1 //===--- FormatToken.h - Format C++ code ------------------------*- C++ -*-===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 /// 10 /// \file 11 /// This file contains the declaration of the FormatToken, a wrapper 12 /// around Token with additional information related to formatting. 13 /// 14 //===----------------------------------------------------------------------===// 15 16 #ifndef LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H 17 #define LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H 18 19 #include "clang/Basic/IdentifierTable.h" 20 #include "clang/Basic/OperatorPrecedence.h" 21 #include "clang/Format/Format.h" 22 #include "clang/Lex/Lexer.h" 23 #include <memory> 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(AttributeParen) \ 34 TYPE(AttributeSquare) \ 35 TYPE(BinaryOperator) \ 36 TYPE(BitFieldColon) \ 37 TYPE(BlockComment) \ 38 TYPE(CastRParen) \ 39 TYPE(ConditionalExpr) \ 40 TYPE(ConflictAlternative) \ 41 TYPE(ConflictEnd) \ 42 TYPE(ConflictStart) \ 43 TYPE(CtorInitializerColon) \ 44 TYPE(CtorInitializerComma) \ 45 TYPE(DesignatedInitializerLSquare) \ 46 TYPE(DesignatedInitializerPeriod) \ 47 TYPE(DictLiteral) \ 48 TYPE(ForEachMacro) \ 49 TYPE(FunctionAnnotationRParen) \ 50 TYPE(FunctionDeclarationName) \ 51 TYPE(FunctionLBrace) \ 52 TYPE(FunctionTypeLParen) \ 53 TYPE(ImplicitStringLiteral) \ 54 TYPE(InheritanceColon) \ 55 TYPE(InheritanceComma) \ 56 TYPE(InlineASMBrace) \ 57 TYPE(InlineASMColon) \ 58 TYPE(JavaAnnotation) \ 59 TYPE(JsComputedPropertyName) \ 60 TYPE(JsExponentiation) \ 61 TYPE(JsExponentiationEqual) \ 62 TYPE(JsFatArrow) \ 63 TYPE(JsNonNullAssertion) \ 64 TYPE(JsTypeColon) \ 65 TYPE(JsTypeOperator) \ 66 TYPE(JsTypeOptionalQuestion) \ 67 TYPE(LambdaArrow) \ 68 TYPE(LambdaLSquare) \ 69 TYPE(LeadingJavaAnnotation) \ 70 TYPE(LineComment) \ 71 TYPE(MacroBlockBegin) \ 72 TYPE(MacroBlockEnd) \ 73 TYPE(ObjCBlockLBrace) \ 74 TYPE(ObjCBlockLParen) \ 75 TYPE(ObjCDecl) \ 76 TYPE(ObjCForIn) \ 77 TYPE(ObjCMethodExpr) \ 78 TYPE(ObjCMethodSpecifier) \ 79 TYPE(ObjCProperty) \ 80 TYPE(ObjCStringLiteral) \ 81 TYPE(OverloadedOperator) \ 82 TYPE(OverloadedOperatorLParen) \ 83 TYPE(PointerOrReference) \ 84 TYPE(PureVirtualSpecifier) \ 85 TYPE(RangeBasedForLoopColon) \ 86 TYPE(RegexLiteral) \ 87 TYPE(SelectorName) \ 88 TYPE(StartOfName) \ 89 TYPE(StructuredBindingLSquare) \ 90 TYPE(TemplateCloser) \ 91 TYPE(TemplateOpener) \ 92 TYPE(TemplateString) \ 93 TYPE(ProtoExtensionLSquare) \ 94 TYPE(TrailingAnnotation) \ 95 TYPE(TrailingReturnArrow) \ 96 TYPE(TrailingUnaryOperator) \ 97 TYPE(UnaryOperator) \ 98 TYPE(Unknown) 99 100 enum TokenType { 101 #define TYPE(X) TT_##X, 102 LIST_TOKEN_TYPES 103 #undef TYPE 104 NUM_TOKEN_TYPES 105 }; 106 107 /// Determines the name of a token type. 108 const char *getTokenTypeName(TokenType Type); 109 110 // Represents what type of block a set of braces open. 111 enum BraceBlockKind { BK_Unknown, BK_Block, BK_BracedInit }; 112 113 // The packing kind of a function's parameters. 114 enum ParameterPackingKind { PPK_BinPacked, PPK_OnePerLine, PPK_Inconclusive }; 115 116 enum FormatDecision { FD_Unformatted, FD_Continue, FD_Break }; 117 118 class TokenRole; 119 class AnnotatedLine; 120 121 /// A wrapper around a \c Token storing information about the 122 /// whitespace characters preceding it. 123 struct FormatToken { FormatTokenFormatToken124 FormatToken() {} 125 126 /// The \c Token. 127 Token Tok; 128 129 /// The number of newlines immediately before the \c Token. 