| File: | clang/lib/Format/ContinuationIndenter.cpp |
| Warning: | line 595, column 9 Access to field 'PartOfMultiVariableDeclStmt' results in a dereference of a null pointer (loaded from variable 'Previous') |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
| 1 | //===--- ContinuationIndenter.cpp - Format C++ code -----------------------===// | |||
| 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 implements the continuation indenter. | |||
| 11 | /// | |||
| 12 | //===----------------------------------------------------------------------===// | |||
| 13 | ||||
| 14 | #include "ContinuationIndenter.h" | |||
| 15 | #include "BreakableToken.h" | |||
| 16 | #include "FormatInternal.h" | |||
| 17 | #include "FormatToken.h" | |||
| 18 | #include "WhitespaceManager.h" | |||
| 19 | #include "clang/Basic/OperatorPrecedence.h" | |||
| 20 | #include "clang/Basic/SourceManager.h" | |||
| 21 | #include "clang/Format/Format.h" | |||
| 22 | #include "llvm/ADT/StringSet.h" | |||
| 23 | #include "llvm/Support/Debug.h" | |||
| 24 | ||||
| 25 | #define DEBUG_TYPE"format-indenter" "format-indenter" | |||
| 26 | ||||
| 27 | namespace clang { | |||
| 28 | namespace format { | |||
| 29 | ||||
| 30 | // Returns true if a TT_SelectorName should be indented when wrapped, | |||
| 31 | // false otherwise. | |||
| 32 | static bool shouldIndentWrappedSelectorName(const FormatStyle &Style, | |||
| 33 | LineType LineType) { | |||
| 34 | return Style.IndentWrappedFunctionNames || LineType == LT_ObjCMethodDecl; | |||
| 35 | } | |||
| 36 | ||||
| 37 | // Returns the length of everything up to the first possible line break after | |||
| 38 | // the ), ], } or > matching \c Tok. | |||
| 39 | static unsigned getLengthToMatchingParen(const FormatToken &Tok, | |||
| 40 | const std::vector<ParenState> &Stack) { | |||
| 41 | // Normally whether or not a break before T is possible is calculated and | |||
| 42 | // stored in T.CanBreakBefore. Braces, array initializers and text proto | |||
| 43 | // messages like `key: < ... >` are an exception: a break is possible | |||
| 44 | // before a closing brace R if a break was inserted after the corresponding | |||
| 45 | // opening brace. The information about whether or not a break is needed | |||
| 46 | // before a closing brace R is stored in the ParenState field | |||
| 47 | // S.BreakBeforeClosingBrace where S is the state that R closes. | |||
| 48 | // | |||
| 49 | // In order to decide whether there can be a break before encountered right | |||
| 50 | // braces, this implementation iterates over the sequence of tokens and over | |||
| 51 | // the paren stack in lockstep, keeping track of the stack level which visited | |||
| 52 | // right braces correspond to in MatchingStackIndex. | |||
| 53 | // | |||
| 54 | // For example, consider: | |||
| 55 | // L. <- line number | |||
| 56 | // 1. { | |||
| 57 | // 2. {1}, | |||
| 58 | // 3. {2}, | |||
| 59 | // 4. {{3}}} | |||
| 60 | // ^ where we call this method with this token. | |||
| 61 | // The paren stack at this point contains 3 brace levels: | |||
| 62 | // 0. { at line 1, BreakBeforeClosingBrace: true | |||
| 63 | // 1. first { at line 4, BreakBeforeClosingBrace: false | |||
| 64 | // 2. second { at line 4, BreakBeforeClosingBrace: false, | |||
| 65 | // where there might be fake parens levels in-between these levels. | |||
| 66 | // The algorithm will start at the first } on line 4, which is the matching | |||
| 67 | // brace of the initial left brace and at level 2 of the stack. Then, | |||
| 68 | // examining BreakBeforeClosingBrace: false at level 2, it will continue to | |||
| 69 | // the second } on line 4, and will traverse the stack downwards until it | |||
| 70 | // finds the matching { on level 1. Then, examining BreakBeforeClosingBrace: | |||
| 71 | // false at level 1, it will continue to the third } on line 4 and will | |||
| 72 | // traverse the stack downwards until it finds the matching { on level 0. | |||
| 73 | // Then, examining BreakBeforeClosingBrace: true at level 0, the algorithm | |||
| 74 | // will stop and will use the second } on line 4 to determine the length to | |||
| 75 | // return, as in this example the range will include the tokens: {3}} | |||
| 76 | // | |||
| 77 | // The algorithm will only traverse the stack if it encounters braces, array | |||
| 78 | // initializer squares or text proto angle brackets. | |||
| 79 | if (!Tok.MatchingParen) | |||
| 80 | return 0; | |||
| 81 | FormatToken *End = Tok.MatchingParen; | |||
| 82 | // Maintains a stack level corresponding to the current End token. | |||
| 83 | int MatchingStackIndex = Stack.size() - 1; | |||
| 84 | // Traverses the stack downwards, looking for the level to which LBrace | |||
| 85 | // corresponds. Returns either a pointer to the matching level or nullptr if | |||
| 86 | // LParen is not found in the initial portion of the stack up to | |||
| 87 | // MatchingStackIndex. | |||
| 88 | auto FindParenState = [&](const FormatToken *LBrace) -> const ParenState * { | |||
| 89 | while (MatchingStackIndex >= 0 && Stack[MatchingStackIndex].Tok != LBrace) | |||
| 90 | --MatchingStackIndex; | |||
| 91 | return MatchingStackIndex >= 0 ? &Stack[MatchingStackIndex] : nullptr; | |||
| 92 | }; | |||
| 93 | for (; End->Next; End = End->Next) { | |||
| 94 | if (End->Next->CanBreakBefore) | |||
| 95 | break; | |||
| 96 | if (!End->Next->closesScope()) | |||
| 97 | continue; | |||
| 98 | if (End->Next->MatchingParen && | |||
| 99 | End->Next->MatchingParen->isOneOf( | |||
| 100 | tok::l_brace, TT_ArrayInitializerLSquare, tok::less)) { | |||
| 101 | const ParenState *State = FindParenState(End->Next->MatchingParen); | |||
| 102 | if (State && State->BreakBeforeClosingBrace) | |||
| 103 | break; | |||
| 104 | } | |||
| 105 | } | |||
| 106 | return End->TotalLength - Tok.TotalLength + 1; | |||
| 107 | } | |||
| 108 | ||||
| 109 | static unsigned getLengthToNextOperator(const FormatToken &Tok) { | |||
| 110 | if (!Tok.NextOperator) | |||
| 111 | return 0; | |||
| 112 | return Tok.NextOperator->TotalLength - Tok.TotalLength; | |||
| 113 | } | |||
| 114 | ||||
| 115 | // Returns \c true if \c Tok is the "." or "->" of a call and starts the next | |||
| 116 | // segment of a builder type call. | |||
| 117 | static bool startsSegmentOfBuilderTypeCall(const FormatToken &Tok) { | |||
| 118 | return Tok.isMemberAccess() && Tok.Previous && Tok.Previous->closesScope(); | |||
| 119 | } | |||
| 120 | ||||
| 121 | // Returns \c true if \c Current starts a new parameter. | |||
| 122 | static bool startsNextParameter(const FormatToken &Current, | |||
| 123 | const FormatStyle &Style) { | |||
| 124 | const FormatToken &Previous = *Current.Previous; | |||
| 125 | if (Current.is(TT_CtorInitializerComma) && | |||
| 126 | Style.BreakConstructorInitializers == FormatStyle::BCIS_BeforeComma) | |||
| 127 | return true; | |||
| 128 | if (Style.Language == FormatStyle::LK_Proto && Current.is(TT_SelectorName)) | |||
| 129 | return true; | |||
| 130 | return Previous.is(tok::comma) && !Current.isTrailingComment() && | |||
| 131 | ((Previous.isNot(TT_CtorInitializerComma) || | |||
| 132 | Style.BreakConstructorInitializers != | |||
| 133 | FormatStyle::BCIS_BeforeComma) && | |||
| 134 | (Previous.isNot(TT_InheritanceComma) || | |||
| 135 | Style.BreakInheritanceList != FormatStyle::BILS_BeforeComma)); | |||
| 136 | } | |||
| 137 | ||||
| 138 | static bool opensProtoMessageField(const FormatToken &LessTok, | |||
| 139 | const FormatStyle &Style) { | |||
| 140 | if (LessTok.isNot(tok::less)) | |||
| 141 | return false; | |||
| 142 | return Style.Language == FormatStyle::LK_TextProto || | |||
| 143 | (Style.Language == FormatStyle::LK_Proto && | |||
| 144 | (LessTok.NestingLevel > 0 || | |||
| 145 | (LessTok.Previous && LessTok.Previous->is(tok::equal)))); | |||
| 146 | } | |||
| 147 | ||||
| 148 | // Returns the delimiter of a raw string literal, or None if TokenText is not | |||
| 149 | // the text of a raw string literal. The delimiter could be the empty string. | |||
| 150 | // For example, the delimiter of R"deli(cont)deli" is deli. | |||
| 151 | static llvm::Optional<StringRef> getRawStringDelimiter(StringRef TokenText) { | |||
| 152 | if (TokenText.size() < 5 // The smallest raw string possible is 'R"()"'. | |||
| 153 | || !TokenText.startswith("R\"") || !TokenText.endswith("\"")) | |||
| 154 | return None; | |||
| 155 | ||||
| 156 | // A raw string starts with 'R"<delimiter>(' and delimiter is ascii and has | |||
| 157 | // size at most 16 by the standard, so the first '(' must be among the first | |||
| 158 | // 19 bytes. | |||
| 159 | size_t LParenPos = TokenText.substr(0, 19).find_first_of('('); | |||
| 160 | if (LParenPos == StringRef::npos) | |||
| 161 | return None; | |||
| 162 | StringRef Delimiter = TokenText.substr(2, LParenPos - 2); | |||
| 163 | ||||
| 164 | // Check that the string ends in ')Delimiter"'. | |||
| 165 | size_t RParenPos = TokenText.size() - Delimiter.size() - 2; | |||
| 166 | if (TokenText[RParenPos] != ')') | |||
| 167 | return None; | |||
| 168 | if (!TokenText.substr(RParenPos + 1).startswith(Delimiter)) | |||
| 169 | return None; | |||
| 170 | return Delimiter; | |||
| 171 | } | |||
| 172 | ||||
| 173 | // Returns the canonical delimiter for \p Language, or the empty string if no | |||
| 174 | // canonical delimiter is specified. | |||
| 175 | static StringRef | |||
| 176 | getCanonicalRawStringDelimiter(const FormatStyle &Style, | |||
| 177 | FormatStyle::LanguageKind Language) { | |||
| 178 | for (const auto &Format : Style.RawStringFormats) { | |||
| 179 | if (Format.Language == Language) | |||
| 180 | return StringRef(Format.CanonicalDelimiter); | |||
| 181 | } | |||
| 182 | return ""; | |||
| 183 | } | |||
| 184 | ||||
| 185 | RawStringFormatStyleManager::RawStringFormatStyleManager( | |||
| 186 | const FormatStyle &CodeStyle) { | |||
| 187 | for (const auto &RawStringFormat : CodeStyle.RawStringFormats) { | |||
| 188 | llvm::Optional<FormatStyle> LanguageStyle = | |||
| 189 | CodeStyle.GetLanguageStyle(RawStringFormat.Language); | |||
| 190 | if (!LanguageStyle) { | |||
| 191 | FormatStyle PredefinedStyle; | |||
| 192 | if (!getPredefinedStyle(RawStringFormat.BasedOnStyle, | |||
| 193 | RawStringFormat.Language, &PredefinedStyle)) { | |||
| 194 | PredefinedStyle = getLLVMStyle(); | |||
| 195 | PredefinedStyle.Language = RawStringFormat.Language; | |||
| 196 | } | |||
| 197 | LanguageStyle = PredefinedStyle; | |||
| 198 | } | |||
| 199 | LanguageStyle->ColumnLimit = CodeStyle.ColumnLimit; | |||
| 200 | for (StringRef Delimiter : RawStringFormat.Delimiters) { | |||
| 201 | DelimiterStyle.insert({Delimiter, *LanguageStyle}); | |||
| 202 | } | |||
| 203 | for (StringRef EnclosingFunction : RawStringFormat.EnclosingFunctions) { | |||
| 204 | EnclosingFunctionStyle.insert({EnclosingFunction, *LanguageStyle}); | |||
| 205 | } | |||
| 206 | } | |||
| 207 | } | |||
| 208 | ||||
| 209 | llvm::Optional<FormatStyle> | |||
| 210 | RawStringFormatStyleManager::getDelimiterStyle(StringRef Delimiter) const { | |||
| 211 | auto It = DelimiterStyle.find(Delimiter); | |||
| 212 | if (It == DelimiterStyle.end()) | |||
| 213 | return None; | |||
| 214 | return It->second; | |||
| 215 | } | |||
| 216 | ||||
| 217 | llvm::Optional<FormatStyle> | |||
| 218 | RawStringFormatStyleManager::getEnclosingFunctionStyle( | |||
| 219 | StringRef EnclosingFunction) const { | |||
| 220 | auto It = EnclosingFunctionStyle.find(EnclosingFunction); | |||
| 221 | if (It == EnclosingFunctionStyle.end()) | |||
| 222 | return None; | |||
| 223 | return It->second; | |||
| 224 | } | |||
| 225 | ||||
| 226 | ContinuationIndenter::ContinuationIndenter(const FormatStyle &Style, | |||
| 227 | const AdditionalKeywords &Keywords, | |||
| 228 | const SourceManager &SourceMgr, | |||
| 229 | WhitespaceManager &Whitespaces, | |||
| 230 | encoding::Encoding Encoding, | |||
| 231 | bool BinPackInconclusiveFunctions) | |||
| 232 | : Style(Style), Keywords(Keywords), SourceMgr(SourceMgr), | |||
| 233 | Whitespaces(Whitespaces), Encoding(Encoding), | |||
| 234 | BinPackInconclusiveFunctions(BinPackInconclusiveFunctions), | |||
| 235 | CommentPragmasRegex(Style.CommentPragmas), RawStringFormats(Style) {} | |||
| 236 | ||||
| 237 | LineState ContinuationIndenter::getInitialState(unsigned FirstIndent, | |||
| 238 | unsigned FirstStartColumn, | |||
| 239 | const AnnotatedLine *Line, | |||
| 240 | bool DryRun) { | |||
| 241 | LineState State; | |||
| 242 | State.FirstIndent = FirstIndent; | |||
| 243 | if (FirstStartColumn && Line->First->NewlinesBefore == 0) | |||
| 244 | State.Column = FirstStartColumn; | |||
| 245 | else | |||
| 246 | State.Column = FirstIndent; | |||
| 247 | // With preprocessor directive indentation, the line starts on column 0 | |||
| 248 | // since it's indented after the hash, but FirstIndent is set to the | |||
| 249 | // preprocessor indent. | |||
| 250 | if (Style.IndentPPDirectives == FormatStyle::PPDIS_AfterHash && | |||
| 251 | (Line->Type == LT_PreprocessorDirective || | |||
| 252 | Line->Type == LT_ImportStatement)) | |||
| 253 | State.Column = 0; | |||
| 254 | State.Line = Line; | |||
| 255 | State.NextToken = Line->First; | |||
| 256 | State.Stack.push_back(ParenState(/*Tok=*/nullptr, FirstIndent, FirstIndent, | |||
| 257 | /*AvoidBinPacking=*/false, | |||
| 258 | /*NoLineBreak=*/false)); | |||
| 259 | State.LineContainsContinuedForLoopSection = false; | |||
| 260 | State.NoContinuation = false; | |||
| 261 | State.StartOfStringLiteral = 0; | |||
| 262 | State.StartOfLineLevel = 0; | |||
| 263 | State.LowestLevelOnLine = 0; | |||
| 264 | State.IgnoreStackForComparison = false; | |||
| 265 | ||||
| 266 | if (Style.Language == FormatStyle::LK_TextProto) { | |||
| 267 | // We need this in order to deal with the bin packing of text fields at | |||
| 268 | // global scope. | |||
| 269 | State.Stack.back().AvoidBinPacking = true; | |||
| 270 | State.Stack.back().BreakBeforeParameter = true; | |||
| 271 | State.Stack.back().AlignColons = false; | |||
| 272 | } | |||
| 273 | ||||
| 274 | // The first token has already been indented and thus consumed. | |||
| 275 | moveStateToNextToken(State, DryRun, /*Newline=*/false); | |||
| 276 | return State; | |||
| 277 | } | |||
| 278 | ||||
| 279 | bool ContinuationIndenter::canBreak(const LineState &State) { | |||
| 280 | const FormatToken &Current = *State.NextToken; | |||
| 281 | const FormatToken &Previous = *Current.Previous; | |||
| 282 | assert(&Previous == Current.Previous)(static_cast <bool> (&Previous == Current.Previous) ? void (0) : __assert_fail ("&Previous == Current.Previous" , "clang/lib/Format/ContinuationIndenter.cpp", 282, __extension__ __PRETTY_FUNCTION__)); | |||
| 283 | if (!Current.CanBreakBefore && !(State.Stack.back().BreakBeforeClosingBrace && | |||
| 284 | Current.closesBlockOrBlockTypeList(Style))) | |||
| 285 | return false; | |||
| 286 | // The opening "{" of a braced list has to be on the same line as the first | |||
| 287 | // element if it is nested in another braced init list or function call. | |||
