInfo *kw_resetall;
1686 IdentifierInfo *kw_sample;
1687 IdentifierInfo *kw_scalared;
1688 IdentifierInfo *kw_sequence;
1689 IdentifierInfo *kw_small;
1690 IdentifierInfo *kw_soft;
1691 IdentifierInfo *kw_solve;
1692 IdentifierInfo *kw_specify;
1693 IdentifierInfo *kw_specparam;
1694 IdentifierInfo *kw_strong0;
1695 IdentifierInfo *kw_strong1;
1696 IdentifierInfo *kw_supply0;
1697 IdentifierInfo *kw_supply1;
1698 IdentifierInfo *kw_table;
1699 IdentifierInfo *kw_tagged;
1700 IdentifierInfo *kw_task;
1701 IdentifierInfo *kw_timescale;
1702 IdentifierInfo *kw_tri0;
1703 IdentifierInfo *kw_tri1;
1704 IdentifierInfo *kw_tri;
1705 IdentifierInfo *kw_triand;
1706 IdentifierInfo *kw_trior;
1707 IdentifierInfo *kw_trireg;
1708 IdentifierInfo *kw_unconnected_drive;
1709 IdentifierInfo *kw_undefineall;
1710 IdentifierInfo *kw_unique;
1711 IdentifierInfo *kw_unique0;
1712 IdentifierInfo *kw_uwire;
1713 IdentifierInfo *kw_vectored;
1714 IdentifierInfo *kw_wait;
1715 IdentifierInfo *kw_wand;
1716 IdentifierInfo *kw_weak0;
1717 IdentifierInfo *kw_weak1;
1718 IdentifierInfo *kw_wildcard;
1719 IdentifierInfo *kw_wire;
1720 IdentifierInfo *kw_with;
1721 IdentifierInfo *kw_wor;
1722
1723 // Workaround for hashes and backticks in Verilog.
1724 IdentifierInfo *kw_verilogHash;
1725 IdentifierInfo *kw_verilogHashHash;
1726
1727 // Symbols in Verilog that don't exist in C++.
1728 IdentifierInfo *kw_apostrophe;
1729
1730 // TableGen keywords
1731 IdentifierInfo *kw_bit;
1732 IdentifierInfo *kw_bits;
1733 IdentifierInfo *kw_code;
1734 IdentifierInfo *kw_dag;
1735 IdentifierInfo *kw_def;
1736 IdentifierInfo *kw_defm;
1737 IdentifierInfo *kw_defset;
1738 IdentifierInfo *kw_defvar;
1739 IdentifierInfo *kw_dump;
1740 IdentifierInfo *kw_include;
1741 IdentifierInfo *kw_list;
1742 IdentifierInfo *kw_multiclass;
1743 IdentifierInfo *kw_then;
1744
1745 /// Returns \c true if \p Tok is a keyword or an identifier.
1746 bool isWordLike(const FormatToken &Tok, bool IsVerilog = true) const {
1747 // getIdentifierinfo returns non-null for keywords as well as identifiers.
1748 return Tok.Tok.getIdentifierInfo() &&
1749 (!IsVerilog || !isVerilogKeywordSymbol(Tok));
1750 }
1751
1752 /// Returns \c true if \p Tok is a true JavaScript identifier, returns
1753 /// \c false if it is a keyword or a pseudo keyword.
1754 /// If \c AcceptIdentifierName is true, returns true not only for keywords,
1755 // but also for IdentifierName tokens (aka pseudo-keywords), such as
1756 // ``yield``.