130 /// 131 /// This can be used to determine what the user wrote in the original code 132 /// and thereby e.g. leave an empty line between two function definitions. 133 unsigned NewlinesBefore = 0; 134 135 /// Whether there is at least one unescaped newline before the \c 136 /// Token. 137 bool HasUnescapedNewline = false; 138 139 /// The range of the whitespace immediately preceding the \c Token. 140 SourceRange WhitespaceRange; 141 142 /// The offset just past the last '\n' in this token's leading 143 /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'. 144 unsigned LastNewlineOffset = 0; 145 146 /// The width of the non-whitespace parts of the token (or its first 147 /// line for multi-line tokens) in columns. 148 /// We need this to correctly measure number of columns a token spans. 149 unsigned ColumnWidth = 0; 150 151 /// Contains the width in columns of the last line of a multi-line 152 /// token. 153 unsigned LastLineColumnWidth = 0; 154 155 /// Whether the token text contains newlines (escaped or not). 156 bool IsMultiline = false; 157 158 /// Indicates that this is the first token of the file. 159 bool IsFirst = false; 160 161 /// Whether there must be a line break before this token. 162 /// 163 /// This happens for example when a preprocessor directive ended directly 164 /// before the token. 165 bool MustBreakBefore = false; 166 167 /// The raw text of the token. 168 /// 169 /// Contains the raw token text without leading whitespace and without leading 170 /// escaped newlines. 171 StringRef TokenText; 172 173 /// Set to \c true if this token is an unterminated literal. 174 bool IsUnterminatedLiteral = 0; 175 176 /// Contains the kind of block if this token is a brace. 177 BraceBlockKind BlockKind = BK_Unknown; 178 179 TokenType Type = TT_Unknown; 180 181 /// The number of spaces that should be inserted before this token. 182 unsigned SpacesRequiredBefore = 0; 183 184 /// \c true if it is allowed to break before this token. 185 bool CanBreakBefore = false; 186 187 /// \c true if this is the ">" of "template<..>". 188 bool ClosesTemplateDeclaration = false; 189 190 /// Number of parameters, if this is "(", "[" or "<". 191 /// 192 /// This is initialized to 1 as we don't need to distinguish functions with 193 /// 0 parameters from functions with 1 parameter. Thus, we can simply count 194 /// the number of commas. 195 unsigned ParameterCount = 0; 196 197 /// Number of parameters that are nested blocks, 198 /// if this is "(", "[" or "<". 199 unsigned BlockParameterCount = 0; 200 201 /// If this is a bracket ("<", "(", "[" or "{"), contains the kind of 202 /// the surrounding bracket. 203 tok::TokenKind ParentBracket = tok::unknown; 204 205 /// A token can have a special role that can carry extra information 206 /// about the token's formatting. 207 std::unique_ptr<TokenRole> Role; 208 209 /// If this is an opening parenthesis, how are the parameters packed? 210 ParameterPackingKind PackingKind = PPK_Inconclusive; 211 212 /// The total length of the unwrapped line up to and including this 213 /// token. 214 unsigned TotalLength = 0; 215 216 /// The original 0-based column of this token, including expanded tabs. 217 /// The configured TabWidth is used as tab width. 218 unsigned OriginalColumn = 0; 219 220 /// The length of following tokens until the next natural split point, 221 /// or the next token that can be broken. 222 unsigned UnbreakableTailLength = 0; 223 224 // FIXME: Come up with a 'cleaner' concept. 