| 288 | if (!Current.MustBreakBefore && Previous.is(tok::l_brace) && | |||
| 289 | Previous.isNot(TT_DictLiteral) && Previous.is(BK_BracedInit) && | |||
| 290 | Previous.Previous && | |||
| 291 | Previous.Previous->isOneOf(tok::l_brace, tok::l_paren, tok::comma)) | |||
| 292 | return false; | |||
| 293 | // This prevents breaks like: | |||
| 294 | // ... | |||
| 295 | // SomeParameter, OtherParameter).DoSomething( | |||
| 296 | // ... | |||
| 297 | // As they hide "DoSomething" and are generally bad for readability. | |||
| 298 | if (Previous.opensScope() && Previous.isNot(tok::l_brace) && | |||
| 299 | State.LowestLevelOnLine < State.StartOfLineLevel && | |||
| 300 | State.LowestLevelOnLine < Current.NestingLevel) | |||
| 301 | return false; | |||
| 302 | if (Current.isMemberAccess() && State.Stack.back().ContainsUnwrappedBuilder) | |||
| 303 | return false; | |||
| 304 | ||||
| 305 | // Don't create a 'hanging' indent if there are multiple blocks in a single | |||
| 306 | // statement. | |||
| 307 | if (Previous.is(tok::l_brace) && State.Stack.size() > 1 && | |||
| 308 | State.Stack[State.Stack.size() - 2].NestedBlockInlined && | |||
| 309 | State.Stack[State.Stack.size() - 2].HasMultipleNestedBlocks) | |||
| 310 | return false; | |||
| 311 | ||||
| 312 | // Don't break after very short return types (e.g. "void") as that is often | |||
| 313 | // unexpected. | |||
| 314 | if (Current.is(TT_FunctionDeclarationName) && State.Column < 6) { | |||
| 315 | if (Style.AlwaysBreakAfterReturnType == FormatStyle::RTBS_None) | |||
| 316 | return false; | |||
| 317 | } | |||
| 318 | ||||
| 319 | // If binary operators are moved to the next line (including commas for some | |||
| 320 | // styles of constructor initializers), that's always ok. | |||
| 321 | if (!Current.isOneOf(TT_BinaryOperator, tok::comma) && | |||
| 322 | State.Stack.back().NoLineBreakInOperand) | |||
| 323 | return false; | |||
| 324 | ||||
| 325 | if (Previous.is(tok::l_square) && Previous.is(TT_ObjCMethodExpr)) | |||
| 326 | return false; | |||
| 327 | ||||
| 328 | return !State.Stack.back().NoLineBreak; | |||
| 329 | } | |||
| 330 | ||||
| 331 | bool ContinuationIndenter::mustBreak(const LineState &State) { | |||
| 332 | const FormatToken &Current = *State.NextToken; | |||
| 333 | const FormatToken &Previous = *Current.Previous; | |||
| 334 | if (Style.BraceWrapping.BeforeLambdaBody && Current.CanBreakBefore && | |||
| 335 | Current.is(TT_LambdaLBrace) && Previous.isNot(TT_LineComment)) { | |||
| 336 | auto LambdaBodyLength = getLengthToMatchingParen(Current, State.Stack); | |||
| 337 | return (LambdaBodyLength > getColumnLimit(State)); | |||
| 338 | } | |||
| 339 | if (Current.MustBreakBefore || Current.is(TT_InlineASMColon)) | |||
| 340 | return true; | |||
| 341 | if (State.Stack.back().BreakBeforeClosingBrace && | |||
| 342 | Current.closesBlockOrBlockTypeList(Style)) | |||
| 343 | return true; | |||
| 344 | if (State.Stack.back().BreakBeforeClosingParen && Current.is(tok::r_paren)) | |||
| 345 | return true; | |||
| 346 | if (Previous.is(tok::semi) && State.LineContainsContinuedForLoopSection) | |||
| 347 | return true; | |||
| 348 | if (Style.Language == FormatStyle::LK_ObjC && | |||
| 349 | Style.ObjCBreakBeforeNestedBlockParam && | |||
| 350 | Current.ObjCSelectorNameParts > 1 && | |||
| 351 | Current.startsSequence(TT_SelectorName, tok::colon, tok::caret)) { | |||
| 352 | return true; | |||
| 353 | } | |||
| 354 | // Avoid producing inconsistent states by requiring breaks where they are not | |||
| 355 | // permitted for C# generic type constraints. | |||
| 356 | if (State.Stack.back().IsCSharpGenericTypeConstraint && | |||
| 357 | Previous.isNot(TT_CSharpGenericTypeConstraintComma)) | |||
| 358 | return false; | |||
| 359 | if ((startsNextParameter(Current, Style) || Previous.is(tok::semi) || | |||
| 360 | (Previous.is(TT_TemplateCloser) && Current.is(TT_StartOfName) && | |||
| 361 | Style.isCpp() && | |||
| 362 | // FIXME: This is a temporary workaround for the case where clang-format | |||
| 363 | // sets BreakBeforeParameter to avoid bin packing and this creates a | |||
| 364 | // completely unnecessary line break after a template type that isn't | |||
| 365 | // line-wrapped. | |||
| 366 | (Previous.NestingLevel == 1 || Style.BinPackParameters)) || | |||
| 367 | (Style.BreakBeforeTernaryOperators && Current.is(TT_ConditionalExpr) && | |||
| 368 | Previous.isNot(tok::question)) || | |||
| 369 | (!Style.BreakBeforeTernaryOperators && | |||
| 370 | Previous.is(TT_ConditionalExpr))) && | |||
| 371 | State.Stack.back().BreakBeforeParameter && !Current.isTrailingComment() && | |||
| 372 | !Current.isOneOf(tok::r_paren, tok::r_brace)) | |||
| 373 | return true; | |||
| 374 | if (State.Stack.back().IsChainedConditional && | |||
| 375 | ((Style.BreakBeforeTernaryOperators && Current.is(TT_ConditionalExpr) && | |||
| 376 | Current.is(tok::colon)) || | |||
| 377 | (!Style.BreakBeforeTernaryOperators && Previous.is(TT_ConditionalExpr) && | |||
| 378 | Previous.is(tok::colon)))) | |||
| 379 | return true; | |||
| 380 | if (((Previous.is(TT_DictLiteral) && Previous.is(tok::l_brace)) || | |||
| 381 | (Previous.is(TT_ArrayInitializerLSquare) && | |||
| 382 | Previous.ParameterCount > 1) || | |||
| 383 | opensProtoMessageField(Previous, Style)) && | |||
| 384 | Style.ColumnLimit > 0 && | |||
| 385 | getLengthToMatchingParen(Previous, State.Stack) + State.Column - 1 > | |||
| 386 | getColumnLimit(State)) | |||
| 387 | return true; | |||
| 388 | ||||
| 389 | const FormatToken &BreakConstructorInitializersToken = | |||
| 390 | Style.BreakConstructorInitializers == FormatStyle::BCIS_AfterColon | |||
| 391 | ? Previous | |||
| 392 | : Current; | |||
| 393 | if (BreakConstructorInitializersToken.is(TT_CtorInitializerColon) && | |||
| 394 | (State.Column + State.Line->Last->TotalLength - Previous.TotalLength > | |||
| 395 | getColumnLimit(State) || | |||
| 396 | State.Stack.back().BreakBeforeParameter) && | |||
| 397 | (Style.AllowShortFunctionsOnASingleLine != FormatStyle::SFS_All || | |||
| 398 | Style.BreakConstructorInitializers != FormatStyle::BCIS_BeforeColon || | |||
| 399 | Style.ColumnLimit != 0)) | |||
| 400 | return true; | |||
| 401 | ||||
| 402 | if (Current.is(TT_ObjCMethodExpr) && !Previous.is(TT_SelectorName) && | |||
| 403 | State.Line->startsWith(TT_ObjCMethodSpecifier)) | |||
| 404 | return true; | |||
| 405 | if (Current.is(TT_SelectorName) && !Previous.is(tok::at) && | |||
| 406 | State.Stack.back().ObjCSelectorNameFound && | |||
| 407 | State.Stack.back().BreakBeforeParameter && | |||
| 408 | (Style.ObjCBreakBeforeNestedBlockParam || | |||
| 409 | !Current.startsSequence(TT_SelectorName, tok::colon, tok::caret))) | |||
| 410 | return true; | |||
| 411 | ||||
| 412 | unsigned NewLineColumn = getNewLineColumn(State); | |||
| 413 | if (Current.isMemberAccess() && Style.ColumnLimit != 0 && | |||
| 414 | State.Column + getLengthToNextOperator(Current) > Style.ColumnLimit && | |||
| 415 | (State.Column > NewLineColumn || | |||
| 416 | Current.NestingLevel < State.StartOfLineLevel)) | |||
| 417 | return true; | |||
| 418 | ||||
| 419 | if (startsSegmentOfBuilderTypeCall(Current) && | |||
| 420 | (State.Stack.back().CallContinuation != 0 || | |||
| 421 | State.Stack.back().BreakBeforeParameter) && | |||
| 422 | // JavaScript is treated different here as there is a frequent pattern: | |||
| 423 | // SomeFunction(function() { | |||
| 424 | // ... | |||
| 425 | // }.bind(...)); | |||
| 426 | // FIXME: We should find a more generic solution to this problem. | |||
| 427 | !(State.Column <= NewLineColumn && Style.isJavaScript()) && | |||
| 428 | !(Previous.closesScopeAfterBlock() && State.Column <= NewLineColumn)) | |||
| 429 | return true; | |||
| 430 | ||||
| 431 | // If the template declaration spans multiple lines, force wrap before the | |||
| 432 | // function/class declaration | |||
| 433 | if (Previous.ClosesTemplateDeclaration && | |||
| 434 | State.Stack.back().BreakBeforeParameter && Current.CanBreakBefore) | |||
| 435 | return true; | |||
| 436 | ||||
| 437 | if (!State.Line->First->is(tok::kw_enum) && State.Column <= NewLineColumn) | |||
| 438 | return false; | |||
| 439 | ||||
| 440 | if (Style.AlwaysBreakBeforeMultilineStrings && | |||
| 441 | (NewLineColumn == State.FirstIndent + Style.ContinuationIndentWidth || | |||
| 442 | Previous.is(tok::comma) || Current.NestingLevel < 2) && | |||
| 443 | !Previous.isOneOf(tok::kw_return, tok::lessless, tok::at, | |||
| 444 | Keywords.kw_dollar) && | |||
| 445 | !Previous.isOneOf(TT_InlineASMColon, TT_ConditionalExpr) && | |||
| 446 | nextIsMultilineString(State)) | |||
| 447 | return true; | |||
| 448 | ||||
| 449 | // Using CanBreakBefore here and below takes care of the decision whether the | |||
| 450 | // current style uses wrapping before or after operators for the given | |||
| 451 | // operator. | |||
| 452 | if (Previous.is(TT_BinaryOperator) && Current.CanBreakBefore) { | |||
| 453 | // If we need to break somewhere inside the LHS of a binary expression, we | |||
| 454 | // should also break after the operator. Otherwise, the formatting would | |||
| 455 | // hide the operator precedence, e.g. in: | |||
| 456 | // if (aaaaaaaaaaaaaa == | |||
| 457 | // bbbbbbbbbbbbbb && c) {.. | |||
| 458 | // For comparisons, we only apply this rule, if the LHS is a binary | |||
| 459 | // expression itself as otherwise, the line breaks seem superfluous. | |||
| 460 | // We need special cases for ">>" which we have split into two ">" while | |||
| 461 | // lexing in order to make template parsing easier. | |||
| 462 | bool IsComparison = (Previous.getPrecedence() == prec::Relational || | |||
| 463 | Previous.getPrecedence() == prec::Equality || | |||
| 464 | Previous.getPrecedence() == prec::Spaceship) && | |||
| 465 | Previous.Previous && | |||
| 466 | Previous.Previous->isNot(TT_BinaryOperator); // For >>. | |||
| 467 | bool LHSIsBinaryExpr = | |||
| 468 | Previous.Previous && Previous.Previous->EndsBinaryExpression; | |||
| 469 | if ((!IsComparison || LHSIsBinaryExpr) && !Current.isTrailingComment() && | |||
| 470 | Previous.getPrecedence() != prec::Assignment && | |||
| 471 | State.Stack.back().BreakBeforeParameter) | |||
| 472 | return true; | |||
| 473 | } else if (Current.is(TT_BinaryOperator) && Current.CanBreakBefore && | |||
| 474 | State.Stack.back().BreakBeforeParameter) { | |||
| 475 | return true; | |||
| 476 | } | |||
| 477 | ||||
| 478 | // Same as above, but for the first "<<" operator. | |||
| 479 | if (Current.is(tok::lessless) && Current.isNot(TT_OverloadedOperator) && | |||
| 480 | State.Stack.back().BreakBeforeParameter && | |||
| 481 | State.Stack.back().FirstLessLess == 0) | |||
| 482 | return true; | |||
| 483 | ||||
| 484 | if (Current.NestingLevel == 0 && !Current.isTrailingComment()) { | |||
| 485 | // Always break after "template <...>" and leading annotations. This is only | |||
| 486 | // for cases where the entire line does not fit on a single line as a | |||
| 487 | // different LineFormatter would be used otherwise. | |||
| 488 | if (Previous.ClosesTemplateDeclaration) | |||
| 489 | return Style.AlwaysBreakTemplateDeclarations != FormatStyle::BTDS_No; | |||
| 490 | if (Previous.is(TT_FunctionAnnotationRParen) && | |||
| 491 | State.Line->Type != LT_PreprocessorDirective) | |||
| 492 | return true; | |||
| 493 | if (Previous.is(TT_LeadingJavaAnnotation) && Current.isNot(tok::l_paren) && | |||
| 494 | Current.isNot(TT_LeadingJavaAnnotation)) | |||
| 495 | return true; | |||
| 496 | } | |||
| 497 | ||||
| 498 | if (Style.isJavaScript() && Previous.is(tok::r_paren) && | |||
| 499 | Previous.is(TT_JavaAnnotation)) { | |||
| 500 | // Break after the closing parenthesis of TypeScript decorators before | |||
| 501 | // functions, getters and setters. | |||
| 502 | static const llvm::StringSet<> BreakBeforeDecoratedTokens = {"get", "set", | |||
| 503 | "function"}; | |||
| 504 | if (BreakBeforeDecoratedTokens.contains(Current.TokenText)) | |||
| 505 | return true; | |||
| 506 | } | |||
| 507 | ||||
| 508 | // If the return type spans multiple lines, wrap before the function name. | |||
| 509 | if (((Current.is(TT_FunctionDeclarationName) && | |||
| 510 | // Don't break before a C# function when no break after return type | |||
| 511 | (!Style.isCSharp() || | |||
| 512 | Style.AlwaysBreakAfterReturnType != FormatStyle::RTBS_None) && | |||
| 513 | // Don't always break between a JavaScript `function` and the function | |||
| 514 | // name. | |||
| 515 | !Style.isJavaScript()) || | |||
| 516 | (Current.is(tok::kw_operator) && !Previous.is(tok::coloncolon))) && | |||
| 517 | !Previous.is(tok::kw_template) && State.Stack.back().BreakBeforeParameter) | |||
| 518 | return true; | |||
| 519 | ||||
| 520 | // The following could be precomputed as they do not depend on the state. | |||
| 521 | // However, as they should take effect only if the UnwrappedLine does not fit | |||
| 522 | // into the ColumnLimit, they are checked here in the ContinuationIndenter. | |||
| 523 | if (Style.ColumnLimit != 0 && Previous.is(BK_Block) && | |||
| 524 | Previous.is(tok::l_brace) && !Current.isOneOf(tok::r_brace, tok::comment)) | |||
| 525 | return true; | |||
| 526 | ||||
| 527 | if (Current.is(tok::lessless) && | |||
| 528 | ((Previous.is(tok::identifier) && Previous.TokenText == "endl") || | |||
| 529 | (Previous.Tok.isLiteral() && (Previous.TokenText.endswith("\\n\"") || | |||
| 530 | Previous.TokenText == "\'\\n\'")))) | |||
| 531 | return true; | |||
| 532 | ||||
| 533 | if (Previous.is(TT_BlockComment) && Previous.IsMultiline) | |||
| 534 | return true; | |||
| 535 | ||||
| 536 | if (State.NoContinuation) | |||
| 537 | return true; | |||
| 538 | ||||
| 539 | return false; | |||
| 540 | } | |||
| 541 | ||||
| 542 | unsigned ContinuationIndenter::addTokenToState(LineState &State, bool Newline, | |||
| 543 | bool DryRun, | |||
| 544 | unsigned ExtraSpaces) { | |||
| 545 | const FormatToken &Current = *State.NextToken; | |||
| 546 | ||||
| 547 | assert(!State.Stack.empty())(static_cast <bool> (!State.Stack.empty()) ? void (0) : __assert_fail ("!State.Stack.empty()", "clang/lib/Format/ContinuationIndenter.cpp" , 547, __extension__ __PRETTY_FUNCTION__)); | |||
| 548 | State.NoContinuation = false; | |||
| 549 | ||||
| 550 | if ((Current.is(TT_ImplicitStringLiteral) && | |||
| 551 | (Current.Previous->Tok.getIdentifierInfo() == nullptr || | |||
| 552 | Current.Previous->Tok.getIdentifierInfo()->getPPKeywordID() == | |||
| 553 | tok::pp_not_keyword))) { | |||
| 554 | unsigned EndColumn = | |||
| 555 | SourceMgr.getSpellingColumnNumber(Current.WhitespaceRange.getEnd()); | |||
| 556 | if (Current.LastNewlineOffset != 0) { | |||
| 557 | // If there is a newline within this token, the final column will solely | |||
| 558 | // determined by the current end column. | |||
| 559 | State.Column = EndColumn; | |||
| 560 | } else { | |||
| 561 | unsigned StartColumn = | |||
| 562 | SourceMgr.getSpellingColumnNumber(Current.WhitespaceRange.getBegin()); | |||