1757 bool isJavaScriptIdentifier(const FormatToken &Tok,
1758 bool AcceptIdentifierName = true) const {
1759 // Based on the list of JavaScript & TypeScript keywords here:
1760 // https://github.com/microsoft/TypeScript/blob/main/src/compiler/scanner.ts#L74
1761 if (Tok.isAccessSpecifierKeyword())
1762 return false;
1763 switch (Tok.Tok.getKind()) {
1764 case tok::kw_break:
1765 case tok::kw_case:
1766 case tok::kw_catch:
1767 case tok::kw_class:
1768 case tok::kw_continue:
1769 case tok::kw_const:
1770 case tok::kw_default:
1771 case tok::kw_delete:
1772 case tok::kw_do:
1773 case tok::kw_else:
1774 case tok::kw_enum:
1775 case tok::kw_export:
1776 case tok::kw_false:
1777 case tok::kw_for:
1778 case tok::kw_if:
1779 case tok::kw_import:
1780 case tok::kw_module:
1781 case tok::kw_new:
1782 case tok::kw_return:
1783 case tok::kw_static:
1784 case tok::kw_switch:
1785 case tok::kw_this:
1786 case tok::kw_throw:
1787 case tok::kw_true:
1788 case tok::kw_try:
1789 case tok::kw_typeof:
1790 case tok::kw_void:
1791 case tok::kw_while:
1792 // These are JS keywords that are lexed by LLVM/clang as keywords.
1793 return false;
1794 case tok::identifier: {
1795 // For identifiers, make sure they are true identifiers, excluding the
1796 // JavaScript pseudo-keywords (not lexed by LLVM/clang as keywords).
1797 bool IsPseudoKeyword =
1798 JsExtraKeywords.find(x: Tok.Tok.getIdentifierInfo()) !=
1799 JsExtraKeywords.end();
1800 return AcceptIdentifierName || !IsPseudoKeyword;
1801 }
1802 default:
1803 // Other keywords are handled in the switch below, to avoid problems due
1804 // to duplicate case labels when using the #include trick.
1805 break;
1806 }
1807
1808 switch (Tok.Tok.getKind()) {
1809 // Handle C++ keywords not included above: these are all JS identifiers.
1810#define KEYWORD(X, Y) case tok::kw_##X:
1811#include "clang/Basic/TokenKinds.def"
1812 // #undef KEYWORD is not needed -- it's #undef-ed at the end of
1813 // TokenKinds.def
1814 return true;
1815 default:
1816 // All other tokens (punctuation etc) are not JS identifiers.
1817 return false;
1818 }
1819 }
1820
1821 /// Returns \c true if \p Tok is a C# keyword, returns \c false if it is
1822 /// anything else.
1823 bool isCSharpKeyword(const FormatToken &Tok) const {
1824 if (Tok.isAccessSpecifierKeyword())
1825 return true;
1826 switch (Tok.Tok.getKind()) {
1827 case tok::kw_bool:
1828 case tok::kw_break:
1829 case tok::kw_case:
1830 case tok::kw_catch:
1831 case tok::kw_char:
1832 case tok::kw_class:
1833 case tok::kw_const:
1834 case tok::kw_continue:
1835 case tok::kw_default:
1836 case tok::kw_do:
1837 case tok::kw_double:
1838 case tok::kw_else:
1839 case tok::kw_enum:
1840 case tok::kw_explicit:
1841 case tok::kw_extern:
1842 case tok::kw_false:
1843 case tok::kw_float:
1844 case tok::kw_for:
1845 case tok::kw_goto:
1846 case tok::kw_if:
1847 case tok::kw_int:
1848 case tok::kw_long:
1849 case tok::kw_namespace:
1850 case tok::kw_new:
1851 case tok::kw_operator:
1852 case tok::kw_return:
1853 case tok::kw_short:
1854 case tok::kw_sizeof:
1855 case tok::kw_static:
1856 case tok::kw_struct:
1857 case tok::kw_switch:
1858 case tok::kw_this:
1859 case tok::kw_throw:
1860 case tok::kw_true:
1861 case tok::kw_try:
1862 case tok::kw_typeof:
1863 case tok::kw_using:
1864 case tok::kw_virtual:
1865 case tok::kw_void:
1866 case tok::kw_volatile:
1867 case tok::kw_while:
1868 return true;
1869 default:
1870 return Tok.is(Kind: tok::identifier) &&
1871 CSharpExtraKeywords.find(x: Tok.Tok.getIdentifierInfo()) ==
1872 CSharpExtraKeywords.end();
1873 }
1874 }
1875
1876 bool isVerilogKeywordSymbol(const FormatToken &Tok) const {
1877 return Tok.isOneOf(K1: kw_verilogHash, K2: kw_verilogHashHash, Ks: kw_apostrophe);
1878 }
1879
1880 bool isVerilogWordOperator(const FormatToken &Tok) const {
1881 return Tok.isOneOf(K1: kw_before, K2: kw_intersect, Ks: kw_dist, Ks: kw_iff, Ks: kw_inside,
1882 Ks: kw_with);
1883 }
1884
1885 bool isVerilogIdentifier(const FormatToken &Tok) const {
1886 switch (Tok.Tok.getKind()) {
1887 case tok::kw_case:
1888 case tok::kw_class:
1889 case tok::kw_const:
1890 case tok::kw_continue:
1891 case tok::kw_default:
1892 case tok::kw_do:
1893 case tok::kw_else:
1894 case tok::kw_enum:
1895 case tok::kw_export:
1896 case tok::kw_extern:
1897 case tok::kw_for:
1898 case tok::kw_if:
1899 case tok::kw_import:
1900 case tok::kw_restrict:
1901 case tok::kw_signed:
1902 case tok::kw_static:
1903 case tok::kw_struct:
1904 case tok::kw_typedef:
1905 case tok::kw_union:
1906 case tok::kw_unsigned:
1907 case tok::kw_virtual:
1908 case tok::kw_while:
1909 return false;
1910 case tok::identifier:
1911 return isWordLike(Tok) &&
1912 VerilogExtraKeywords.find(x: Tok.Tok.getIdentifierInfo()) ==
1913 VerilogExtraKeywords.end();
1914 default:
1915 // getIdentifierInfo returns non-null for both identifiers and keywords.
1916 return Tok.Tok.getIdentifierInfo();
1917 }
1918 }
1919
1920 /// Returns whether \p Tok is a Verilog preprocessor directive. This is
1921 /// needed because macro expansions start with a backtick as well and they
1922 /// need to be treated differently.
1923 bool isVerilogPPDirective(const FormatToken &Tok) const {
1924 auto Info = Tok.Tok.getIdentifierInfo();
1925 if (!Info)
1926 return false;
1927 switch (Info->getPPKeywordID()) {
1928 case tok::pp_define:
1929 case tok::pp_else:
1930 case tok::pp_endif:
1931 case tok::pp_ifdef:
1932 case tok::pp_ifndef:
1933 case tok::pp_include:
1934 case tok::pp_line:
1935 case tok::pp_pragma:
1936 case tok::pp_undef:
1937 return true;
1938 default:
1939 return Tok.isOneOf(K1: kw_begin_keywords, K2: kw_celldefine,
1940 Ks: kw_default_decay_time, Ks: kw_default_nettype,
1941 Ks: kw_default_trireg_strength, Ks: kw_delay_mode_distributed,
1942 Ks: kw_delay_mode_path, Ks: kw_delay_mode_unit,
1943 Ks: kw_delay_mode_zero, Ks: kw_elsif, Ks: kw_end_keywords,
1944 Ks: kw_endcelldefine, Ks: kw_nounconnected_drive, Ks: kw_resetall,
1945 Ks: kw_timescale, Ks: kw_unconnected_drive, Ks: kw_undefineall);
1946 }
1947 }
1948
1949 /// Returns whether \p Tok is a Verilog keyword that opens a block.
1950 bool isVerilogBegin(const FormatToken &Tok) const {
1951 // `table` is not included since it needs to be treated specially.