225 /// The binding strength of a token. This is a combined value of 226 /// operator precedence, parenthesis nesting, etc. 227 unsigned BindingStrength = 0; 228 229 /// The nesting level of this token, i.e. the number of surrounding (), 230 /// [], {} or <>. 231 unsigned NestingLevel = 0; 232 233 /// The indent level of this token. Copied from the surrounding line. 234 unsigned IndentLevel = 0; 235 236 /// Penalty for inserting a line break before this token. 237 unsigned SplitPenalty = 0; 238 239 /// If this is the first ObjC selector name in an ObjC method 240 /// definition or call, this contains the length of the longest name. 241 /// 242 /// This being set to 0 means that the selectors should not be colon-aligned, 243 /// e.g. because several of them are block-type. 244 unsigned LongestObjCSelectorName = 0; 245 246 /// If this is the first ObjC selector name in an ObjC method 247 /// definition or call, this contains the number of parts that the whole 248 /// selector consist of. 249 unsigned ObjCSelectorNameParts = 0; 250 251 /// The 0-based index of the parameter/argument. For ObjC it is set 252 /// for the selector name token. 253 /// For now calculated only for ObjC. 254 unsigned ParameterIndex = 0; 255 256 /// Stores the number of required fake parentheses and the 257 /// corresponding operator precedence. 258 /// 259 /// If multiple fake parentheses start at a token, this vector stores them in 260 /// reverse order, i.e. inner fake parenthesis first. 261 SmallVector<prec::Level, 4> FakeLParens; 262 /// Insert this many fake ) after this token for correct indentation. 263 unsigned FakeRParens = 0; 264 265 /// \c true if this token starts a binary expression, i.e. has at least 266 /// one fake l_paren with a precedence greater than prec::Unknown. 267 bool StartsBinaryExpression = false; 268 /// \c true if this token ends a binary expression. 269 bool EndsBinaryExpression = false; 270 271 /// Is this is an operator (or "."/"->") in a sequence of operators 272 /// with the same precedence, contains the 0-based operator index. 273 unsigned OperatorIndex = 0; 274 275 /// If this is an operator (or "."/"->") in a sequence of operators 276 /// with the same precedence, points to the next operator. 277 FormatToken *NextOperator = nullptr; 278 279 /// Is this token part of a \c DeclStmt defining multiple variables? 280 /// 281 /// Only set if \c Type == \c TT_StartOfName. 282 bool PartOfMultiVariableDeclStmt = false; 283 284 /// Does this line comment continue a line comment section? 285 /// 286 /// Only set to true if \c Type == \c TT_LineComment. 287 bool ContinuesLineCommentSection = false; 288 289 /// If this is a bracket, this points to the matching one. 290 FormatToken *MatchingParen = nullptr; 291 292 /// The previous token in the unwrapped line. 293 FormatToken *Previous = nullptr; 294 295 /// The next token in the unwrapped line. 296 FormatToken *Next = nullptr; 297 298 /// If this token starts a block, this contains all the unwrapped lines 299 /// in it. 300 SmallVector<AnnotatedLine *, 1> Children; 301 302 /// Stores the formatting decision for the token once it was made. 303 FormatDecision Decision = FD_Unformatted; 304 305 /// If \c true, this token has been fully formatted (indented and 306 /// potentially re-formatted inside), and we do not allow further formatting 307 /// changes. 