| 563 | assert(EndColumn >= StartColumn)(static_cast <bool> (EndColumn >= StartColumn) ? void (0) : __assert_fail ("EndColumn >= StartColumn", "clang/lib/Format/ContinuationIndenter.cpp" , 563, __extension__ __PRETTY_FUNCTION__)); | |||
| 564 | State.Column += EndColumn - StartColumn; | |||
| 565 | } | |||
| 566 | moveStateToNextToken(State, DryRun, /*Newline=*/false); | |||
| 567 | return 0; | |||
| 568 | } | |||
| 569 | ||||
| 570 | unsigned Penalty = 0; | |||
| 571 | if (Newline) | |||
| 572 | Penalty = addTokenOnNewLine(State, DryRun); | |||
| 573 | else | |||
| 574 | addTokenOnCurrentLine(State, DryRun, ExtraSpaces); | |||
| 575 | ||||
| 576 | return moveStateToNextToken(State, DryRun, Newline) + Penalty; | |||
| 577 | } | |||
| 578 | ||||
| 579 | void ContinuationIndenter::addTokenOnCurrentLine(LineState &State, bool DryRun, | |||
| 580 | unsigned ExtraSpaces) { | |||
| 581 | FormatToken &Current = *State.NextToken; | |||
| 582 | const FormatToken &Previous = *State.NextToken->Previous; | |||
| ||||
| 583 | if (Current.is(tok::equal) && | |||
| 584 | (State.Line->First->is(tok::kw_for) || Current.NestingLevel == 0) && | |||
| 585 | State.Stack.back().VariablePos == 0) { | |||
| 586 | State.Stack.back().VariablePos = State.Column; | |||
| 587 | // Move over * and & if they are bound to the variable name. | |||
| 588 | const FormatToken *Tok = &Previous; | |||
| 589 | while (Tok && State.Stack.back().VariablePos >= Tok->ColumnWidth) { | |||
| 590 | State.Stack.back().VariablePos -= Tok->ColumnWidth; | |||
| 591 | if (Tok->SpacesRequiredBefore != 0) | |||
| 592 | break; | |||
| 593 | Tok = Tok->Previous; | |||
| 594 | } | |||
| 595 | if (Previous.PartOfMultiVariableDeclStmt) | |||
| ||||
| 596 | State.Stack.back().LastSpace = State.Stack.back().VariablePos; | |||
| 597 | } | |||
| 598 | ||||
| 599 | unsigned Spaces = Current.SpacesRequiredBefore + ExtraSpaces; | |||
| 600 | ||||
| 601 | // Indent preprocessor directives after the hash if required. | |||
| 602 | int PPColumnCorrection = 0; | |||
| 603 | if (Style.IndentPPDirectives == FormatStyle::PPDIS_AfterHash && | |||
| 604 | Previous.is(tok::hash) && State.FirstIndent > 0 && | |||
| 605 | (State.Line->Type == LT_PreprocessorDirective || | |||
| 606 | State.Line->Type == LT_ImportStatement)) { | |||
| 607 | Spaces += State.FirstIndent; | |||
| 608 | ||||
| 609 | // For preprocessor indent with tabs, State.Column will be 1 because of the | |||
| 610 | // hash. This causes second-level indents onward to have an extra space | |||
| 611 | // after the tabs. We avoid this misalignment by subtracting 1 from the | |||
| 612 | // column value passed to replaceWhitespace(). | |||
| 613 | if (Style.UseTab != FormatStyle::UT_Never) | |||
| 614 | PPColumnCorrection = -1; | |||
| 615 | } | |||
| 616 | ||||
| 617 | if (!DryRun) | |||
| 618 | Whitespaces.replaceWhitespace(Current, /*Newlines=*/0, Spaces, | |||
| 619 | State.Column + Spaces + PPColumnCorrection); | |||
| 620 | ||||
| 621 | // If "BreakBeforeInheritanceComma" mode, don't break within the inheritance | |||
| 622 | // declaration unless there is multiple inheritance. | |||
| 623 | if (Style.BreakInheritanceList == FormatStyle::BILS_BeforeComma && | |||
| 624 | Current.is(TT_InheritanceColon)) | |||
| 625 | State.Stack.back().NoLineBreak = true; | |||
| 626 | if (Style.BreakInheritanceList == FormatStyle::BILS_AfterColon && | |||
| 627 | Previous.is(TT_InheritanceColon)) | |||
| 628 | State.Stack.back().NoLineBreak = true; | |||
| 629 | ||||
| 630 | if (Current.is(TT_SelectorName) && | |||
| 631 | !State.Stack.back().ObjCSelectorNameFound) { | |||
| 632 | unsigned MinIndent = | |||
| 633 | std::max(State.FirstIndent + Style.ContinuationIndentWidth, | |||
| 634 | State.Stack.back().Indent); | |||
| 635 | unsigned FirstColonPos = State.Column + Spaces + Current.ColumnWidth; | |||
| 636 | if (Current.LongestObjCSelectorName == 0) | |||
| 637 | State.Stack.back().AlignColons = false; | |||
| 638 | else if (MinIndent + Current.LongestObjCSelectorName > FirstColonPos) | |||
| 639 | State.Stack.back().ColonPos = MinIndent + Current.LongestObjCSelectorName; | |||
| 640 | else | |||
| 641 | State.Stack.back().ColonPos = FirstColonPos; | |||
| 642 | } | |||
| 643 | ||||
| 644 | // In "AlwaysBreak" or "BlockIndent" mode, enforce wrapping directly after the | |||
| 645 | // parenthesis by disallowing any further line breaks if there is no line | |||
| 646 | // break after the opening parenthesis. Don't break if it doesn't conserve | |||
| 647 | // columns. | |||
| 648 | if ((Style.AlignAfterOpenBracket == FormatStyle::BAS_AlwaysBreak || | |||
| 649 | Style.AlignAfterOpenBracket == FormatStyle::BAS_BlockIndent) && | |||
| 650 | (Previous.isOneOf(tok::l_paren, TT_TemplateOpener, tok::l_square) || | |||
| 651 | (Previous.is(tok::l_brace) && Previous.isNot(BK_Block) && | |||
| 652 | Style.Cpp11BracedListStyle)) && | |||
| 653 | State.Column > getNewLineColumn(State) && | |||
| 654 | (!Previous.Previous || !Previous.Previous->isOneOf( | |||
| 655 | tok::kw_for, tok::kw_while, tok::kw_switch)) && | |||
| 656 | // Don't do this for simple (no expressions) one-argument function calls | |||
| 657 | // as that feels like needlessly wasting whitespace, e.g.: | |||
| 658 | // | |||
| 659 | // caaaaaaaaaaaall( | |||
| 660 | // caaaaaaaaaaaall( | |||
| 661 | // caaaaaaaaaaaall( | |||
| 662 | // caaaaaaaaaaaaaaaaaaaaaaall(aaaaaaaaaaaaaa, aaaaaaaaa)))); | |||
| 663 | Current.FakeLParens.size() > 0 && | |||
| 664 | Current.FakeLParens.back() > prec::Unknown) | |||
| 665 | State.Stack.back().NoLineBreak = true; | |||
| 666 | if (Previous.is(TT_TemplateString) && Previous.opensScope()) | |||
| 667 | State.Stack.back().NoLineBreak = true; | |||
| 668 | ||||
| 669 | if (Style.AlignAfterOpenBracket != FormatStyle::BAS_DontAlign && | |||
| 670 | !State.Stack.back().IsCSharpGenericTypeConstraint && | |||
| 671 | Previous.opensScope() && Previous.isNot(TT_ObjCMethodExpr) && | |||
| 672 | (Current.isNot(TT_LineComment) || Previous.is(BK_BracedInit))) { | |||
| 673 | State.Stack.back().Indent = State.Column + Spaces; | |||
| 674 | State.Stack.back().IsAligned = true; | |||
| 675 | } | |||
| 676 | if (State.Stack.back().AvoidBinPacking && startsNextParameter(Current, Style)) | |||
| 677 | State.Stack.back().NoLineBreak = true; | |||
| 678 | if (startsSegmentOfBuilderTypeCall(Current) && | |||
| 679 | State.Column > getNewLineColumn(State)) | |||
| 680 | State.Stack.back().ContainsUnwrappedBuilder = true; | |||
| 681 | ||||
| 682 | if (Current.is(TT_LambdaArrow) && Style.Language == FormatStyle::LK_Java) | |||
| 683 | State.Stack.back().NoLineBreak = true; | |||
| 684 | if (Current.isMemberAccess() && Previous.is(tok::r_paren) && | |||
| 685 | (Previous.MatchingParen && | |||
| 686 | (Previous.TotalLength - Previous.MatchingParen->TotalLength > 10))) | |||
| 687 | // If there is a function call with long parameters, break before trailing | |||
| 688 | // calls. This prevents things like: | |||
| 689 | // EXPECT_CALL(SomeLongParameter).Times( | |||
| 690 | // 2); | |||
| 691 | // We don't want to do this for short parameters as they can just be | |||
| 692 | // indexes. | |||
| 693 | State.Stack.back().NoLineBreak = true; | |||
| 694 | ||||
| 695 | // Don't allow the RHS of an operator to be split over multiple lines unless | |||
| 696 | // there is a line-break right after the operator. | |||
| 697 | // Exclude relational operators, as there, it is always more desirable to | |||
| 698 | // have the LHS 'left' of the RHS. | |||
| 699 | const FormatToken *P = Current.getPreviousNonComment(); | |||
| 700 | if (!Current.is(tok::comment) && P && | |||
| 701 | (P->isOneOf(TT_BinaryOperator, tok::comma) || | |||
| 702 | (P->is(TT_ConditionalExpr) && P->is(tok::colon))) && | |||
| 703 | !P->isOneOf(TT_OverloadedOperator, TT_CtorInitializerComma) && | |||
| 704 | P->getPrecedence() != prec::Assignment && | |||
| 705 | P->getPrecedence() != prec::Relational && | |||
| 706 | P->getPrecedence() != prec::Spaceship) { | |||
| 707 | bool BreakBeforeOperator = | |||
| 708 | P->MustBreakBefore || P->is(tok::lessless) || | |||
| 709 | (P->is(TT_BinaryOperator) && | |||
| 710 | Style.BreakBeforeBinaryOperators != FormatStyle::BOS_None) || | |||
| 711 | (P->is(TT_ConditionalExpr) && Style.BreakBeforeTernaryOperators); | |||
| 712 | // Don't do this if there are only two operands. In these cases, there is | |||
| 713 | // always a nice vertical separation between them and the extra line break | |||
| 714 | // does not help. | |||
| 715 | bool HasTwoOperands = | |||
| 716 | P->OperatorIndex == 0 && !P->NextOperator && !P->is(TT_ConditionalExpr); | |||
| 717 | if ((!BreakBeforeOperator && | |||
| 718 | !(HasTwoOperands && | |||
| 719 | Style.AlignOperands != FormatStyle::OAS_DontAlign)) || | |||
| 720 | (!State.Stack.back().LastOperatorWrapped && BreakBeforeOperator)) | |||
| 721 | State.Stack.back().NoLineBreakInOperand = true; | |||
| 722 | } | |||
| 723 | ||||
| 724 | State.Column += Spaces; | |||
| 725 | if (Current.isNot(tok::comment) && Previous.is(tok::l_paren) && | |||
| 726 | Previous.Previous && | |||
| 727 | (Previous.Previous->is(tok::kw_for) || Previous.Previous->isIf())) { | |||
| 728 | // Treat the condition inside an if as if it was a second function | |||
| 729 | // parameter, i.e. let nested calls have a continuation indent. | |||
| 730 | State.Stack.back().LastSpace = State.Column; | |||
| 731 | State.Stack.back().NestedBlockIndent = State.Column; | |||
| 732 | } else if (!Current.isOneOf(tok::comment, tok::caret) && | |||
| 733 | ((Previous.is(tok::comma) && | |||
| 734 | !Previous.is(TT_OverloadedOperator)) || | |||
| 735 | (Previous.is(tok::colon) && Previous.is(TT_ObjCMethodExpr)))) { | |||
| 736 | State.Stack.back().LastSpace = State.Column; | |||
| 737 | } else if (Previous.is(TT_CtorInitializerColon) && | |||
| 738 | Style.BreakConstructorInitializers == | |||
| 739 | FormatStyle::BCIS_AfterColon) { | |||
| 740 | State.Stack.back().Indent = State.Column; | |||
| 741 | State.Stack.back().LastSpace = State.Column; | |||
| 742 | } else if ((Previous.isOneOf(TT_BinaryOperator, TT_ConditionalExpr, | |||
| 743 | TT_CtorInitializerColon)) && | |||
| 744 | ((Previous.getPrecedence() != prec::Assignment && | |||
| 745 | (Previous.isNot(tok::lessless) || Previous.OperatorIndex != 0 || | |||
| 746 | Previous.NextOperator)) || | |||
| 747 | Current.StartsBinaryExpression)) { | |||
| 748 | // Indent relative to the RHS of the expression unless this is a simple | |||
| 749 | // assignment without binary expression on the RHS. Also indent relative to | |||
| 750 | // unary operators and the colons of constructor initializers. | |||
| 751 | if (Style.BreakBeforeBinaryOperators == FormatStyle::BOS_None) | |||
| 752 | State.Stack.back().LastSpace = State.Column; | |||
| 753 | } else if (Previous.is(TT_InheritanceColon)) { | |||
| 754 | State.Stack.back().Indent = State.Column; | |||
| 755 | State.Stack.back().LastSpace = State.Column; | |||
| 756 | } else if (Current.is(TT_CSharpGenericTypeConstraintColon)) { | |||
| 757 | State.Stack.back().ColonPos = State.Column; | |||
| 758 | } else if (Previous.opensScope()) { | |||
| 759 | // If a function has a trailing call, indent all parameters from the | |||
| 760 | // opening parenthesis. This avoids confusing indents like: | |||
| 761 | // OuterFunction(InnerFunctionCall( // break | |||
| 762 | // ParameterToInnerFunction)) // break | |||
| 763 | // .SecondInnerFunctionCall(); | |||
| 764 | bool HasTrailingCall = false; | |||
| 765 | if (Previous.MatchingParen) { | |||
| 766 | const FormatToken *Next = Previous.MatchingParen->getNextNonComment(); | |||
| 767 | HasTrailingCall = Next && Next->isMemberAccess(); | |||
| 768 | } | |||
| 769 | if (HasTrailingCall && State.Stack.size() > 1 && | |||
| 770 | State.Stack[State.Stack.size() - 2].CallContinuation == 0) | |||
| 771 | State.Stack.back().LastSpace = State.Column; | |||
| 772 | } | |||
| 773 | } | |||
| 774 | ||||
| 775 | unsigned ContinuationIndenter::addTokenOnNewLine(LineState &State, | |||
| 776 | bool DryRun) { | |||
| 777 | FormatToken &Current = *State.NextToken; | |||
| 778 | const FormatToken &Previous = *State.NextToken->Previous; | |||
| 779 | ||||
| 780 | // Extra penalty that needs to be added because of the way certain line | |||
| 781 | // breaks are chosen. | |||
| 782 | unsigned Penalty = 0; | |||
| 783 | ||||
| 784 | const FormatToken *PreviousNonComment = Current.getPreviousNonComment(); | |||
| 785 | const FormatToken *NextNonComment = Previous.getNextNonComment(); | |||
| 786 | if (!NextNonComment) | |||
| 787 | NextNonComment = &Current; | |||
| 788 | // The first line break on any NestingLevel causes an extra penalty in order | |||
| 789 | // prefer similar line breaks. | |||
| 790 | if (!State.Stack.back().ContainsLineBreak) | |||
| 791 | Penalty += 15; | |||
| 792 | State.Stack.back().ContainsLineBreak = true; | |||
| 793 | ||||
| 794 | Penalty += State.NextToken->SplitPenalty; | |||
| 795 | ||||
| 796 | // Breaking before the first "<<" is generally not desirable if the LHS is | |||
| 797 | // short. Also always add the penalty if the LHS is split over multiple lines | |||
| 798 | // to avoid unnecessary line breaks that just work around this penalty. | |||
| 799 | if (NextNonComment->is(tok::lessless) && | |||
| 800 | State.Stack.back().FirstLessLess == 0 && | |||
| 801 | (State.Column <= Style.ColumnLimit / 3 || | |||
| 802 | State.Stack.back().BreakBeforeParameter)) | |||
| 803 | Penalty += Style.PenaltyBreakFirstLessLess; | |||
| 804 | ||||
| 805 | State.Column = getNewLineColumn(State); | |||
| 806 | ||||
| 807 | // Add Penalty proportional to amount of whitespace away from FirstColumn | |||
| 808 | // This tends to penalize several lines that are far-right indented, | |||
| 809 | // and prefers a line-break prior to such a block, e.g: | |||
| 810 | // | |||
| 811 | // Constructor() : | |||
| 812 | // member(value), looooooooooooooooong_member( | |||
| 813 | // looooooooooong_call(param_1, param_2, param_3)) | |||
| 814 | // would then become | |||
| 815 | // Constructor() : | |||
| 816 | // member(value), | |||
| 817 | // looooooooooooooooong_member( | |||
| 818 | // looooooooooong_call(param_1, param_2, param_3)) | |||
| 819 | if (State.Column > State.FirstIndent) | |||
| 820 | Penalty += | |||
| 821 | Style.PenaltyIndentedWhitespace * (State.Column - State.FirstIndent); | |||
| 822 | ||||
| 823 | // Indent nested blocks relative to this column, unless in a very specific | |||
| 824 | // JavaScript special case where: | |||
| 825 | // | |||
| 826 | // var loooooong_name = | |||
| 827 | // function() { | |||
| 828 | // // code | |||
| 829 | // } | |||
| 830 | // | |||
| 831 | // is common and should be formatted like a free-standing function. The same | |||
| 832 | // goes for wrapping before the lambda return type arrow. | |||
| 833 | if (!Current.is(TT_LambdaArrow) && | |||
| 834 | (!Style.isJavaScript() || Current.NestingLevel != 0 || | |||
| 835 | !PreviousNonComment || !PreviousNonComment->is(tok::equal) || | |||
| 836 | !Current.isOneOf(Keywords.kw_async, Keywords.kw_function))) | |||
| 837 | State.Stack.back().NestedBlockIndent = State.Column; | |||