1952 if (Tok.isOneOf(K1: kw_begin, K2: kw_generate, Ks: kw_specify))
1953 return true;
1954 if (Tok.isNot(Kind: kw_fork))
1955 return false;
1956 const auto *Prev = Tok.getPreviousNonComment();
1957 return !(Prev && Prev->isOneOf(K1: kw_disable, K2: kw_wait));
1958 }
1959
1960 /// Returns whether \p Tok is a Verilog keyword that closes a block.
1961 bool isVerilogEnd(const FormatToken &Tok) const {
1962 return !Tok.endsSequence(K1: kw_join, Tokens: kw_rand) &&
1963 Tok.isOneOf(K1: TT_MacroBlockEnd, K2: kw_end, Ks: kw_endcase, Ks: kw_endclass,
1964 Ks: kw_endclocking, Ks: kw_endchecker, Ks: kw_endfunction,
1965 Ks: kw_endgenerate, Ks: kw_endgroup, Ks: kw_endinterface,
1966 Ks: kw_endmodule, Ks: kw_endpackage, Ks: kw_endprimitive,
1967 Ks: kw_endprogram, Ks: kw_endproperty, Ks: kw_endsequence,
1968 Ks: kw_endspecify, Ks: kw_endtable, Ks: kw_endtask, Ks: kw_join,
1969 Ks: kw_join_any, Ks: kw_join_none);
1970 }
1971
1972 /// Returns whether \p Tok is a Verilog keyword that opens a module, etc.
1973 bool isVerilogHierarchy(const FormatToken &Tok) const {
1974 if (Tok.endsSequence(K1: kw_function, Tokens: kw_with))
1975 return false;
1976 if (Tok.is(II: kw_property)) {
1977 const FormatToken *Prev = Tok.getPreviousNonComment();
1978 return !(Prev &&
1979 Prev->isOneOf(K1: tok::kw_restrict, K2: kw_assert, Ks: kw_assume, Ks: kw_cover));
1980 }
1981 return Tok.isOneOf(K1: tok::kw_case, K2: tok::kw_class, Ks: kw_function, Ks: kw_module,
1982 Ks: kw_interface, Ks: kw_package, Ks: kw_casex, Ks: kw_casez, Ks: kw_checker,
1983 Ks: kw_clocking, Ks: kw_covergroup, Ks: kw_macromodule, Ks: kw_primitive,
1984 Ks: kw_program, Ks: kw_property, Ks: kw_randcase, Ks: kw_randsequence,
1985 Ks: kw_task);
1986 }
1987
1988 bool isVerilogEndOfLabel(const FormatToken &Tok) const {
1989 const FormatToken *Next = Tok.getNextNonComment();
1990 // In Verilog the colon in a default label is optional.
1991 return Tok.is(TT: TT_CaseLabelColon) ||
1992 (Tok.is(Kind: tok::kw_default) &&
1993 !(Next && Next->isOneOf(K1: tok::colon, K2: tok::semi, Ks: kw_clocking, Ks: kw_iff,
1994 Ks: kw_input, Ks: kw_output, Ks: kw_sequence)));
1995 }
1996
1997 /// Returns whether \p Tok is a Verilog keyword that starts a
1998 /// structured procedure like 'always'.