308 bool Finalized = false; 309 isFormatToken310 bool is(tok::TokenKind Kind) const { return Tok.is(Kind); } isFormatToken311 bool is(TokenType TT) const { return Type == TT; } isFormatToken312 bool is(const IdentifierInfo *II) const { 313 return II && II == Tok.getIdentifierInfo(); 314 } isFormatToken315 bool is(tok::PPKeywordKind Kind) const { 316 return Tok.getIdentifierInfo() && 317 Tok.getIdentifierInfo()->getPPKeywordID() == Kind; 318 } isOneOfFormatToken319 template <typename A, typename B> bool isOneOf(A K1, B K2) const { 320 return is(K1) || is(K2); 321 } 322 template <typename A, typename B, typename... Ts> isOneOfFormatToken323 bool isOneOf(A K1, B K2, Ts... Ks) const { 324 return is(K1) || isOneOf(K2, Ks...); 325 } isNotFormatToken326 template <typename T> bool isNot(T Kind) const { return !is(Kind); } 327 328 /// \c true if this token starts a sequence with the given tokens in order, 329 /// following the ``Next`` pointers, ignoring comments. 330 template <typename A, typename... Ts> startsSequenceFormatToken331 bool startsSequence(A K1, Ts... Tokens) const { 332 return startsSequenceInternal(K1, Tokens...); 333 } 334 335 /// \c true if this token ends a sequence with the given tokens in order, 336 /// following the ``Previous`` pointers, ignoring comments. 337 template <typename A, typename... Ts> endsSequenceFormatToken338 bool endsSequence(A K1, Ts... Tokens) const { 339 return endsSequenceInternal(K1, Tokens...); 340 } 341 isStringLiteralFormatToken342 bool isStringLiteral() const { return tok::isStringLiteral(Tok.getKind()); } 343 isObjCAtKeywordFormatToken344 bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const { 345 return Tok.isObjCAtKeyword(Kind); 346 } 347 348 bool isAccessSpecifier(bool ColonRequired = true) const { 349 return isOneOf(tok::kw_public, tok::kw_protected, tok::kw_private) && 350 (!ColonRequired || (Next && Next->is(tok::colon))); 351 } 352 353 /// Determine whether the token is a simple-type-specifier. 354 bool isSimpleTypeSpecifier() const; 355 isObjCAccessSpecifierFormatToken356 bool isObjCAccessSpecifier() const { 357 return is(tok::at) && Next && 358 (Next->isObjCAtKeyword(tok::objc_public) || 359 Next->isObjCAtKeyword(tok::objc_protected) || 360 Next->isObjCAtKeyword(tok::objc_package) || 361 Next->isObjCAtKeyword(tok::objc_private)); 362 } 363 364 /// Returns whether \p Tok is ([{ or an opening < of a template or in 365 /// protos. opensScopeFormatToken366 bool opensScope() const { 367 if (is(TT_TemplateString) && TokenText.endswith("${")) 368 return true; 369 if (is(TT_DictLiteral) && is(tok::less)) 370 return true; 371 return isOneOf(tok::l_paren, tok::l_brace, tok::l_square, 372 TT_TemplateOpener); 373 } 374 /// Returns whether \p Tok is )]} or a closing > of a template or in 375 /// protos. closesScopeFormatToken376 bool closesScope() const { 377 if (is(TT_TemplateString) && TokenText.startswith("}")) 378 return true; 379 if (is(TT_DictLiteral) && is(tok::greater)) 380 return true; 381 return isOneOf(tok::r_paren, tok::r_brace, tok::r_square, 382 TT_TemplateCloser); 383 } 384 385 /// Returns \c true if this is a "." or "->" accessing a member. isMemberAccessFormatToken386 bool isMemberAccess() const { 387 return isOneOf(tok::arrow, tok::period, tok::arrowstar) && 388 !isOneOf(TT_DesignatedInitializerPeriod, TT_TrailingReturnArrow, 389 TT_LambdaArrow); 390 } 391 isUnaryOperatorFormatToken392 bool isUnaryOperator() const { 393 switch (Tok.getKind()) { 394 case tok::plus: 395 case tok::plusplus: 396 case tok::minus: 397 case tok::minusminus: 398 case tok::exclaim: 399 case tok::tilde: 400 case tok::kw_sizeof: 401 case tok::kw_alignof: 402 return true; 403 default: 404 return false; 405 } 406 } 407 isBinaryOperatorFormatToken408 bool isBinaryOperator() const { 409 // Comma is a binary operator, but does not behave as such wrt. formatting. 