| 838 | ||||
| 839 | if (NextNonComment->isMemberAccess()) { | |||
| 840 | if (State.Stack.back().CallContinuation == 0) | |||
| 841 | State.Stack.back().CallContinuation = State.Column; | |||
| 842 | } else if (NextNonComment->is(TT_SelectorName)) { | |||
| 843 | if (!State.Stack.back().ObjCSelectorNameFound) { | |||
| 844 | if (NextNonComment->LongestObjCSelectorName == 0) { | |||
| 845 | State.Stack.back().AlignColons = false; | |||
| 846 | } else { | |||
| 847 | State.Stack.back().ColonPos = | |||
| 848 | (shouldIndentWrappedSelectorName(Style, State.Line->Type) | |||
| 849 | ? std::max(State.Stack.back().Indent, | |||
| 850 | State.FirstIndent + Style.ContinuationIndentWidth) | |||
| 851 | : State.Stack.back().Indent) + | |||
| 852 | std::max(NextNonComment->LongestObjCSelectorName, | |||
| 853 | NextNonComment->ColumnWidth); | |||
| 854 | } | |||
| 855 | } else if (State.Stack.back().AlignColons && | |||
| 856 | State.Stack.back().ColonPos <= NextNonComment->ColumnWidth) { | |||
| 857 | State.Stack.back().ColonPos = State.Column + NextNonComment->ColumnWidth; | |||
| 858 | } | |||
| 859 | } else if (PreviousNonComment && PreviousNonComment->is(tok::colon) && | |||
| 860 | PreviousNonComment->isOneOf(TT_ObjCMethodExpr, TT_DictLiteral)) { | |||
| 861 | // FIXME: This is hacky, find a better way. The problem is that in an ObjC | |||
| 862 | // method expression, the block should be aligned to the line starting it, | |||
| 863 | // e.g.: | |||
| 864 | // [aaaaaaaaaaaaaaa aaaaaaaaa: \\ break for some reason | |||
| 865 | // ^(int *i) { | |||
| 866 | // // ... | |||
| 867 | // }]; | |||
| 868 | // Thus, we set LastSpace of the next higher NestingLevel, to which we move | |||
| 869 | // when we consume all of the "}"'s FakeRParens at the "{". | |||
| 870 | if (State.Stack.size() > 1) | |||
| 871 | State.Stack[State.Stack.size() - 2].LastSpace = | |||
| 872 | std::max(State.Stack.back().LastSpace, State.Stack.back().Indent) + | |||
| 873 | Style.ContinuationIndentWidth; | |||
| 874 | } | |||
| 875 | ||||
| 876 | if ((PreviousNonComment && | |||
| 877 | PreviousNonComment->isOneOf(tok::comma, tok::semi) && | |||
| 878 | !State.Stack.back().AvoidBinPacking) || | |||
| 879 | Previous.is(TT_BinaryOperator)) | |||
| 880 | State.Stack.back().BreakBeforeParameter = false; | |||
| 881 | if (PreviousNonComment && | |||
| 882 | PreviousNonComment->isOneOf(TT_TemplateCloser, TT_JavaAnnotation) && | |||
| 883 | Current.NestingLevel == 0) | |||
| 884 | State.Stack.back().BreakBeforeParameter = false; | |||
| 885 | if (NextNonComment->is(tok::question) || | |||
| 886 | (PreviousNonComment && PreviousNonComment->is(tok::question))) | |||
| 887 | State.Stack.back().BreakBeforeParameter = true; | |||
| 888 | if (Current.is(TT_BinaryOperator) && Current.CanBreakBefore) | |||
| 889 | State.Stack.back().BreakBeforeParameter = false; | |||
| 890 | ||||
| 891 | if (!DryRun) { | |||
| 892 | unsigned MaxEmptyLinesToKeep = Style.MaxEmptyLinesToKeep + 1; | |||
| 893 | if (Current.is(tok::r_brace) && Current.MatchingParen && | |||
| 894 | // Only strip trailing empty lines for l_braces that have children, i.e. | |||
| 895 | // for function expressions (lambdas, arrows, etc). | |||
| 896 | !Current.MatchingParen->Children.empty()) { | |||
| 897 | // lambdas and arrow functions are expressions, thus their r_brace is not | |||
| 898 | // on its own line, and thus not covered by UnwrappedLineFormatter's logic | |||
| 899 | // about removing empty lines on closing blocks. Special case them here. | |||
| 900 | MaxEmptyLinesToKeep = 1; | |||
| 901 | } | |||
| 902 | unsigned Newlines = | |||
| 903 | std::max(1u, std::min(Current.NewlinesBefore, MaxEmptyLinesToKeep)); | |||
| 904 | bool ContinuePPDirective = | |||
| 905 | State.Line->InPPDirective && State.Line->Type != LT_ImportStatement; | |||
| 906 | Whitespaces.replaceWhitespace(Current, Newlines, State.Column, State.Column, | |||
| 907 | State.Stack.back().IsAligned, | |||
| 908 | ContinuePPDirective); | |||
| 909 | } | |||
| 910 | ||||
| 911 | if (!Current.isTrailingComment()) | |||
| 912 | State.Stack.back().LastSpace = State.Column; | |||
| 913 | if (Current.is(tok::lessless)) | |||
| 914 | // If we are breaking before a "<<", we always want to indent relative to | |||
| 915 | // RHS. This is necessary only for "<<", as we special-case it and don't | |||
| 916 | // always indent relative to the RHS. | |||
| 917 | State.Stack.back().LastSpace += 3; // 3 -> width of "<< ". | |||
| 918 | ||||
| 919 | State.StartOfLineLevel = Current.NestingLevel; | |||
| 920 | State.LowestLevelOnLine = Current.NestingLevel; | |||
| 921 | ||||
| 922 | // Any break on this level means that the parent level has been broken | |||
| 923 | // and we need to avoid bin packing there. | |||
| 924 | bool NestedBlockSpecialCase = | |||
| 925 | (!Style.isCpp() && Current.is(tok::r_brace) && State.Stack.size() > 1 && | |||
| 926 | State.Stack[State.Stack.size() - 2].NestedBlockInlined) || | |||
| 927 | (Style.Language == FormatStyle::LK_ObjC && Current.is(tok::r_brace) && | |||
| 928 | State.Stack.size() > 1 && !Style.ObjCBreakBeforeNestedBlockParam); | |||
| 929 | if (!NestedBlockSpecialCase) | |||
| 930 | for (unsigned i = 0, e = State.Stack.size() - 1; i != e; ++i) | |||
| 931 | State.Stack[i].BreakBeforeParameter = true; | |||
| 932 | ||||
| 933 | if (PreviousNonComment && | |||
| 934 | !PreviousNonComment->isOneOf(tok::comma, tok::colon, tok::semi) && | |||
| 935 | (PreviousNonComment->isNot(TT_TemplateCloser) || | |||
| 936 | Current.NestingLevel != 0) && | |||
| 937 | !PreviousNonComment->isOneOf( | |||
| 938 | TT_BinaryOperator, TT_FunctionAnnotationRParen, TT_JavaAnnotation, | |||
| 939 | TT_LeadingJavaAnnotation) && | |||
| 940 | Current.isNot(TT_BinaryOperator) && !PreviousNonComment->opensScope()) | |||
| 941 | State.Stack.back().BreakBeforeParameter = true; | |||
| 942 | ||||
| 943 | // If we break after { or the [ of an array initializer, we should also break | |||
| 944 | // before the corresponding } or ]. | |||
| 945 | if (PreviousNonComment && | |||
| 946 | (PreviousNonComment->isOneOf(tok::l_brace, TT_ArrayInitializerLSquare) || | |||
| 947 | opensProtoMessageField(*PreviousNonComment, Style))) | |||
| 948 | State.Stack.back().BreakBeforeClosingBrace = true; | |||
| 949 | ||||
| 950 | if (PreviousNonComment && PreviousNonComment->is(tok::l_paren)) | |||
| 951 | State.Stack.back().BreakBeforeClosingParen = | |||
| 952 | Style.AlignAfterOpenBracket == FormatStyle::BAS_BlockIndent; | |||
| 953 | ||||
| 954 | if (State.Stack.back().AvoidBinPacking) { | |||
| 955 | // If we are breaking after '(', '{', '<', or this is the break after a ':' | |||
| 956 | // to start a member initializater list in a constructor, this should not | |||
| 957 | // be considered bin packing unless the relevant AllowAll option is false or | |||
| 958 | // this is a dict/object literal. | |||
| 959 | bool PreviousIsBreakingCtorInitializerColon = | |||
| 960 | Previous.is(TT_CtorInitializerColon) && | |||
| 961 | Style.BreakConstructorInitializers == FormatStyle::BCIS_AfterColon; | |||
| 962 | if (!(Previous.isOneOf(tok::l_paren, tok::l_brace, TT_BinaryOperator) || | |||
| 963 | PreviousIsBreakingCtorInitializerColon) || | |||
| 964 | (!Style.AllowAllParametersOfDeclarationOnNextLine && | |||
| 965 | State.Line->MustBeDeclaration) || | |||
| 966 | (!Style.AllowAllArgumentsOnNextLine && | |||
| 967 | !State.Line->MustBeDeclaration) || | |||
| 968 | (Style.PackConstructorInitializers != FormatStyle::PCIS_NextLine && | |||
| 969 | PreviousIsBreakingCtorInitializerColon) || | |||
| 970 | Previous.is(TT_DictLiteral)) | |||
| 971 | State.Stack.back().BreakBeforeParameter = true; | |||
| 972 | ||||
| 973 | // If we are breaking after a ':' to start a member initializer list, | |||
| 974 | // and we allow all arguments on the next line, we should not break | |||
| 975 | // before the next parameter. | |||
| 976 | if (PreviousIsBreakingCtorInitializerColon && | |||
| 977 | Style.PackConstructorInitializers == FormatStyle::PCIS_NextLine) | |||
| 978 | State.Stack.back().BreakBeforeParameter = false; | |||
| 979 | } | |||
| 980 | ||||
| 981 | return Penalty; | |||
| 982 | } | |||
| 983 | ||||
| 984 | unsigned ContinuationIndenter::getNewLineColumn(const LineState &State) { | |||
| 985 | if (!State.NextToken || !State.NextToken->Previous) | |||
| 986 | return 0; | |||
| 987 | ||||
| 988 | FormatToken &Current = *State.NextToken; | |||
| 989 | ||||
| 990 | if (State.Stack.back().IsCSharpGenericTypeConstraint && | |||
| 991 | Current.isNot(TT_CSharpGenericTypeConstraint)) | |||
| 992 | return State.Stack.back().ColonPos + 2; | |||
| 993 | ||||
| 994 | const FormatToken &Previous = *Current.Previous; | |||
| 995 | // If we are continuing an expression, we want to use the continuation indent. | |||
| 996 | unsigned ContinuationIndent = | |||
| 997 | std::max(State.Stack.back().LastSpace, State.Stack.back().Indent) + | |||
| 998 | Style.ContinuationIndentWidth; | |||
| 999 | const FormatToken *PreviousNonComment = Current.getPreviousNonComment(); | |||
| 1000 | const FormatToken *NextNonComment = Previous.getNextNonComment(); | |||
| 1001 | if (!NextNonComment) | |||
| 1002 | NextNonComment = &Current; | |||
| 1003 | ||||
| 1004 | // Java specific bits. | |||
| 1005 | if (Style.Language == FormatStyle::LK_Java && | |||
| 1006 | Current.isOneOf(Keywords.kw_implements, Keywords.kw_extends)) | |||
| 1007 | return std::max(State.Stack.back().LastSpace, | |||
| 1008 | State.Stack.back().Indent + Style.ContinuationIndentWidth); | |||
| 1009 | ||||
| 1010 | if (Style.BreakBeforeBraces == FormatStyle::BS_Whitesmiths && | |||
| 1011 | State.Line->First->is(tok::kw_enum)) | |||
| 1012 | return (Style.IndentWidth * State.Line->First->IndentLevel) + | |||
| 1013 | Style.IndentWidth; | |||
| 1014 | ||||
| 1015 | if (NextNonComment->is(tok::l_brace) && NextNonComment->is(BK_Block)) | |||
| 1016 | return Current.NestingLevel == 0 ? State.FirstIndent | |||
| 1017 | : State.Stack.back().Indent; | |||
| 1018 | if ((Current.isOneOf(tok::r_brace, tok::r_square) || | |||
| 1019 | (Current.is(tok::greater) && | |||
| 1020 | (Style.Language == FormatStyle::LK_Proto || | |||
| 1021 | Style.Language == FormatStyle::LK_TextProto))) && | |||
| 1022 | State.Stack.size() > 1) { | |||
| 1023 | if (Current.closesBlockOrBlockTypeList(Style)) | |||
| 1024 | return State.Stack[State.Stack.size() - 2].NestedBlockIndent; | |||
| 1025 | if (Current.MatchingParen && Current.MatchingParen->is(BK_BracedInit)) | |||
| 1026 | return State.Stack[State.Stack.size() - 2].LastSpace; | |||
| 1027 | return State.FirstIndent; | |||
| 1028 | } | |||
| 1029 | // Indent a closing parenthesis at the previous level if followed by a semi, | |||
| 1030 | // const, or opening brace. This allows indentations such as: | |||
| 1031 | // foo( | |||
| 1032 | // a, | |||
| 1033 | // ); | |||
| 1034 | // int Foo::getter( | |||
| 1035 | // // | |||
| 1036 | // ) const { | |||
| 1037 | // return foo; | |||
| 1038 | // } | |||
| 1039 | // function foo( | |||
| 1040 | // a, | |||
| 1041 | // ) { | |||
| 1042 | // code(); // | |||
| 1043 | // } | |||
| 1044 | if (Current.is(tok::r_paren) && State.Stack.size() > 1 && | |||
| 1045 | (!Current.Next || | |||
| 1046 | Current.Next->isOneOf(tok::semi, tok::kw_const, tok::l_brace))) | |||
| 1047 | return State.Stack[State.Stack.size() - 2].LastSpace; | |||
| 1048 | if (Style.AlignAfterOpenBracket == FormatStyle::BAS_BlockIndent && | |||
| 1049 | Current.is(tok::r_paren) && State.Stack.size() > 1) | |||
| 1050 | return State.Stack[State.Stack.size() - 2].LastSpace; | |||
| 1051 | if (NextNonComment->is(TT_TemplateString) && NextNonComment->closesScope()) | |||
| 1052 | return State.Stack[State.Stack.size() - 2].LastSpace; | |||
| 1053 | if (Current.is(tok::identifier) && Current.Next && | |||
| 1054 | (Current.Next->is(TT_DictLiteral) || | |||
| 1055 | ((Style.Language == FormatStyle::LK_Proto || | |||
| 1056 | Style.Language == FormatStyle::LK_TextProto) && | |||
| 1057 | Current.Next->isOneOf(tok::less, tok::l_brace)))) | |||
| 1058 | return State.Stack.back().Indent; | |||
| 1059 | if (NextNonComment->is(TT_ObjCStringLiteral) && | |||
| 1060 | State.StartOfStringLiteral != 0) | |||
| 1061 | return State.StartOfStringLiteral - 1; | |||
| 1062 | if (NextNonComment->isStringLiteral() && State.StartOfStringLiteral != 0) | |||
| 1063 | return State.StartOfStringLiteral; | |||
| 1064 | if (NextNonComment->is(tok::lessless) && | |||
| 1065 | State.Stack.back().FirstLessLess != 0) | |||
| 1066 | return State.Stack.back().FirstLessLess; | |||
| 1067 | if (NextNonComment->isMemberAccess()) { | |||
| 1068 | if (State.Stack.back().CallContinuation == 0) | |||
| 1069 | return ContinuationIndent; | |||
| 1070 | return State.Stack.back().CallContinuation; | |||
| 1071 | } | |||
| 1072 | if (State.Stack.back().QuestionColumn != 0 && | |||
| 1073 | ((NextNonComment->is(tok::colon) && | |||
| 1074 | NextNonComment->is(TT_ConditionalExpr)) || | |||
| 1075 | Previous.is(TT_ConditionalExpr))) { | |||
| 1076 | if (((NextNonComment->is(tok::colon) && NextNonComment->Next && | |||
| 1077 | !NextNonComment->Next->FakeLParens.empty() && | |||
| 1078 | NextNonComment->Next->FakeLParens.back() == prec::Conditional) || | |||
| 1079 | (Previous.is(tok::colon) && !Current.FakeLParens.empty() && | |||
| 1080 | Current.FakeLParens.back() == prec::Conditional)) && | |||
| 1081 | !State.Stack.back().IsWrappedConditional) { | |||
| 1082 | // NOTE: we may tweak this slightly: | |||
| 1083 | // * not remove the 'lead' ContinuationIndentWidth | |||
| 1084 | // * always un-indent by the operator when | |||
| 1085 | // BreakBeforeTernaryOperators=true | |||
| 1086 | unsigned Indent = State.Stack.back().Indent; | |||
| 1087 | if (Style.AlignOperands != FormatStyle::OAS_DontAlign) { | |||
| 1088 | Indent -= Style.ContinuationIndentWidth; | |||
| 1089 | } | |||
| 1090 | if (Style.BreakBeforeTernaryOperators && | |||
| 1091 | State.Stack.back().UnindentOperator) | |||
| 1092 | Indent -= 2; | |||
| 1093 | return Indent; | |||
| 1094 | } | |||
| 1095 | return State.Stack.back().QuestionColumn; | |||
| 1096 | } | |||
| 1097 | if (Previous.is(tok::comma) && State.Stack.back().VariablePos != 0) | |||
| 1098 | return State.Stack.back().VariablePos; | |||
| 1099 | if ((PreviousNonComment && | |||
| 1100 | (PreviousNonComment->ClosesTemplateDeclaration || | |||
| 1101 | PreviousNonComment->isOneOf( | |||
| 1102 | TT_AttributeParen, TT_AttributeSquare, TT_FunctionAnnotationRParen, | |||
| 1103 | TT_JavaAnnotation, TT_LeadingJavaAnnotation))) || | |||
| 1104 | (!Style.IndentWrappedFunctionNames && | |||
| 1105 | NextNonComment->isOneOf(tok::kw_operator, TT_FunctionDeclarationName))) | |||
| 1106 | return std::max(State.Stack.back().LastSpace, State.Stack.back().Indent); | |||
| 1107 | if (NextNonComment->is(TT_SelectorName)) { | |||
| 1108 | if (!State.Stack.back().ObjCSelectorNameFound) { | |||
| 1109 | unsigned MinIndent = State.Stack.back().Indent; | |||
| 1110 | if (shouldIndentWrappedSelectorName(Style, State.Line->Type)) | |||