1999 bool isVerilogStructuredProcedure(const FormatToken &Tok) const {
2000 return Tok.isOneOf(K1: kw_always, K2: kw_always_comb, Ks: kw_always_ff, Ks: kw_always_latch,
2001 Ks: kw_final, Ks: kw_forever, Ks: kw_initial);
2002 }
2003
2004 bool isVerilogQualifier(const FormatToken &Tok) const {
2005 switch (Tok.Tok.getKind()) {
2006 case tok::kw_extern:
2007 case tok::kw_signed:
2008 case tok::kw_static:
2009 case tok::kw_unsigned:
2010 case tok::kw_virtual:
2011 return true;
2012 case tok::identifier:
2013 return Tok.isOneOf(
2014 K1: kw_let, K2: kw_var, Ks: kw_ref, Ks: kw_automatic, Ks: kw_bins, Ks: kw_coverpoint,
2015 Ks: kw_ignore_bins, Ks: kw_illegal_bins, Ks: kw_inout, Ks: kw_input, Ks: kw_interconnect,
2016 Ks: kw_local, Ks: kw_localparam, Ks: kw_output, Ks: kw_parameter, Ks: kw_pure, Ks: kw_rand,
2017 Ks: kw_randc, Ks: kw_scalared, Ks: kw_specparam, Ks: kw_tri, Ks: kw_tri0, Ks: kw_tri1,
2018 Ks: kw_triand, Ks: kw_trior, Ks: kw_trireg, Ks: kw_uwire, Ks: kw_vectored, Ks: kw_wand,
2019 Ks: kw_wildcard, Ks: kw_wire, Ks: kw_wor);
2020 default:
2021 return false;
2022 }
2023 }
2024
2025 bool isTableGenDefinition(const FormatToken &Tok) const {
2026 return Tok.isOneOf(K1: kw_def, K2: kw_defm, Ks: kw_defset, Ks: kw_defvar, Ks: kw_multiclass,
2027 Ks: kw_let, Ks: tok::kw_class);
2028 }
2029
2030 bool isTableGenKeyword(const FormatToken &Tok) const {
2031 switch (Tok.Tok.getKind()) {
2032 case tok::kw_class:
2033 case tok::kw_else:
2034 case tok::kw_false:
2035 case tok::kw_if:
2036 case tok::kw_int:
2037 case tok::kw_true:
2038 return true;
2039 default:
2040 return Tok.is(Kind: tok::identifier) &&
2041 TableGenExtraKeywords.find(x: Tok.Tok.getIdentifierInfo()) !=
2042 TableGenExtraKeywords.end();
2043 }
2044 }
2045
2046private:
2047 /// The JavaScript keywords beyond the C++ keyword set.
2048 std::unordered_set<IdentifierInfo *> JsExtraKeywords;
2049
2050 /// The C# keywords beyond the C++ keyword set.
2051 std::unordered_set<IdentifierInfo *> CSharpExtraKeywords;
2052
2053 /// The Verilog keywords beyond the C++ keyword set.
2054 std::unordered_set<IdentifierInfo *> VerilogExtraKeywords;
2055
2056 /// The TableGen keywords beyond the C++ keyword set.
2057 std::unordered_set<IdentifierInfo *> TableGenExtraKeywords;
2058};
2059
2060inline bool isLineComment(const FormatToken &FormatTok) {
2061 return FormatTok.is(Kind: tok::comment) && !FormatTok.TokenText.starts_with(Prefix: "/*");
2062}
2063
2064// Checks if \p FormatTok is a line comment that continues the line comment
2065// \p Previous. The original column of \p MinColumnToken is used to determine
2066// whether \p FormatTok is indented enough to the right to continue \p Previous.
2067inline bool continuesLineComment(const FormatToken &FormatTok,
2068 const FormatToken *Previous,
2069 const FormatToken *MinColumnToken) {
2070 if (!Previous || !MinColumnToken)
2071 return false;
2072 unsigned MinContinueColumn =
2073 MinColumnToken->OriginalColumn + (isLineComment(FormatTok: *MinColumnToken) ? 0 : 1);
2074 return isLineComment(FormatTok) && FormatTok.NewlinesBefore == 1 &&
2075 isLineComment(FormatTok: *Previous) &&
2076 FormatTok.OriginalColumn >= MinContinueColumn;
2077}
2078
2079// Returns \c true if \c Current starts a new parameter.
2080bool startsNextParameter(const FormatToken &Current, const FormatStyle &Style);
2081
2082} // namespace format
2083} // namespace clang
2084
2085#endif
2086