410 return getPrecedence() > prec::Comma; 411 } 412 isTrailingCommentFormatToken413 bool isTrailingComment() const { 414 return is(tok::comment) && 415 (is(TT_LineComment) || !Next || Next->NewlinesBefore > 0); 416 } 417 418 /// Returns \c true if this is a keyword that can be used 419 /// like a function call (e.g. sizeof, typeid, ...). isFunctionLikeKeywordFormatToken420 bool isFunctionLikeKeyword() const { 421 switch (Tok.getKind()) { 422 case tok::kw_throw: 423 case tok::kw_typeid: 424 case tok::kw_return: 425 case tok::kw_sizeof: 426 case tok::kw_alignof: 427 case tok::kw_alignas: 428 case tok::kw_decltype: 429 case tok::kw_noexcept: 430 case tok::kw_static_assert: 431 case tok::kw___attribute: 432 return true; 433 default: 434 return false; 435 } 436 } 437 438 /// Returns \c true if this is a string literal that's like a label, 439 /// e.g. ends with "=" or ":". isLabelStringFormatToken440 bool isLabelString() const { 441 if (!is(tok::string_literal)) 442 return false; 443 StringRef Content = TokenText; 444 if (Content.startswith("\"") || Content.startswith("'")) 445 Content = Content.drop_front(1); 446 if (Content.endswith("\"") || Content.endswith("'")) 447 Content = Content.drop_back(1); 448 Content = Content.trim(); 449 return Content.size() > 1 && 450 (Content.back() == ':' || Content.back() == '='); 451 } 452 453 /// Returns actual token start location without leading escaped 454 /// newlines and whitespace. 455 /// 456 /// This can be different to Tok.getLocation(), which includes leading escaped 457 /// newlines. getStartOfNonWhitespaceFormatToken458 SourceLocation getStartOfNonWhitespace() const { 459 return WhitespaceRange.getEnd(); 460 } 461 getPrecedenceFormatToken462 prec::Level getPrecedence() const { 463 return getBinOpPrecedence(Tok.getKind(), /*GreaterThanIsOperator=*/true, 464 /*CPlusPlus11=*/true); 465 } 466 467 /// Returns the previous token ignoring comments. getPreviousNonCommentFormatToken468 FormatToken *getPreviousNonComment() const { 469 FormatToken *Tok = Previous; 470 while (Tok && Tok->is(tok::comment)) 471 Tok = Tok->Previous; 472 return Tok; 473 } 474 475 /// Returns the next token ignoring comments. getNextNonCommentFormatToken476 const FormatToken *getNextNonComment() const { 477 const FormatToken *Tok = Next; 478 while (Tok && Tok->is(tok::comment)) 479 Tok = Tok->Next; 480 return Tok; 481 } 482 483 /// Returns \c true if this tokens starts a block-type list, i.e. a 484 /// list that should be indented with a block indent. opensBlockOrBlockTypeListFormatToken485 bool opensBlockOrBlockTypeList(const FormatStyle &Style) const { 486 if (is(TT_TemplateString) && opensScope()) 487 return true; 488 return is(TT_ArrayInitializerLSquare) || 489 is(TT_ProtoExtensionLSquare) || 490 (is(tok::l_brace) && 491 (BlockKind == BK_Block || is(TT_DictLiteral) || 492 (!Style.Cpp11BracedListStyle && NestingLevel == 0))) || 493 (is(tok::less) && (Style.Language == FormatStyle::LK_Proto || 494 Style.Language == FormatStyle::LK_TextProto)); 495 } 496 497 /// Returns whether the token is the left square bracket of a C++ 498 /// structured binding declaration. isCppStructuredBindingFormatToken499 bool isCppStructuredBinding(const FormatStyle &Style) const { 500 if (!Style.isCpp() || isNot(tok::l_square)) 501 return false; 502 const FormatToken *T = this; 503 do { 504 T = T->getPreviousNonComment(); 505 } while (T && T->isOneOf(tok::kw_const, tok::kw_volatile, tok::amp, 506 tok::ampamp)); 507 return T && T->is(tok::kw_auto); 508 } 509 510 /// Same as