| 1111 | MinIndent = std::max(MinIndent, | |||
| 1112 | State.FirstIndent + Style.ContinuationIndentWidth); | |||
| 1113 | // If LongestObjCSelectorName is 0, we are indenting the first | |||
| 1114 | // part of an ObjC selector (or a selector component which is | |||
| 1115 | // not colon-aligned due to block formatting). | |||
| 1116 | // | |||
| 1117 | // Otherwise, we are indenting a subsequent part of an ObjC | |||
| 1118 | // selector which should be colon-aligned to the longest | |||
| 1119 | // component of the ObjC selector. | |||
| 1120 | // | |||
| 1121 | // In either case, we want to respect Style.IndentWrappedFunctionNames. | |||
| 1122 | return MinIndent + | |||
| 1123 | std::max(NextNonComment->LongestObjCSelectorName, | |||
| 1124 | NextNonComment->ColumnWidth) - | |||
| 1125 | NextNonComment->ColumnWidth; | |||
| 1126 | } | |||
| 1127 | if (!State.Stack.back().AlignColons) | |||
| 1128 | return State.Stack.back().Indent; | |||
| 1129 | if (State.Stack.back().ColonPos > NextNonComment->ColumnWidth) | |||
| 1130 | return State.Stack.back().ColonPos - NextNonComment->ColumnWidth; | |||
| 1131 | return State.Stack.back().Indent; | |||
| 1132 | } | |||
| 1133 | if (NextNonComment->is(tok::colon) && NextNonComment->is(TT_ObjCMethodExpr)) | |||
| 1134 | return State.Stack.back().ColonPos; | |||
| 1135 | if (NextNonComment->is(TT_ArraySubscriptLSquare)) { | |||
| 1136 | if (State.Stack.back().StartOfArraySubscripts != 0) | |||
| 1137 | return State.Stack.back().StartOfArraySubscripts; | |||
| 1138 | else if (Style.isCSharp()) // C# allows `["key"] = value` inside object | |||
| 1139 | // initializers. | |||
| 1140 | return State.Stack.back().Indent; | |||
| 1141 | return ContinuationIndent; | |||
| 1142 | } | |||
| 1143 | ||||
| 1144 | // This ensure that we correctly format ObjC methods calls without inputs, | |||
| 1145 | // i.e. where the last element isn't selector like: [callee method]; | |||
| 1146 | if (NextNonComment->is(tok::identifier) && NextNonComment->FakeRParens == 0 && | |||
| 1147 | NextNonComment->Next && NextNonComment->Next->is(TT_ObjCMethodExpr)) | |||
| 1148 | return State.Stack.back().Indent; | |||
| 1149 | ||||
| 1150 | if (NextNonComment->isOneOf(TT_StartOfName, TT_PointerOrReference) || | |||
| 1151 | Previous.isOneOf(tok::coloncolon, tok::equal, TT_JsTypeColon)) | |||
| 1152 | return ContinuationIndent; | |||
| 1153 | if (PreviousNonComment && PreviousNonComment->is(tok::colon) && | |||
| 1154 | PreviousNonComment->isOneOf(TT_ObjCMethodExpr, TT_DictLiteral)) | |||
| 1155 | return ContinuationIndent; | |||
| 1156 | if (NextNonComment->is(TT_CtorInitializerComma)) | |||
| 1157 | return State.Stack.back().Indent; | |||
| 1158 | if (PreviousNonComment && PreviousNonComment->is(TT_CtorInitializerColon) && | |||
| 1159 | Style.BreakConstructorInitializers == FormatStyle::BCIS_AfterColon) | |||
| 1160 | return State.Stack.back().Indent; | |||
| 1161 | if (PreviousNonComment && PreviousNonComment->is(TT_InheritanceColon) && | |||
| 1162 | Style.BreakInheritanceList == FormatStyle::BILS_AfterColon) | |||
| 1163 | return State.Stack.back().Indent; | |||
| 1164 | if (NextNonComment->isOneOf(TT_CtorInitializerColon, TT_InheritanceColon, | |||
| 1165 | TT_InheritanceComma)) | |||
| 1166 | return State.FirstIndent + Style.ConstructorInitializerIndentWidth; | |||
| 1167 | if (Previous.is(tok::r_paren) && !Current.isBinaryOperator() && | |||
| 1168 | !Current.isOneOf(tok::colon, tok::comment)) | |||
| 1169 | return ContinuationIndent; | |||
| 1170 | if (Current.is(TT_ProtoExtensionLSquare)) | |||
| 1171 | return State.Stack.back().Indent; | |||
| 1172 | if (Current.isBinaryOperator() && State.Stack.back().UnindentOperator) | |||
| 1173 | return State.Stack.back().Indent - Current.Tok.getLength() - | |||
| 1174 | Current.SpacesRequiredBefore; | |||
| 1175 | if (Current.isOneOf(tok::comment, TT_BlockComment, TT_LineComment) && | |||
| 1176 | NextNonComment->isBinaryOperator() && State.Stack.back().UnindentOperator) | |||
| 1177 | return State.Stack.back().Indent - NextNonComment->Tok.getLength() - | |||
| 1178 | NextNonComment->SpacesRequiredBefore; | |||
| 1179 | if (State.Stack.back().Indent == State.FirstIndent && PreviousNonComment && | |||
| 1180 | !PreviousNonComment->isOneOf(tok::r_brace, TT_CtorInitializerComma)) | |||
| 1181 | // Ensure that we fall back to the continuation indent width instead of | |||
| 1182 | // just flushing continuations left. | |||
| 1183 | return State.Stack.back().Indent + Style.ContinuationIndentWidth; | |||
| 1184 | return State.Stack.back().Indent; | |||
| 1185 | } | |||
| 1186 | ||||
| 1187 | static bool hasNestedBlockInlined(const FormatToken *Previous, | |||
| 1188 | const FormatToken &Current, | |||
| 1189 | const FormatStyle &Style) { | |||
| 1190 | if (Previous->isNot(tok::l_paren)) | |||
| 1191 | return true; | |||
| 1192 | if (Previous->ParameterCount > 1) | |||
| 1193 | return true; | |||
| 1194 | ||||
| 1195 | // Also a nested block if contains a lambda inside function with 1 parameter | |||
| 1196 | return (Style.BraceWrapping.BeforeLambdaBody && Current.is(TT_LambdaLSquare)); | |||
| 1197 | } | |||
| 1198 | ||||
| 1199 | unsigned ContinuationIndenter::moveStateToNextToken(LineState &State, | |||
| 1200 | bool DryRun, bool Newline) { | |||
| 1201 | assert(State.Stack.size())(static_cast <bool> (State.Stack.size()) ? void (0) : __assert_fail ("State.Stack.size()", "clang/lib/Format/ContinuationIndenter.cpp" , 1201, __extension__ __PRETTY_FUNCTION__)); | |||
| 1202 | const FormatToken &Current = *State.NextToken; | |||
| 1203 | ||||
| 1204 | if (Current.is(TT_CSharpGenericTypeConstraint)) | |||
| 1205 | State.Stack.back().IsCSharpGenericTypeConstraint = true; | |||
| 1206 | if (Current.isOneOf(tok::comma, TT_BinaryOperator)) | |||
| 1207 | State.Stack.back().NoLineBreakInOperand = false; | |||
| 1208 | if (Current.isOneOf(TT_InheritanceColon, TT_CSharpGenericTypeConstraintColon)) | |||
| 1209 | State.Stack.back().AvoidBinPacking = true; | |||
| 1210 | if (Current.is(tok::lessless) && Current.isNot(TT_OverloadedOperator)) { | |||
| 1211 | if (State.Stack.back().FirstLessLess == 0) | |||
| 1212 | State.Stack.back().FirstLessLess = State.Column; | |||
| 1213 | else | |||
| 1214 | State.Stack.back().LastOperatorWrapped = Newline; | |||
| 1215 | } | |||
| 1216 | if (Current.is(TT_BinaryOperator) && Current.isNot(tok::lessless)) | |||
| 1217 | State.Stack.back().LastOperatorWrapped = Newline; | |||
| 1218 | if (Current.is(TT_ConditionalExpr) && Current.Previous && | |||
| 1219 | !Current.Previous->is(TT_ConditionalExpr)) | |||
| 1220 | State.Stack.back().LastOperatorWrapped = Newline; | |||
| 1221 | if (Current.is(TT_ArraySubscriptLSquare) && | |||
| 1222 | State.Stack.back().StartOfArraySubscripts == 0) | |||
| 1223 | State.Stack.back().StartOfArraySubscripts = State.Column; | |||
| 1224 | if (Current.is(TT_ConditionalExpr) && Current.is(tok::question) && | |||
| 1225 | ((Current.MustBreakBefore) || | |||
| 1226 | (Current.getNextNonComment() && | |||
| 1227 | Current.getNextNonComment()->MustBreakBefore))) | |||
| 1228 | State.Stack.back().IsWrappedConditional = true; | |||
| 1229 | if (Style.BreakBeforeTernaryOperators && Current.is(tok::question)) | |||
| 1230 | State.Stack.back().QuestionColumn = State.Column; | |||
| 1231 | if (!Style.BreakBeforeTernaryOperators && Current.isNot(tok::colon)) { | |||
| 1232 | const FormatToken *Previous = Current.Previous; | |||
| 1233 | while (Previous && Previous->isTrailingComment()) | |||
| 1234 | Previous = Previous->Previous; | |||
| 1235 | if (Previous && Previous->is(tok::question)) | |||
| 1236 | State.Stack.back().QuestionColumn = State.Column; | |||
| 1237 | } | |||
| 1238 | if (!Current.opensScope() && !Current.closesScope() && | |||
| 1239 | !Current.is(TT_PointerOrReference)) | |||
| 1240 | State.LowestLevelOnLine = | |||
| 1241 | std::min(State.LowestLevelOnLine, Current.NestingLevel); | |||
| 1242 | if (Current.isMemberAccess()) | |||
| 1243 | State.Stack.back().StartOfFunctionCall = | |||
| 1244 | !Current.NextOperator ? 0 : State.Column; | |||
| 1245 | if (Current.is(TT_SelectorName)) | |||
| 1246 | State.Stack.back().ObjCSelectorNameFound = true; | |||
| 1247 | if (Current.is(TT_CtorInitializerColon) && | |||
| 1248 | Style.BreakConstructorInitializers != FormatStyle::BCIS_AfterColon) { | |||
| 1249 | // Indent 2 from the column, so: | |||
| 1250 | // SomeClass::SomeClass() | |||
| 1251 | // : First(...), ... | |||
| 1252 | // Next(...) | |||
| 1253 | // ^ line up here. | |||
| 1254 | State.Stack.back().Indent = | |||
| 1255 | State.Column + | |||
| 1256 | (Style.BreakConstructorInitializers == FormatStyle::BCIS_BeforeComma | |||
| 1257 | ? 0 | |||
| 1258 | : 2); | |||
| 1259 | State.Stack.back().NestedBlockIndent = State.Stack.back().Indent; | |||
| 1260 | if (Style.PackConstructorInitializers > FormatStyle::PCIS_BinPack) { | |||
| 1261 | State.Stack.back().AvoidBinPacking = true; | |||
| 1262 | State.Stack.back().BreakBeforeParameter = | |||
| 1263 | Style.PackConstructorInitializers != FormatStyle::PCIS_NextLine; | |||
| 1264 | } else { | |||
| 1265 | State.Stack.back().BreakBeforeParameter = false; | |||
| 1266 | } | |||
| 1267 | } | |||
| 1268 | if (Current.is(TT_CtorInitializerColon) && | |||
| 1269 | Style.BreakConstructorInitializers == FormatStyle::BCIS_AfterColon) { | |||
| 1270 | State.Stack.back().Indent = | |||
| 1271 | State.FirstIndent + Style.ConstructorInitializerIndentWidth; | |||
| 1272 | State.Stack.back().NestedBlockIndent = State.Stack.back().Indent; | |||
| 1273 | if (Style.PackConstructorInitializers > FormatStyle::PCIS_BinPack) | |||
| 1274 | State.Stack.back().AvoidBinPacking = true; | |||
| 1275 | } | |||
| 1276 | if (Current.is(TT_InheritanceColon)) | |||
| 1277 | State.Stack.back().Indent = | |||
| 1278 | State.FirstIndent + Style.ConstructorInitializerIndentWidth; | |||
| 1279 | if (Current.isOneOf(TT_BinaryOperator, TT_ConditionalExpr) && Newline) | |||
| 1280 | State.Stack.back().NestedBlockIndent = | |||
| 1281 | State.Column + Current.ColumnWidth + 1; | |||
| 1282 | if (Current.isOneOf(TT_LambdaLSquare, TT_LambdaArrow)) | |||
| 1283 | State.Stack.back().LastSpace = State.Column; | |||
| 1284 | ||||
| 1285 | // Insert scopes created by fake parenthesis. | |||
| 1286 | const FormatToken *Previous = Current.getPreviousNonComment(); | |||
| 1287 | ||||
| 1288 | // Add special behavior to support a format commonly used for JavaScript | |||
| 1289 | // closures: | |||
| 1290 | // SomeFunction(function() { | |||
| 1291 | // foo(); | |||
| 1292 | // bar(); | |||
| 1293 | // }, a, b, c); | |||
| 1294 | if (Current.isNot(tok::comment) && Previous && | |||
| 1295 | Previous->isOneOf(tok::l_brace, TT_ArrayInitializerLSquare) && | |||
| 1296 | !Previous->is(TT_DictLiteral) && State.Stack.size() > 1 && | |||
| 1297 | !State.Stack.back().HasMultipleNestedBlocks) { | |||
| 1298 | if (State.Stack[State.Stack.size() - 2].NestedBlockInlined && Newline) | |||
| 1299 | for (unsigned i = 0, e = State.Stack.size() - 1; i != e; ++i) | |||
| 1300 | State.Stack[i].NoLineBreak = true; | |||
| 1301 | State.Stack[State.Stack.size() - 2].NestedBlockInlined = false; | |||
| 1302 | } | |||
| 1303 | if (Previous && (Previous->isOneOf(TT_BinaryOperator, TT_ConditionalExpr) || | |||
| 1304 | (Previous->isOneOf(tok::l_paren, tok::comma, tok::colon) && | |||
| 1305 | !Previous->isOneOf(TT_DictLiteral, TT_ObjCMethodExpr)))) { | |||
| 1306 | State.Stack.back().NestedBlockInlined = | |||
| 1307 | !Newline && hasNestedBlockInlined(Previous, Current, Style); | |||
| 1308 | } | |||
| 1309 | ||||
| 1310 | moveStatePastFakeLParens(State, Newline); | |||
| 1311 | moveStatePastScopeCloser(State); | |||
| 1312 | bool AllowBreak = !State.Stack.back().NoLineBreak && | |||
| 1313 | !State.Stack.back().NoLineBreakInOperand; | |||
| 1314 | moveStatePastScopeOpener(State, Newline); | |||
| 1315 | moveStatePastFakeRParens(State); | |||
| 1316 | ||||
| 1317 | if (Current.is(TT_ObjCStringLiteral) && State.StartOfStringLiteral == 0) | |||
| 1318 | State.StartOfStringLiteral = State.Column + 1; | |||
| 1319 | if (Current.is(TT_CSharpStringLiteral) && State.StartOfStringLiteral == 0) | |||
| 1320 | State.StartOfStringLiteral = State.Column + 1; | |||
| 1321 | else if (Current.isStringLiteral() && State.StartOfStringLiteral == 0) | |||
| 1322 | State.StartOfStringLiteral = State.Column; | |||
| 1323 | else if (!Current.isOneOf(tok::comment, tok::identifier, tok::hash) && | |||
| 1324 | !Current.isStringLiteral()) | |||
| 1325 | State.StartOfStringLiteral = 0; | |||
| 1326 | ||||
| 1327 | State.Column += Current.ColumnWidth; | |||
| 1328 | State.NextToken = State.NextToken->Next; | |||
| 1329 | ||||
| 1330 | unsigned Penalty = | |||
| 1331 | handleEndOfLine(Current, State, DryRun, AllowBreak, Newline); | |||
| 1332 | ||||
| 1333 | if (Current.Role) | |||
| 1334 | Current.Role->formatFromToken(State, this, DryRun); | |||
| 1335 | // If the previous has a special role, let it consume tokens as appropriate. | |||
| 1336 | // It is necessary to start at the previous token for the only implemented | |||
| 1337 | // role (comma separated list). That way, the decision whether or not to break | |||
| 1338 | // after the "{" is already done and both options are tried and evaluated. | |||
| 1339 | // FIXME: This is ugly, find a better way. | |||
| 1340 | if (Previous && Previous->Role) | |||
| 1341 | Penalty += Previous->Role->formatAfterToken(State, this, DryRun); | |||
| 1342 | ||||
| 1343 | return Penalty; | |||
| 1344 | } | |||
| 1345 | ||||
| 1346 | void ContinuationIndenter::moveStatePastFakeLParens(LineState &State, | |||
| 1347 | bool Newline) { | |||
| 1348 | const FormatToken &Current = *State.NextToken; | |||
| 1349 | if (Current.FakeLParens.empty()) | |||
| 1350 | return; | |||
| 1351 | ||||
| 1352 | const FormatToken *Previous = Current.getPreviousNonComment(); | |||
| 1353 | ||||
| 1354 | // Don't add extra indentation for the first fake parenthesis after | |||
| 1355 | // 'return', assignments or opening <({[. The indentation for these cases | |||
| 1356 | // is special cased. | |||
| 1357 | bool SkipFirstExtraIndent = | |||
| 1358 | (Previous && (Previous->opensScope() || | |||
| 1359 | Previous->isOneOf(tok::semi, tok::kw_return) || | |||
| 1360 | (Previous->getPrecedence() == prec::Assignment && | |||
| 1361 | Style.AlignOperands != FormatStyle::OAS_DontAlign) || | |||
| 1362 | Previous->is(TT_ObjCMethodExpr))); | |||
| 1363 | for (const auto &PrecedenceLevel : llvm::reverse(Current.FakeLParens)) { | |||
| 1364 | ParenState NewParenState = State.Stack.back(); | |||
| 1365 | NewParenState.Tok = nullptr; | |||
| 1366 | NewParenState.ContainsLineBreak = false; | |||
| 1367 | NewParenState.LastOperatorWrapped = true; | |||
| 1368 | NewParenState.IsChainedConditional = false; | |||
| 1369 | NewParenState.IsWrappedConditional = false; | |||
| 1370 | NewParenState.UnindentOperator = false; | |||
| 1371 | NewParenState.NoLineBreak = | |||
| 1372 | NewParenState.NoLineBreak || State.Stack.back().NoLineBreakInOperand; | |||