opensBlockOrBlockTypeList, but for the closing token. closesBlockOrBlockTypeListFormatToken511 bool closesBlockOrBlockTypeList(const FormatStyle &Style) const { 512 if (is(TT_TemplateString) && closesScope()) 513 return true; 514 return MatchingParen && MatchingParen->opensBlockOrBlockTypeList(Style); 515 } 516 517 /// Return the actual namespace token, if this token starts a namespace 518 /// block. getNamespaceTokenFormatToken519 const FormatToken *getNamespaceToken() const { 520 const FormatToken *NamespaceTok = this; 521 if (is(tok::comment)) 522 NamespaceTok = NamespaceTok->getNextNonComment(); 523 // Detect "(inline)? namespace" in the beginning of a line. 524 if (NamespaceTok && NamespaceTok->is(tok::kw_inline)) 525 NamespaceTok = NamespaceTok->getNextNonComment(); 526 return NamespaceTok && NamespaceTok->is(tok::kw_namespace) ? NamespaceTok 527 : nullptr; 528 } 529 530 private: 531 // Disallow copying. 532 FormatToken(const FormatToken &) = delete; 533 void operator=(const FormatToken &) = delete; 534 535 template <typename A, typename... Ts> startsSequenceInternalFormatToken536 bool startsSequenceInternal(A K1, Ts... Tokens) const { 537 if (is(tok::comment) && Next) 538 return Next->startsSequenceInternal(K1, Tokens...); 539 return is(K1) && Next && Next->startsSequenceInternal(Tokens...); 540 } 541 startsSequenceInternalFormatToken542 template <typename A> bool startsSequenceInternal(A K1) const { 543 if (is(tok::comment) && Next) 544 return Next->startsSequenceInternal(K1); 545 return is(K1); 546 } 547 endsSequenceInternalFormatToken548 template <typename A, typename... Ts> bool endsSequenceInternal(A K1) const { 549 if (is(tok::comment) && Previous) 550 return Previous->endsSequenceInternal(K1); 551 return is(K1); 552 } 553 554 template <typename A, typename... Ts> endsSequenceInternalFormatToken555 bool endsSequenceInternal(A K1, Ts... Tokens) const { 556 if (is(tok::comment) && Previous) 557 return Previous->endsSequenceInternal(K1, Tokens...); 558 return is(K1) && Previous && Previous->endsSequenceInternal(Tokens...); 559 } 560 }; 561 562 class ContinuationIndenter; 563 struct LineState; 564 565 class TokenRole { 566 public: TokenRole(const FormatStyle & Style)567 TokenRole(const FormatStyle &Style) : Style(Style) {} 568 virtual ~TokenRole(); 569 570 /// After the \c TokenAnnotator has finished annotating all the tokens, 571 /// this function precomputes required information for formatting. 572 virtual void precomputeFormattingInfos(const FormatToken *Token); 573 574 /// Apply the special formatting that the given role demands. 575 /// 576 /// Assumes that the token having this role is already formatted. 577 /// 578 /// Continues formatting from \p State leaving indentation to \p Indenter and 579 /// returns the total penalty that this formatting incurs. formatFromToken(LineState & State,ContinuationIndenter * Indenter,bool DryRun)580 virtual unsigned formatFromToken(LineState &State, 581 ContinuationIndenter *Indenter, 582 bool DryRun) { 583 return 0; 584 } 585 586 /// Same as \c formatFromToken, but assumes that the first token has 587 /// already been set thereby deciding on the first line break. formatAfterToken(LineState & State,ContinuationIndenter * Indenter,bool DryRun)588 virtual unsigned formatAfterToken(LineState &State, 589 ContinuationIndenter *Indenter, 590 bool DryRun) { 591 return 0; 592 } 593 594 /// Notifies the \c Role that a comma was found. CommaFound(const FormatToken * Token)595 virtual void CommaFound(const FormatToken *Token) {} 596 597 protected: 