| 1373 | ||||
| 1374 | // Don't propagate AvoidBinPacking into subexpressions of arg/param lists. | |||
| 1375 | if (PrecedenceLevel > prec::Comma) | |||
| 1376 | NewParenState.AvoidBinPacking = false; | |||
| 1377 | ||||
| 1378 | // Indent from 'LastSpace' unless these are fake parentheses encapsulating | |||
| 1379 | // a builder type call after 'return' or, if the alignment after opening | |||
| 1380 | // brackets is disabled. | |||
| 1381 | if (!Current.isTrailingComment() && | |||
| 1382 | (Style.AlignOperands != FormatStyle::OAS_DontAlign || | |||
| 1383 | PrecedenceLevel < prec::Assignment) && | |||
| 1384 | (!Previous || Previous->isNot(tok::kw_return) || | |||
| 1385 | (Style.Language != FormatStyle::LK_Java && PrecedenceLevel > 0)) && | |||
| 1386 | (Style.AlignAfterOpenBracket != FormatStyle::BAS_DontAlign || | |||
| 1387 | PrecedenceLevel != prec::Comma || Current.NestingLevel == 0)) { | |||
| 1388 | NewParenState.Indent = | |||
| 1389 | std::max(std::max(State.Column, NewParenState.Indent), | |||
| 1390 | State.Stack.back().LastSpace); | |||
| 1391 | } | |||
| 1392 | ||||
| 1393 | if (Previous && | |||
| 1394 | (Previous->getPrecedence() == prec::Assignment || | |||
| 1395 | Previous->is(tok::kw_return) || | |||
| 1396 | (PrecedenceLevel == prec::Conditional && Previous->is(tok::question) && | |||
| 1397 | Previous->is(TT_ConditionalExpr))) && | |||
| 1398 | !Newline) { | |||
| 1399 | // If BreakBeforeBinaryOperators is set, un-indent a bit to account for | |||
| 1400 | // the operator and keep the operands aligned | |||
| 1401 | if (Style.AlignOperands == FormatStyle::OAS_AlignAfterOperator) | |||
| 1402 | NewParenState.UnindentOperator = true; | |||
| 1403 | // Mark indentation as alignment if the expression is aligned. | |||
| 1404 | if (Style.AlignOperands != FormatStyle::OAS_DontAlign) | |||
| 1405 | NewParenState.IsAligned = true; | |||
| 1406 | } | |||
| 1407 | ||||
| 1408 | // Do not indent relative to the fake parentheses inserted for "." or "->". | |||
| 1409 | // This is a special case to make the following to statements consistent: | |||
| 1410 | // OuterFunction(InnerFunctionCall( // break | |||
| 1411 | // ParameterToInnerFunction)); | |||
| 1412 | // OuterFunction(SomeObject.InnerFunctionCall( // break | |||
| 1413 | // ParameterToInnerFunction)); | |||
| 1414 | if (PrecedenceLevel > prec::Unknown) | |||
| 1415 | NewParenState.LastSpace = std::max(NewParenState.LastSpace, State.Column); | |||
| 1416 | if (PrecedenceLevel != prec::Conditional && !Current.is(TT_UnaryOperator) && | |||
| 1417 | Style.AlignAfterOpenBracket != FormatStyle::BAS_DontAlign) | |||
| 1418 | NewParenState.StartOfFunctionCall = State.Column; | |||
| 1419 | ||||
| 1420 | // Indent conditional expressions, unless they are chained "else-if" | |||
| 1421 | // conditionals. Never indent expression where the 'operator' is ',', ';' or | |||
| 1422 | // an assignment (i.e. *I <= prec::Assignment) as those have different | |||
| 1423 | // indentation rules. Indent other expression, unless the indentation needs | |||
| 1424 | // to be skipped. | |||
| 1425 | if (PrecedenceLevel == prec::Conditional && Previous && | |||
| 1426 | Previous->is(tok::colon) && Previous->is(TT_ConditionalExpr) && | |||
| 1427 | &PrecedenceLevel == &Current.FakeLParens.back() && | |||
| 1428 | !State.Stack.back().IsWrappedConditional) { | |||
| 1429 | NewParenState.IsChainedConditional = true; | |||
| 1430 | NewParenState.UnindentOperator = State.Stack.back().UnindentOperator; | |||
| 1431 | } else if (PrecedenceLevel == prec::Conditional || | |||
| 1432 | (!SkipFirstExtraIndent && PrecedenceLevel > prec::Assignment && | |||
| 1433 | !Current.isTrailingComment())) { | |||
| 1434 | NewParenState.Indent += Style.ContinuationIndentWidth; | |||
| 1435 | } | |||
| 1436 | if ((Previous && !Previous->opensScope()) || PrecedenceLevel != prec::Comma) | |||
| 1437 | NewParenState.BreakBeforeParameter = false; | |||
| 1438 | State.Stack.push_back(NewParenState); | |||
| 1439 | SkipFirstExtraIndent = false; | |||
| 1440 | } | |||
| 1441 | } | |||
| 1442 | ||||
| 1443 | void ContinuationIndenter::moveStatePastFakeRParens(LineState &State) { | |||
| 1444 | for (unsigned i = 0, e = State.NextToken->FakeRParens; i != e; ++i) { | |||
| 1445 | unsigned VariablePos = State.Stack.back().VariablePos; | |||
| 1446 | if (State.Stack.size() == 1) { | |||
| 1447 | // Do not pop the last element. | |||
| 1448 | break; | |||
| 1449 | } | |||
| 1450 | State.Stack.pop_back(); | |||
| 1451 | State.Stack.back().VariablePos = VariablePos; | |||
| 1452 | } | |||
| 1453 | } | |||
| 1454 | ||||
| 1455 | void ContinuationIndenter::moveStatePastScopeOpener(LineState &State, | |||
| 1456 | bool Newline) { | |||
| 1457 | const FormatToken &Current = *State.NextToken; | |||
| 1458 | if (!Current.opensScope()) | |||
| 1459 | return; | |||
| 1460 | ||||
| 1461 | // Don't allow '<' or '(' in C# generic type constraints to start new scopes. | |||
| 1462 | if (Current.isOneOf(tok::less, tok::l_paren) && | |||
| 1463 | State.Stack.back().IsCSharpGenericTypeConstraint) | |||
| 1464 | return; | |||
| 1465 | ||||
| 1466 | if (Current.MatchingParen && Current.is(BK_Block)) { | |||
| 1467 | moveStateToNewBlock(State); | |||
| 1468 | return; | |||
| 1469 | } | |||
| 1470 | ||||
| 1471 | unsigned NewIndent; | |||
| 1472 | unsigned LastSpace = State.Stack.back().LastSpace; | |||
| 1473 | bool AvoidBinPacking; | |||
| 1474 | bool BreakBeforeParameter = false; | |||
| 1475 | unsigned NestedBlockIndent = std::max(State.Stack.back().StartOfFunctionCall, | |||
| 1476 | State.Stack.back().NestedBlockIndent); | |||
| 1477 | if (Current.isOneOf(tok::l_brace, TT_ArrayInitializerLSquare) || | |||
| 1478 | opensProtoMessageField(Current, Style)) { | |||
| 1479 | if (Current.opensBlockOrBlockTypeList(Style)) { | |||
| 1480 | NewIndent = Style.IndentWidth + | |||
| 1481 | std::min(State.Column, State.Stack.back().NestedBlockIndent); | |||
| 1482 | } else { | |||
| 1483 | NewIndent = State.Stack.back().LastSpace + Style.ContinuationIndentWidth; | |||
| 1484 | } | |||
| 1485 | const FormatToken *NextNoComment = Current.getNextNonComment(); | |||
| 1486 | bool EndsInComma = Current.MatchingParen && | |||
| 1487 | Current.MatchingParen->Previous && | |||
| 1488 | Current.MatchingParen->Previous->is(tok::comma); | |||
| 1489 | AvoidBinPacking = EndsInComma || Current.is(TT_DictLiteral) || | |||
| 1490 | Style.Language == FormatStyle::LK_Proto || | |||
| 1491 | Style.Language == FormatStyle::LK_TextProto || | |||
| 1492 | !Style.BinPackArguments || | |||
| 1493 | (NextNoComment && | |||
| 1494 | NextNoComment->isOneOf(TT_DesignatedInitializerPeriod, | |||
| 1495 | TT_DesignatedInitializerLSquare)); | |||
| 1496 | BreakBeforeParameter = EndsInComma; | |||
| 1497 | if (Current.ParameterCount > 1) | |||
| 1498 | NestedBlockIndent = std::max(NestedBlockIndent, State.Column + 1); | |||
| 1499 | } else { | |||
| 1500 | NewIndent = Style.ContinuationIndentWidth + | |||
| 1501 | std::max(State.Stack.back().LastSpace, | |||
| 1502 | State.Stack.back().StartOfFunctionCall); | |||
| 1503 | ||||
| 1504 | // Ensure that different different brackets force relative alignment, e.g.: | |||
| 1505 | // void SomeFunction(vector< // break | |||
| 1506 | // int> v); | |||
| 1507 | // FIXME: We likely want to do this for more combinations of brackets. | |||
| 1508 | if (Current.is(tok::less) && Current.ParentBracket == tok::l_paren) { | |||
| 1509 | NewIndent = std::max(NewIndent, State.Stack.back().Indent); | |||
| 1510 | LastSpace = std::max(LastSpace, State.Stack.back().Indent); | |||
| 1511 | } | |||
| 1512 | ||||
| 1513 | bool EndsInComma = | |||
| 1514 | Current.MatchingParen && | |||
| 1515 | Current.MatchingParen->getPreviousNonComment() && | |||
| 1516 | Current.MatchingParen->getPreviousNonComment()->is(tok::comma); | |||
| 1517 | ||||
| 1518 | // If ObjCBinPackProtocolList is unspecified, fall back to BinPackParameters | |||
| 1519 | // for backwards compatibility. | |||
| 1520 | bool ObjCBinPackProtocolList = | |||
| 1521 | (Style.ObjCBinPackProtocolList == FormatStyle::BPS_Auto && | |||
| 1522 | Style.BinPackParameters) || | |||
| 1523 | Style.ObjCBinPackProtocolList == FormatStyle::BPS_Always; | |||
| 1524 | ||||
| 1525 | bool BinPackDeclaration = | |||
| 1526 | (State.Line->Type != LT_ObjCDecl && Style.BinPackParameters) || | |||
| 1527 | (State.Line->Type == LT_ObjCDecl && ObjCBinPackProtocolList); | |||
| 1528 | ||||
| 1529 | AvoidBinPacking = | |||
| 1530 | (State.Stack.back().IsCSharpGenericTypeConstraint) || | |||
| 1531 | (Style.isJavaScript() && EndsInComma) || | |||
| 1532 | (State.Line->MustBeDeclaration && !BinPackDeclaration) || | |||
| 1533 | (!State.Line->MustBeDeclaration && !Style.BinPackArguments) || | |||
| 1534 | (Style.ExperimentalAutoDetectBinPacking && | |||
| 1535 | (Current.is(PPK_OnePerLine) || | |||
| 1536 | (!BinPackInconclusiveFunctions && Current.is(PPK_Inconclusive)))); | |||
| 1537 | ||||
| 1538 | if (Current.is(TT_ObjCMethodExpr) && Current.MatchingParen && | |||
| 1539 | Style.ObjCBreakBeforeNestedBlockParam) { | |||
| 1540 | if (Style.ColumnLimit) { | |||
| 1541 | // If this '[' opens an ObjC call, determine whether all parameters fit | |||
| 1542 | // into one line and put one per line if they don't. | |||
| 1543 | if (getLengthToMatchingParen(Current, State.Stack) + State.Column > | |||
| 1544 | getColumnLimit(State)) | |||
| 1545 | BreakBeforeParameter = true; | |||
| 1546 | } else { | |||
| 1547 | // For ColumnLimit = 0, we have to figure out whether there is or has to | |||
| 1548 | // be a line break within this call. | |||
| 1549 | for (const FormatToken *Tok = &Current; | |||
| 1550 | Tok && Tok != Current.MatchingParen; Tok = Tok->Next) { | |||
| 1551 | if (Tok->MustBreakBefore || | |||
| 1552 | (Tok->CanBreakBefore && Tok->NewlinesBefore > 0)) { | |||
| 1553 | BreakBeforeParameter = true; | |||
| 1554 | break; | |||
| 1555 | } | |||
| 1556 | } | |||
| 1557 | } | |||
| 1558 | } | |||
| 1559 | ||||
| 1560 | if (Style.isJavaScript() && EndsInComma) | |||
| 1561 | BreakBeforeParameter = true; | |||
| 1562 | } | |||
| 1563 | // Generally inherit NoLineBreak from the current scope to nested scope. | |||
| 1564 | // However, don't do this for non-empty nested blocks, dict literals and | |||
| 1565 | // array literals as these follow different indentation rules. | |||
| 1566 | bool NoLineBreak = | |||
| 1567 | Current.Children.empty() && | |||
| 1568 | !Current.isOneOf(TT_DictLiteral, TT_ArrayInitializerLSquare) && | |||
| 1569 | (State.Stack.back().NoLineBreak || | |||
| 1570 | State.Stack.back().NoLineBreakInOperand || | |||
| 1571 | (Current.is(TT_TemplateOpener) && | |||
| 1572 | State.Stack.back().ContainsUnwrappedBuilder)); | |||
| 1573 | State.Stack.push_back( | |||
| 1574 | ParenState(&Current, NewIndent, LastSpace, AvoidBinPacking, NoLineBreak)); | |||
| 1575 | State.Stack.back().NestedBlockIndent = NestedBlockIndent; | |||
| 1576 | State.Stack.back().BreakBeforeParameter = BreakBeforeParameter; | |||
| 1577 | State.Stack.back().HasMultipleNestedBlocks = | |||
| 1578 | (Current.BlockParameterCount > 1); | |||
| 1579 | ||||
| 1580 | if (Style.BraceWrapping.BeforeLambdaBody && Current.Next != nullptr && | |||
| 1581 | Current.Tok.is(tok::l_paren)) { | |||
| 1582 | // Search for any parameter that is a lambda | |||
| 1583 | FormatToken const *next = Current.Next; | |||
| 1584 | while (next != nullptr) { | |||
| 1585 | if (next->is(TT_LambdaLSquare)) { | |||
| 1586 | State.Stack.back().HasMultipleNestedBlocks = true; | |||
| 1587 | break; | |||
| 1588 | } | |||
| 1589 | next = next->Next; | |||
| 1590 | } | |||
| 1591 | } | |||
| 1592 | ||||
| 1593 | State.Stack.back().IsInsideObjCArrayLiteral = | |||
| 1594 | Current.is(TT_ArrayInitializerLSquare) && Current.Previous && | |||
| 1595 | Current.Previous->is(tok::at); | |||
| 1596 | } | |||
| 1597 | ||||
| 1598 | void ContinuationIndenter::moveStatePastScopeCloser(LineState &State) { | |||
| 1599 | const FormatToken &Current = *State.NextToken; | |||
| 1600 | if (!Current.closesScope()) | |||
| 1601 | return; | |||
| 1602 | ||||
| 1603 | // If we encounter a closing ), ], } or >, we can remove a level from our | |||
| 1604 | // stacks. | |||
| 1605 | if (State.Stack.size() > 1 && | |||
| 1606 | (Current.isOneOf(tok::r_paren, tok::r_square, TT_TemplateString) || | |||
| 1607 | (Current.is(tok::r_brace) && State.NextToken != State.Line->First) || | |||
| 1608 | State.NextToken->is(TT_TemplateCloser) || | |||
| 1609 | (Current.is(tok::greater) && Current.is(TT_DictLiteral)))) | |||
| 1610 | State.Stack.pop_back(); | |||
| 1611 | ||||
| 1612 | // Reevaluate whether ObjC message arguments fit into one line. | |||
| 1613 | // If a receiver spans multiple lines, e.g.: | |||
| 1614 | // [[object block:^{ | |||
| 1615 | // return 42; | |||
| 1616 | // }] a:42 b:42]; | |||
| 1617 | // BreakBeforeParameter is calculated based on an incorrect assumption | |||
| 1618 | // (it is checked whether the whole expression fits into one line without | |||
| 1619 | // considering a line break inside a message receiver). | |||
| 1620 | // We check whether arguments fit after receiver scope closer (into the same | |||
| 1621 | // line). | |||
| 1622 | if (State.Stack.back().BreakBeforeParameter && Current.MatchingParen && | |||
| 1623 | Current.MatchingParen->Previous) { | |||
| 1624 | const FormatToken &CurrentScopeOpener = *Current.MatchingParen->Previous; | |||
| 1625 | if (CurrentScopeOpener.is(TT_ObjCMethodExpr) && | |||
| 1626 | CurrentScopeOpener.MatchingParen) { | |||
| 1627 | int NecessarySpaceInLine = | |||
| 1628 | getLengthToMatchingParen(CurrentScopeOpener, State.Stack) + | |||
| 1629 | CurrentScopeOpener.TotalLength - Current.TotalLength - 1; | |||
| 1630 | if (State.Column + Current.ColumnWidth + NecessarySpaceInLine <= | |||
| 1631 | Style.ColumnLimit) | |||
| 1632 | State.Stack.back().BreakBeforeParameter = false; | |||
| 1633 | } | |||
| 1634 | } | |||
| 1635 | ||||
| 1636 | if (Current.is(tok::r_square)) { | |||
| 1637 | // If this ends the array subscript expr, reset the corresponding value. | |||
| 1638 | const FormatToken *NextNonComment = Current.getNextNonComment(); | |||
| 1639 | if (NextNonComment && NextNonComment->isNot(tok::l_square)) | |||
| 1640 | State.Stack.back().StartOfArraySubscripts = 0; | |||
| 1641 | } | |||
| 1642 | } | |||
| 1643 | ||||
| 1644 | void ContinuationIndenter::moveStateToNewBlock(LineState &State) { | |||
| 1645 | unsigned NestedBlockIndent = State.Stack.back().NestedBlockIndent; | |||
| 1646 | // ObjC block sometimes follow special indentation rules. | |||
| 1647 | unsigned NewIndent = | |||
| 1648 | NestedBlockIndent + (State.NextToken->is(TT_ObjCBlockLBrace) | |||
| 1649 | ? Style.ObjCBlockIndentWidth | |||
| 1650 | : Style.IndentWidth); | |||
| 1651 | State.Stack.push_back(ParenState(State.NextToken, NewIndent, | |||
| 1652 | State.Stack.back().LastSpace, | |||
| 1653 | /*AvoidBinPacking=*/true, | |||
| 1654 | /*NoLineBreak=*/false)); | |||
| 1655 | State.Stack.back().NestedBlockIndent = NestedBlockIndent; | |||
| 1656 | State.Stack.back().BreakBeforeParameter = true; | |||
| 1657 | } | |||
| 1658 | ||||
| 1659 | static unsigned getLastLineEndColumn(StringRef Text, unsigned StartColumn, | |||
| 1660 | unsigned TabWidth, | |||
| 1661 | encoding::Encoding Encoding) { | |||
| 1662 | size_t LastNewlinePos = Text.find_last_of("\n"); | |||
| 1663 | if (LastNewlinePos == StringRef::npos) { | |||
| 1664 | return StartColumn + | |||
| 1665 | encoding::columnWidthWithTabs(Text, StartColumn, TabWidth, Encoding); | |||
| 1666 | } else { | |||
| 1667 | return encoding::columnWidthWithTabs(Text.substr(LastNewlinePos), | |||
| 1668 | /*StartColumn=*/0, TabWidth, Encoding); | |||
| 1669 | } | |||
| 1670 | } | |||
| 1671 | ||||
| 1672 | unsigned ContinuationIndenter::reformatRawStringLiteral( | |||
| 1673 | const FormatToken &Current, LineState &State, | |||
| 1674 | const FormatStyle &RawStringStyle, bool DryRun, bool Newline) { | |||
| 1675 | unsigned StartColumn = State.Column - Current.ColumnWidth; | |||
| 1676 | StringRef OldDelimiter = *getRawStringDelimiter(Current.TokenText); | |||
| 1677 | StringRef NewDelimiter = | |||
| 1678 | getCanonicalRawStringDelimiter(Style, RawStringStyle.Language); | |||
| 1679 | if (NewDelimiter.empty()) | |||
| 1680 | NewDelimiter = OldDelimiter; | |||
| 1681 | // The text of a raw string is between the leading 'R"delimiter(' and the | |||
| 1682 | // trailing 'delimiter)"'. | |||
| 1683 | unsigned OldPrefixSize = 3 + OldDelimiter.size(); | |||
| 1684 | unsigned OldSuffixSize = 2 + OldDelimiter.size(); | |||
| 1685 | // We create a virtual text environment which expects a null-terminated | |||
| 1686 | // string, so we cannot use StringRef. | |||
| 1687 | std::string RawText = std::string( | |||
| 1688 | Current.TokenText.substr(OldPrefixSize).drop_back(OldSuffixSize)); | |||
| 1689 | if (NewDelimiter != OldDelimiter) { | |||
| 1690 | // Don't update to the canonical delimiter 'deli' if ')deli"' occurs in the | |||
| 1691 | // raw string. | |||
| 1692 | std::string CanonicalDelimiterSuffix = (")" + NewDelimiter + "\"").str(); | |||
| 1693 | if (StringRef(RawText).contains(CanonicalDelimiterSuffix)) | |||
| 1694 | NewDelimiter = OldDelimiter; | |||
| 1695 | } | |||
| 1696 | ||||
| 1697 | unsigned NewPrefixSize = 3 + NewDelimiter.size(); | |||
| 1698 | unsigned NewSuffixSize = 2 + NewDelimiter.size(); | |||
| 1699 | ||||
| 1700 | // The first start column is the column the raw text starts after formatting. | |||
| 1701 | unsigned FirstStartColumn = StartColumn + NewPrefixSize; | |||
| 1702 | ||||
| 1703 | // The next start column is the intended indentation a line break inside | |||
| 1704 | // the raw string at level 0. It is determined by the following rules: | |||
| 1705 | // - if the content starts on newline, it is one level more than the current | |||
| 1706 | // indent, and | |||
| 1707 | // - if the content does not start on a newline, it is the first start | |||
| 1708 | // column. | |||
| 1709 | // These rules have the advantage that the formatted content both does not | |||
| 1710 | // violate the rectangle rule and visually flows within the surrounding | |||
| 1711 | // source. | |||
| 1712 | bool ContentStartsOnNewline = Current.TokenText[OldPrefixSize] == '\n'; | |||
| 1713 | // If this token is the last parameter (checked by looking if it's followed by | |||
| 1714 | // `)` and is not on a newline, the base the indent off the line's nested | |||
| 1715 | // block indent. Otherwise, base the indent off the arguments indent, so we | |||
| 1716 | // can achieve: | |||
| 1717 | // | |||
| 1718 | // fffffffffff(1, 2, 3, R"pb( | |||
| 1719 | // key1: 1 # | |||
| 1720 | // key2: 2)pb"); | |||
| 1721 | // | |||
| 1722 | // fffffffffff(1, 2, 3, | |||
| 1723 | // R"pb( | |||
| 1724 | // key1: 1 # | |||
| 1725 | // key2: 2 | |||
| 1726 | // )pb"); | |||
| 1727 | // | |||
| 1728 | // fffffffffff(1, 2, 3, | |||
| 1729 | // R"pb( | |||
| 1730 | // key1: 1 # | |||
| 1731 | // key2: 2 | |||
| 1732 | // )pb", | |||
| 1733 | // 5); | |||
| 1734 | unsigned CurrentIndent = | |||
| 1735 | (!Newline && Current.Next && Current.Next->is(tok::r_paren)) | |||
| 1736 | ? State.Stack.back().NestedBlockIndent | |||
| 1737 | : State.Stack.back().Indent; | |||
| 1738 | unsigned NextStartColumn = ContentStartsOnNewline | |||
| 1739 | ? CurrentIndent + Style.IndentWidth | |||
| 1740 | : FirstStartColumn; | |||
| 1741 | ||||
| 1742 | // The last start column is the column the raw string suffix starts if it is | |||
| 1743 | // put on a newline. | |||
| 1744 | // The last start column is the intended indentation of the raw string postfix | |||
| 1745 | // if it is put on a newline. It is determined by the following rules: | |||
| 1746 | // - if the raw string prefix starts on a newline, it is the column where | |||
| 1747 | // that raw string prefix starts, and | |||
| 1748 | // - if the raw string prefix does not start on a newline, it is the current | |||
| 1749 | // indent. | |||
| 1750 | unsigned LastStartColumn = | |||
| 1751 | Current.NewlinesBefore ? FirstStartColumn - NewPrefixSize : CurrentIndent; | |||
| 1752 | ||||
| 1753 | std::pair<tooling::Replacements, unsigned> Fixes = internal::reformat( | |||
| 1754 | RawStringStyle, RawText, {tooling::Range(0, RawText.size())}, | |||
| 1755 | FirstStartColumn, NextStartColumn, LastStartColumn, "<stdin>", | |||
| 1756 | /*Status=*/nullptr); | |||
| 1757 | ||||
| 1758 | auto NewCode = applyAllReplacements(RawText, Fixes.first); | |||
| 1759 | tooling::Replacements NoFixes; | |||
| 1760 | if (!NewCode) { | |||
| 1761 | return addMultilineToken(Current, State); | |||
| 1762 | } | |||
| 1763 | if (!DryRun) { | |||
| 1764 | if (NewDelimiter != OldDelimiter) { | |||
| 1765 | // In 'R"delimiter(...', the delimiter starts 2 characters after the start | |||
| 1766 | // of the token. | |||
| 1767 | SourceLocation PrefixDelimiterStart = | |||
| 1768 | Current.Tok.getLocation().getLocWithOffset(2); | |||
| 1769 | auto PrefixErr = Whitespaces.addReplacement(tooling::Replacement( | |||
| 1770 | SourceMgr, PrefixDelimiterStart, OldDelimiter.size(), NewDelimiter)); | |||
| 1771 | if (PrefixErr) { | |||
| 1772 | llvm::errs() | |||
| 1773 | << "Failed to update the prefix delimiter of a raw string: " | |||
| 1774 | << llvm::toString(std::move(PrefixErr)) << "\n"; | |||
| 1775 | } | |||
| 1776 | // In 'R"delimiter(...)delimiter"', the suffix delimiter starts at | |||
| 1777 | // position length - 1 - |delimiter|. | |||
| 1778 | SourceLocation SuffixDelimiterStart = | |||
| 1779 | Current.Tok.getLocation().getLocWithOffset(Current.TokenText.size() - | |||
| 1780 | 1 - OldDelimiter.size()); | |||
| 1781 | auto SuffixErr = Whitespaces.addReplacement(tooling::Replacement( | |||
| 1782 | SourceMgr, SuffixDelimiterStart, OldDelimiter.size(), NewDelimiter)); | |||
| 1783 | if (SuffixErr) { | |||
| 1784 | llvm::errs() | |||
| 1785 | << "Failed to update the suffix delimiter of a raw string: " | |||
| 1786 | << llvm::toString(std::move(SuffixErr)) << "\n"; | |||
| 1787 | } | |||
| 1788 | } | |||
| 1789 | SourceLocation OriginLoc = | |||
| 1790 | Current.Tok.getLocation().getLocWithOffset(OldPrefixSize); | |||
| 1791 | for (const tooling::Replacement &Fix : Fixes.first) { | |||
| 1792 | auto Err = Whitespaces.addReplacement(tooling::Replacement( | |||
| 1793 | SourceMgr, OriginLoc.getLocWithOffset(Fix.getOffset()), | |||
| 1794 | Fix.getLength(), Fix.getReplacementText())); | |||
| 1795 | if (Err) { | |||
| 1796 | llvm::errs() << "Failed to reformat raw string: " | |||
| 1797 | << llvm::toString(std::move(Err)) << "\n"; | |||
| 1798 | } | |||
| 1799 | } | |||
| 1800 | } | |||
| 1801 | unsigned RawLastLineEndColumn = getLastLineEndColumn( | |||
| 1802 | *NewCode, FirstStartColumn, Style.TabWidth, Encoding); | |||
| 1803 | State.Column = RawLastLineEndColumn + NewSuffixSize; | |||
| 1804 | // Since we're updating the column to after the raw string literal here, we | |||
| 1805 | // have to manually add the penalty for the prefix R"delim( over the column | |||
| 1806 | // limit. | |||
| 1807 | unsigned PrefixExcessCharacters = | |||
| 1808 | StartColumn + NewPrefixSize > Style.ColumnLimit | |||
| 1809 | ? StartColumn + NewPrefixSize - Style.ColumnLimit | |||
| 1810 | : 0; | |||
| 1811 | bool IsMultiline = | |||
| 1812 | ContentStartsOnNewline || (NewCode->find('\n') != std::string::npos); | |||
| 1813 | if (IsMultiline) { | |||
| 1814 | // Break before further function parameters on all levels. | |||
| 1815 | for (unsigned i = 0, e = State.Stack.size(); i != e; ++i) | |||
| 1816 | State.Stack[i].BreakBeforeParameter = true; | |||
| 1817 | } | |||
| 1818 | return Fixes.second + PrefixExcessCharacters * Style.PenaltyExcessCharacter; | |||
| 1819 | } | |||
| 1820 | ||||
| 1821 | unsigned ContinuationIndenter::addMultilineToken(const FormatToken &Current, | |||
| 1822 | LineState &State) { | |||
| 1823 | // Break before further function parameters on all levels. | |||
| 1824 | for (unsigned i = 0, e = State.Stack.size(); i != e; ++i) | |||
| 1825 | State.Stack[i].BreakBeforeParameter = true; | |||
| 1826 | ||||
| 1827 | unsigned ColumnsUsed = State.Column; | |||
| 1828 | // We can only affect layout of the first and the last line, so the penalty | |||
| 1829 | // for all other lines is constant, and we ignore it. | |||
| 1830 | State.Column = Current.LastLineColumnWidth; | |||
| 1831 | ||||
| 1832 | if (ColumnsUsed > getColumnLimit(State)) | |||
| 1833 | return Style.PenaltyExcessCharacter * (ColumnsUsed - getColumnLimit(State)); | |||
| 1834 | return 0; | |||
| 1835 | } | |||
| 1836 | ||||
| 1837 | unsigned ContinuationIndenter::handleEndOfLine(const FormatToken &Current, | |||
| 1838 | LineState &State, bool DryRun, | |||
| 1839 | bool AllowBreak, bool Newline) { | |||
| 1840 | unsigned Penalty = 0; | |||
| 1841 | // Compute the raw string style to use in case this is a raw string literal | |||
| 1842 | // that can be reformatted. | |||
| 1843 | auto RawStringStyle = getRawStringStyle(Current, State); | |||
| 1844 | if (RawStringStyle && !Current.Finalized) { | |||
| 1845 | Penalty = reformatRawStringLiteral(Current, State, *RawStringStyle, DryRun, | |||
| 1846 | Newline); | |||
| 1847 | } else if (Current.IsMultiline && Current.isNot(TT_BlockComment)) { | |||
| 1848 | // Don't break multi-line tokens other than block comments and raw string | |||
| 1849 | // literals. Instead, just update the state. | |||
| 1850 | Penalty = addMultilineToken(Current, State); | |||
| 1851 | } else if (State.Line->Type != LT_ImportStatement) { | |||
| 1852 | // We generally don't break import statements. | |||
| 1853 | LineState OriginalState = State; | |||
| 1854 | ||||
| 1855 | // Whether we force the reflowing algorithm to stay strictly within the | |||
| 1856 | // column limit. | |||
| 1857 | bool Strict = false; | |||
| 1858 | // Whether the first non-strict attempt at reflowing did intentionally | |||
| 1859 | // exceed the column limit. | |||
| 1860 | bool Exceeded = false; | |||
| 1861 | std::tie(Penalty, Exceeded) = breakProtrudingToken( | |||
| 1862 | Current, State, AllowBreak, /*DryRun=*/true, Strict); | |||
| 1863 | if (Exceeded) { | |||
| 1864 | // If non-strict reflowing exceeds the column limit, try whether strict | |||
| 1865 | // reflowing leads to an overall lower penalty. | |||
| 1866 | LineState StrictState = OriginalState; | |||
| 1867 | unsigned StrictPenalty = | |||
| 1868 | breakProtrudingToken(Current, StrictState, AllowBreak, | |||
| 1869 | /*DryRun=*/true, /*Strict=*/true) | |||
| 1870 | .first; | |||
| 1871 | Strict = StrictPenalty <= Penalty; | |||
| 1872 | if (Strict) { | |||
| 1873 | Penalty = StrictPenalty; | |||
| 1874 | State = StrictState; | |||
| 1875 | } | |||
| 1876 | } | |||
| 1877 | if (!DryRun) { | |||
| 1878 | // If we're not in dry-run mode, apply the changes with the decision on | |||
| 1879 | // strictness made above. | |||
| 1880 | breakProtrudingToken(Current, OriginalState, AllowBreak, /*DryRun=*/false, | |||
| 1881 | Strict); | |||
| 1882 | } | |||
| 1883 | } | |||
| 1884 | if (State.Column > getColumnLimit(State)) { | |||
| 1885 | unsigned ExcessCharacters = State.Column - getColumnLimit(State); | |||
| 1886 | Penalty += Style.PenaltyExcessCharacter * ExcessCharacters; | |||
| 1887 | } | |||
| 1888 | return Penalty; | |||
| 1889 | } | |||
| 1890 | ||||
| 1891 | // Returns the enclosing function name of a token, or the empty string if not | |||
| 1892 | // found. | |||
| 1893 | static StringRef getEnclosingFunctionName(const FormatToken &Current) { | |||
| 1894 | // Look for: 'function(' or 'function<templates>(' before Current. | |||
| 1895 | auto Tok = Current.getPreviousNonComment(); | |||
| 1896 | if (!Tok || !Tok->is(tok::l_paren)) | |||
| 1897 | return ""; | |||
| 1898 | Tok = Tok->getPreviousNonComment(); | |||
| 1899 | if (!Tok) | |||
| 1900 | return ""; | |||
| 1901 | if (Tok->is(TT_TemplateCloser)) { | |||
| 1902 | Tok = Tok->MatchingParen; | |||
| 1903 | if (Tok) | |||
| 1904 | Tok = Tok->getPreviousNonComment(); | |||
| 1905 | } | |||
| 1906 | if (!Tok || !Tok->is(tok::identifier)) | |||
| 1907 | return ""; | |||
| 1908 | return Tok->TokenText; | |||
| 1909 | } | |||
| 1910 | ||||
| 1911 | llvm::Optional<FormatStyle> | |||
| 1912 | ContinuationIndenter::getRawStringStyle(const FormatToken &Current, | |||
| 1913 | const LineState &State) { | |||
| 1914 | if (!Current.isStringLiteral()) | |||
| 1915 | return None; | |||
| 1916 | auto Delimiter = getRawStringDelimiter(Current.TokenText); | |||
| 1917 | if (!Delimiter) | |||
| 1918 | return None; | |||
| 1919 | auto RawStringStyle = RawStringFormats.getDelimiterStyle(*Delimiter); | |||
| 1920 | if (!RawStringStyle && Delimiter->empty()) | |||
| 1921 | RawStringStyle = RawStringFormats.getEnclosingFunctionStyle( | |||
| 1922 | getEnclosingFunctionName(Current)); | |||
| 1923 | if (!RawStringStyle) | |||
| 1924 | return None; | |||
| 1925 | RawStringStyle->ColumnLimit = getColumnLimit(State); | |||
| 1926 | return RawStringStyle; | |||
| 1927 | } | |||
| 1928 | ||||
| 1929 | std::unique_ptr<BreakableToken> | |||
| 1930 | ContinuationIndenter::createBreakableToken(const FormatToken &Current, | |||
| 1931 | LineState &State, bool AllowBreak) { | |||
| 1932 | unsigned StartColumn = State.Column - Current.ColumnWidth; | |||
| 1933 | if (Current.isStringLiteral()) { | |||
| 1934 | // FIXME: String literal breaking is currently disabled for C#, Java, Json | |||
| 1935 | // and JavaScript, as it requires strings to be merged using "+" which we | |||
| 1936 | // don't support. | |||
| 1937 | if (Style.Language == FormatStyle::LK_Java || Style.isJavaScript() || | |||
| 1938 | Style.isCSharp() || Style.isJson() || !Style.BreakStringLiterals || | |||
| 1939 | !AllowBreak) | |||
| 1940 | return nullptr; | |||
| 1941 | ||||
| 1942 | // Don't break string literals inside preprocessor directives (except for | |||
| 1943 | // #define directives, as their contents are stored in separate lines and | |||
| 1944 | // are not affected by this check). | |||
| 1945 | // This way we avoid breaking code with line directives and unknown | |||