598 const FormatStyle &Style; 599 }; 600 601 class CommaSeparatedList : public TokenRole { 602 public: CommaSeparatedList(const FormatStyle & Style)603 CommaSeparatedList(const FormatStyle &Style) 604 : TokenRole(Style), HasNestedBracedList(false) {} 605 606 void precomputeFormattingInfos(const FormatToken *Token) override; 607 608 unsigned formatAfterToken(LineState &State, ContinuationIndenter *Indenter, 609 bool DryRun) override; 610 611 unsigned formatFromToken(LineState &State, ContinuationIndenter *Indenter, 612 bool DryRun) override; 613 614 /// Adds \p Token as the next comma to the \c CommaSeparated list. CommaFound(const FormatToken * Token)615 void CommaFound(const FormatToken *Token) override { 616 Commas.push_back(Token); 617 } 618 619 private: 620 /// A struct that holds information on how to format a given list with 621 /// a specific number of columns. 622 struct ColumnFormat { 623 /// The number of columns to use. 624 unsigned Columns; 625 626 /// The total width in characters. 627 unsigned TotalWidth; 628 629 /// The number of lines required for this format. 630 unsigned LineCount; 631 632 /// The size of each column in characters. 633 SmallVector<unsigned, 8> ColumnSizes; 634 }; 635 636 /// Calculate which \c ColumnFormat fits best into 637 /// \p RemainingCharacters. 638 const ColumnFormat *getColumnFormat(unsigned RemainingCharacters) const; 639 640 /// The ordered \c FormatTokens making up the commas of this list. 641 SmallVector<const FormatToken *, 8> Commas; 642 643 /// The length of each of the list's items in characters including the 644 /// trailing comma. 645 SmallVector<unsigned, 8> ItemLengths; 646 647 /// Precomputed formats that can be used for this list. 648 SmallVector<ColumnFormat, 4> Formats; 649 650 bool HasNestedBracedList; 651 }; 652 653 /// Encapsulates keywords that are context sensitive or for languages not 654 /// properly supported by Clang's lexer. 655 struct AdditionalKeywords { AdditionalKeywordsAdditionalKeywords656 AdditionalKeywords(IdentifierTable &IdentTable) { 657 kw_final = &IdentTable.get("final"); 658 kw_override = &IdentTable.get("override"); 659 kw_in = &IdentTable.get("in"); 660 kw_of = &IdentTable.get("of"); 661 kw_CF_ENUM = &IdentTable.get("CF_ENUM"); 662 kw_CF_OPTIONS = &IdentTable.get("CF_OPTIONS"); 663 kw_NS_ENUM = &IdentTable.get("NS_ENUM"); 664 kw_NS_OPTIONS = &IdentTable.get("NS_OPTIONS"); 665 666 kw_as = &IdentTable.get("as"); 667 kw_async = &IdentTable.get("async"); 668 kw_await = &IdentTable.get("await"); 669 kw_declare = &IdentTable.get("declare"); 670 kw_finally = &IdentTable.get("finally"); 671 kw_from = &IdentTable.get("from"); 672 kw_function = &IdentTable.get("function"); 673 kw_get = &IdentTable.get("get"); 674 kw_import = &IdentTable.get("import"); 675 kw_is = &IdentTable.get("is"); 676 kw_let = &IdentTable.get("let"); 677 kw_module = &IdentTable.get("module"); 678 kw_readonly = &IdentTable.get("readonly"); 679 kw_set = &IdentTable.get("set"); 680 kw_type = &IdentTable.get("type"); 681 kw_typeof = &IdentTable.get("typeof"); 682 kw_var = &IdentTable.get("var"); 683 kw_yield = &IdentTable.get("yield"); 684 685 kw_abstract = &IdentTable.get("abstract"); 686 kw_assert = &IdentTable.get("assert"); 687 kw_extends = &IdentTable.get("extends"); 688 kw_implements = &IdentTable.get("implements"); 689 kw_instanceof = &IdentTable.get("instanceof"); 690 kw_interface = &IdentTable.get("interface"); 691 kw_native = &IdentTable.get("native"); 692 kw_package = &IdentTable.get("package"); 693 kw_synchronized = &IdentTable.get("synchronized"); 694 kw_throws = &IdentTable.get("throws"); 695 kw___except = &IdentTable.get("__except"); 696 kw___has_include = &IdentTable.get("__has_include"); 697 kw___has_include_next = &IdentTable.get("__has_include_next"); 698 699 kw_mark = &IdentTable.get("mark"); 700 701 kw_extend = &IdentTable.get("extend"); 702 kw_option = &IdentTable.get("option"); 703 kw_optional = &IdentTable.get("optional"); 704 kw_repeated = &IdentTable.get("repeated"); 705 kw_required = &IdentTable.get("required"); 706 kw_returns = &IdentTable.get("returns"); 707 708 kw_signals = &IdentTable.get("signals"); 709 kw_qsignals = &IdentTable.get("Q_SIGNALS"); 710 kw_slots = &IdentTable.get("slots"); 711 kw_qslots = &IdentTable.get("Q_SLOTS"); 712 713 // Keep this at the end of the constructor to make sure everything here is 714 // already initialized. 715 JsExtraKeywords = std::unordered_set<IdentifierInfo *>( 716 {kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from, 717 kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_readonly, 718 kw_set, kw_type, kw_typeof, kw_var, kw_yield, 719 // Keywords from the Java section. 720 kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface}); 721 } 722 723 // Context sensitive keywords. 724 IdentifierInfo *kw_final; 725 IdentifierInfo *kw_override; 726 IdentifierInfo *kw_in; 727 IdentifierInfo *kw_of; 728 IdentifierInfo *kw_CF_ENUM; 729 IdentifierInfo *kw_CF_OPTIONS; 730 IdentifierInfo *kw_NS_ENUM; 731 IdentifierInfo *kw_NS_OPTIONS; 732 IdentifierInfo *kw___except; 733 IdentifierInfo *kw___has_include; 734 IdentifierInfo *kw___has_include_next; 735 736 // JavaScript keywords. 737 IdentifierInfo *kw_as; 738 IdentifierInfo *kw_async; 739 IdentifierInfo *kw_await; 740 IdentifierInfo *kw_declare; 741 IdentifierInfo *kw_finally; 742 IdentifierInfo *kw_from; 743 IdentifierInfo *kw_function; 744 IdentifierInfo *kw_get; 745 IdentifierInfo *kw_import; 746 IdentifierInfo *kw_is; 747 IdentifierInfo *kw_let; 748 IdentifierInfo *kw_module; 749 IdentifierInfo *kw_readonly; 750 IdentifierInfo *kw_set; 751 IdentifierInfo *kw_type; 752 IdentifierInfo *kw_typeof; 753 IdentifierInfo *kw_var; 754 IdentifierInfo *kw_yield; 755 756 // Java keywords. 757 IdentifierInfo *kw_abstract; 758 IdentifierInfo *kw_assert; 759 IdentifierInfo *kw_extends; 760 IdentifierInfo *kw_implements; 761 IdentifierInfo *kw_instanceof; 762 IdentifierInfo *kw_interface; 763 IdentifierInfo *kw_native; 764 IdentifierInfo *kw_package; 765 IdentifierInfo *kw_synchronized; 766 IdentifierInfo *kw_throws; 767 768 // Pragma keywords. 769 IdentifierInfo *kw_mark; 770 771 // Proto keywords. 772 IdentifierInfo *kw_extend; 773 IdentifierInfo *kw_option; 774 IdentifierInfo *kw_optional; 775 IdentifierInfo *kw_repeated; 776 IdentifierInfo *kw_required; 777 IdentifierInfo *kw_returns; 778 779 // QT keywords. 780 IdentifierInfo *kw_signals; 781 IdentifierInfo *kw_qsignals; 782 IdentifierInfo *kw_slots; 783 IdentifierInfo *kw_qslots; 784 785 /// Returns \c true if \p Tok is a true JavaScript identifier, returns 786 /// \c false if it is a keyword or a pseudo keyword. IsJavaScriptIdentifierAdditionalKeywords787 bool IsJavaScriptIdentifier(const FormatToken &Tok) const { 788 return Tok.is(tok::identifier) && 789 JsExtraKeywords.find(Tok.Tok.getIdentifierInfo()) == 790 JsExtraKeywords.end(); 791 } 792 793 private: 794 /// The JavaScript keywords beyond the C++ keyword set. 795 std::unordered_set<IdentifierInfo *> JsExtraKeywords; 796 }; 797 798 } // namespace format 799 } // namespace clang 800 801 #endif 802