| 1946 | // preprocessor directives that contain long string literals. | |||
| 1947 | if (State.Line->Type == LT_PreprocessorDirective) | |||
| 1948 | return nullptr; | |||
| 1949 | // Exempts unterminated string literals from line breaking. The user will | |||
| 1950 | // likely want to terminate the string before any line breaking is done. | |||
| 1951 | if (Current.IsUnterminatedLiteral) | |||
| 1952 | return nullptr; | |||
| 1953 | // Don't break string literals inside Objective-C array literals (doing so | |||
| 1954 | // raises the warning -Wobjc-string-concatenation). | |||
| 1955 | if (State.Stack.back().IsInsideObjCArrayLiteral) { | |||
| 1956 | return nullptr; | |||
| 1957 | } | |||
| 1958 | ||||
| 1959 | StringRef Text = Current.TokenText; | |||
| 1960 | StringRef Prefix; | |||
| 1961 | StringRef Postfix; | |||
| 1962 | // FIXME: Handle whitespace between '_T', '(', '"..."', and ')'. | |||
| 1963 | // FIXME: Store Prefix and Suffix (or PrefixLength and SuffixLength to | |||
| 1964 | // reduce the overhead) for each FormatToken, which is a string, so that we | |||
| 1965 | // don't run multiple checks here on the hot path. | |||
| 1966 | if ((Text.endswith(Postfix = "\"") && | |||
| 1967 | (Text.startswith(Prefix = "@\"") || Text.startswith(Prefix = "\"") || | |||
| 1968 | Text.startswith(Prefix = "u\"") || Text.startswith(Prefix = "U\"") || | |||
| 1969 | Text.startswith(Prefix = "u8\"") || | |||
| 1970 | Text.startswith(Prefix = "L\""))) || | |||
| 1971 | (Text.startswith(Prefix = "_T(\"") && Text.endswith(Postfix = "\")"))) { | |||
| 1972 | // We need this to address the case where there is an unbreakable tail | |||
| 1973 | // only if certain other formatting decisions have been taken. The | |||
| 1974 | // UnbreakableTailLength of Current is an overapproximation is that case | |||
| 1975 | // and we need to be correct here. | |||
| 1976 | unsigned UnbreakableTailLength = (State.NextToken && canBreak(State)) | |||
| 1977 | ? 0 | |||
| 1978 | : Current.UnbreakableTailLength; | |||
| 1979 | return std::make_unique<BreakableStringLiteral>( | |||
| 1980 | Current, StartColumn, Prefix, Postfix, UnbreakableTailLength, | |||
| 1981 | State.Line->InPPDirective, Encoding, Style); | |||
| 1982 | } | |||
| 1983 | } else if (Current.is(TT_BlockComment)) { | |||
| 1984 | if (!Style.ReflowComments || | |||
| 1985 | // If a comment token switches formatting, like | |||
| 1986 | // /* clang-format on */, we don't want to break it further, | |||
| 1987 | // but we may still want to adjust its indentation. | |||
| 1988 | switchesFormatting(Current)) { | |||
| 1989 | return nullptr; | |||
| 1990 | } | |||
| 1991 | return std::make_unique<BreakableBlockComment>( | |||
| 1992 | Current, StartColumn, Current.OriginalColumn, !Current.Previous, | |||
| 1993 | State.Line->InPPDirective, Encoding, Style, Whitespaces.useCRLF()); | |||
| 1994 | } else if (Current.is(TT_LineComment) && | |||
| 1995 | (Current.Previous == nullptr || | |||
| 1996 | Current.Previous->isNot(TT_ImplicitStringLiteral))) { | |||
| 1997 | bool RegularComments = [&]() { | |||
| 1998 | for (const FormatToken *T = &Current; T && T->is(TT_LineComment); | |||
| 1999 | T = T->Next) { | |||
| 2000 | if (!(T->TokenText.startswith("//") || T->TokenText.startswith("#"))) | |||
| 2001 | return false; | |||
| 2002 | } | |||
| 2003 | return true; | |||
| 2004 | }(); | |||
| 2005 | if (!Style.ReflowComments || | |||
| 2006 | CommentPragmasRegex.match(Current.TokenText.substr(2)) || | |||
| 2007 | switchesFormatting(Current) || !RegularComments) | |||
| 2008 | return nullptr; | |||
| 2009 | return std::make_unique<BreakableLineCommentSection>( | |||
| 2010 | Current, StartColumn, /*InPPDirective=*/false, Encoding, Style); | |||
| 2011 | } | |||
| 2012 | return nullptr; | |||
| 2013 | } | |||
| 2014 | ||||
| 2015 | std::pair<unsigned, bool> | |||
| 2016 | ContinuationIndenter::breakProtrudingToken(const FormatToken &Current, | |||
| 2017 | LineState &State, bool AllowBreak, | |||
| 2018 | bool DryRun, bool Strict) { | |||
| 2019 | std::unique_ptr<const BreakableToken> Token = | |||
| 2020 | createBreakableToken(Current, State, AllowBreak); | |||
| 2021 | if (!Token) | |||
| 2022 | return {0, false}; | |||
| 2023 | assert(Token->getLineCount() > 0)(static_cast <bool> (Token->getLineCount() > 0) ? void (0) : __assert_fail ("Token->getLineCount() > 0", "clang/lib/Format/ContinuationIndenter.cpp", 2023, __extension__ __PRETTY_FUNCTION__)); | |||
| 2024 | unsigned ColumnLimit = getColumnLimit(State); | |||
| 2025 | if (Current.is(TT_LineComment)) { | |||
| 2026 | // We don't insert backslashes when breaking line comments. | |||
| 2027 | ColumnLimit = Style.ColumnLimit; | |||
| 2028 | } | |||
| 2029 | if (ColumnLimit == 0) { | |||
| 2030 | // To make the rest of the function easier set the column limit to the | |||
| 2031 | // maximum, if there should be no limit. | |||
| 2032 | ColumnLimit = std::numeric_limits<decltype(ColumnLimit)>::max(); | |||
| 2033 | } | |||
| 2034 | if (Current.UnbreakableTailLength >= ColumnLimit) | |||
| 2035 | return {0, false}; | |||
| 2036 | // ColumnWidth was already accounted into State.Column before calling | |||
| 2037 | // breakProtrudingToken. | |||
| 2038 | unsigned StartColumn = State.Column - Current.ColumnWidth; | |||
| 2039 | unsigned NewBreakPenalty = Current.isStringLiteral() | |||
| 2040 | ? Style.PenaltyBreakString | |||
| 2041 | : Style.PenaltyBreakComment; | |||
| 2042 | // Stores whether we intentionally decide to let a line exceed the column | |||
| 2043 | // limit. | |||
| 2044 | bool Exceeded = false; | |||
| 2045 | // Stores whether we introduce a break anywhere in the token. | |||
| 2046 | bool BreakInserted = Token->introducesBreakBeforeToken(); | |||
| 2047 | // Store whether we inserted a new line break at the end of the previous | |||
| 2048 | // logical line. | |||
| 2049 | bool NewBreakBefore = false; | |||
| 2050 | // We use a conservative reflowing strategy. Reflow starts after a line is | |||
| 2051 | // broken or the corresponding whitespace compressed. Reflow ends as soon as a | |||
| 2052 | // line that doesn't get reflown with the previous line is reached. | |||
| 2053 | bool Reflow = false; | |||
| 2054 | // Keep track of where we are in the token: | |||
| 2055 | // Where we are in the content of the current logical line. | |||
| 2056 | unsigned TailOffset = 0; | |||
| 2057 | // The column number we're currently at. | |||
| 2058 | unsigned ContentStartColumn = | |||
| 2059 | Token->getContentStartColumn(0, /*Break=*/false); | |||
| 2060 | // The number of columns left in the current logical line after TailOffset. | |||
| 2061 | unsigned RemainingTokenColumns = | |||
| 2062 | Token->getRemainingLength(0, TailOffset, ContentStartColumn); | |||
| 2063 | // Adapt the start of the token, for example indent. | |||
| 2064 | if (!DryRun) | |||
| 2065 | Token->adaptStartOfLine(0, Whitespaces); | |||
| 2066 | ||||
| 2067 | unsigned ContentIndent = 0; | |||
| 2068 | unsigned Penalty = 0; | |||
| 2069 | LLVM_DEBUG(llvm::dbgs() << "Breaking protruding token at column "do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << "Breaking protruding token at column " << StartColumn << ".\n"; } } while (false) | |||
| 2070 | << StartColumn << ".\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << "Breaking protruding token at column " << StartColumn << ".\n"; } } while (false); | |||
| 2071 | for (unsigned LineIndex = 0, EndIndex = Token->getLineCount(); | |||
| 2072 | LineIndex != EndIndex; ++LineIndex) { | |||
| 2073 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Line: " << LineIndex << " (Reflow: " << Reflow << ")\n" ; } } while (false) | |||
| 2074 | << " Line: " << LineIndex << " (Reflow: " << Reflow << ")\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Line: " << LineIndex << " (Reflow: " << Reflow << ")\n" ; } } while (false); | |||
| 2075 | NewBreakBefore = false; | |||
| 2076 | // If we did reflow the previous line, we'll try reflowing again. Otherwise | |||
| 2077 | // we'll start reflowing if the current line is broken or whitespace is | |||
| 2078 | // compressed. | |||
| 2079 | bool TryReflow = Reflow; | |||
| 2080 | // Break the current token until we can fit the rest of the line. | |||
| 2081 | while (ContentStartColumn + RemainingTokenColumns > ColumnLimit) { | |||
| 2082 | LLVM_DEBUG(llvm::dbgs() << " Over limit, need: "do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2083 | << (ContentStartColumn + RemainingTokenColumns)do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2084 | << ", space: " << ColumnLimitdo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2085 | << ", reflown prefix: " << ContentStartColumndo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2086 | << ", offset in line: " << TailOffset << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false); | |||
| 2087 | // If the current token doesn't fit, find the latest possible split in the | |||
| 2088 | // current line so that breaking at it will be under the column limit. | |||
| 2089 | // FIXME: Use the earliest possible split while reflowing to correctly | |||
| 2090 | // compress whitespace within a line. | |||
| 2091 | BreakableToken::Split Split = | |||
| 2092 | Token->getSplit(LineIndex, TailOffset, ColumnLimit, | |||
| 2093 | ContentStartColumn, CommentPragmasRegex); | |||
| 2094 | if (Split.first == StringRef::npos) { | |||
| 2095 | // No break opportunity - update the penalty and continue with the next | |||
| 2096 | // logical line. | |||
| 2097 | if (LineIndex < EndIndex - 1) | |||
| 2098 | // The last line's penalty is handled in addNextStateToQueue() or when | |||
| 2099 | // calling replaceWhitespaceAfterLastLine below. | |||
| 2100 | Penalty += Style.PenaltyExcessCharacter * | |||
| 2101 | (ContentStartColumn + RemainingTokenColumns - ColumnLimit); | |||
| 2102 | LLVM_DEBUG(llvm::dbgs() << " No break opportunity.\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " No break opportunity.\n" ; } } while (false); | |||
| 2103 | break; | |||
| 2104 | } | |||
| 2105 | assert(Split.first != 0)(static_cast <bool> (Split.first != 0) ? void (0) : __assert_fail ("Split.first != 0", "clang/lib/Format/ContinuationIndenter.cpp" , 2105, __extension__ __PRETTY_FUNCTION__)); | |||
| 2106 | ||||
| 2107 | if (Token->supportsReflow()) { | |||
| 2108 | // Check whether the next natural split point after the current one can | |||
| 2109 | // still fit the line, either because we can compress away whitespace, | |||
| 2110 | // or because the penalty the excess characters introduce is lower than | |||
| 2111 | // the break penalty. | |||
| 2112 | // We only do this for tokens that support reflowing, and thus allow us | |||
| 2113 | // to change the whitespace arbitrarily (e.g. comments). | |||
| 2114 | // Other tokens, like string literals, can be broken on arbitrary | |||
| 2115 | // positions. | |||
| 2116 | ||||
| 2117 | // First, compute the columns from TailOffset to the next possible split | |||
| 2118 | // position. | |||
| 2119 | // For example: | |||
| 2120 | // ColumnLimit: | | |||
| 2121 | // // Some text that breaks | |||
| 2122 | // ^ tail offset | |||
| 2123 | // ^-- split | |||
| 2124 | // ^-------- to split columns | |||
| 2125 | // ^--- next split | |||
| 2126 | // ^--------------- to next split columns | |||
| 2127 | unsigned ToSplitColumns = Token->getRangeLength( | |||
| 2128 | LineIndex, TailOffset, Split.first, ContentStartColumn); | |||
| 2129 | LLVM_DEBUG(llvm::dbgs() << " ToSplit: " << ToSplitColumns << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ToSplit: " << ToSplitColumns << "\n"; } } while (false); | |||
| 2130 | ||||
| 2131 | BreakableToken::Split NextSplit = Token->getSplit( | |||
| 2132 | LineIndex, TailOffset + Split.first + Split.second, ColumnLimit, | |||
| 2133 | ContentStartColumn + ToSplitColumns + 1, CommentPragmasRegex); | |||
| 2134 | // Compute the columns necessary to fit the next non-breakable sequence | |||
| 2135 | // into the current line. | |||
| 2136 | unsigned ToNextSplitColumns = 0; | |||
| 2137 | if (NextSplit.first == StringRef::npos) { | |||
| 2138 | ToNextSplitColumns = Token->getRemainingLength(LineIndex, TailOffset, | |||
| 2139 | ContentStartColumn); | |||
| 2140 | } else { | |||
| 2141 | ToNextSplitColumns = Token->getRangeLength( | |||
| 2142 | LineIndex, TailOffset, | |||
| 2143 | Split.first + Split.second + NextSplit.first, ContentStartColumn); | |||
| 2144 | } | |||
| 2145 | // Compress the whitespace between the break and the start of the next | |||
| 2146 | // unbreakable sequence. | |||
| 2147 | ToNextSplitColumns = | |||
| 2148 | Token->getLengthAfterCompression(ToNextSplitColumns, Split); | |||
| 2149 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ContentStartColumn: " << ContentStartColumn << "\n"; } } while (false) | |||
| 2150 | << " ContentStartColumn: " << ContentStartColumn << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ContentStartColumn: " << ContentStartColumn << "\n"; } } while (false); | |||
| 2151 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ToNextSplit: " << ToNextSplitColumns << "\n"; } } while (false) | |||
| 2152 | << " ToNextSplit: " << ToNextSplitColumns << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ToNextSplit: " << ToNextSplitColumns << "\n"; } } while (false); | |||
| 2153 | // If the whitespace compression makes us fit, continue on the current | |||
| 2154 | // line. | |||
| 2155 | bool ContinueOnLine = | |||
| 2156 | ContentStartColumn + ToNextSplitColumns <= ColumnLimit; | |||
| 2157 | unsigned ExcessCharactersPenalty = 0; | |||
| 2158 | if (!ContinueOnLine && !Strict) { | |||
| 2159 | // Similarly, if the excess characters' penalty is lower than the | |||
| 2160 | // penalty of introducing a new break, continue on the current line. | |||
| 2161 | ExcessCharactersPenalty = | |||
| 2162 | (ContentStartColumn + ToNextSplitColumns - ColumnLimit) * | |||
| 2163 | Style.PenaltyExcessCharacter; | |||
| 2164 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Penalty excess: " << ExcessCharactersPenalty << "\n break : " << NewBreakPenalty << "\n"; } } while (false) | |||
| 2165 | << " Penalty excess: " << ExcessCharactersPenaltydo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Penalty excess: " << ExcessCharactersPenalty << "\n break : " << NewBreakPenalty << "\n"; } } while (false) | |||
| 2166 | << "\n break : " << NewBreakPenalty << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Penalty excess: " << ExcessCharactersPenalty << "\n break : " << NewBreakPenalty << "\n"; } } while (false); | |||
| 2167 | if (ExcessCharactersPenalty < NewBreakPenalty) { | |||
| 2168 | Exceeded = true; | |||
| 2169 | ContinueOnLine = true; | |||
| 2170 | } | |||
| 2171 | } | |||
| 2172 | if (ContinueOnLine) { | |||
| 2173 | LLVM_DEBUG(llvm::dbgs() << " Continuing on line...\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Continuing on line...\n" ; } } while (false); | |||
| 2174 | // The current line fits after compressing the whitespace - reflow | |||
| 2175 | // the next line into it if possible. | |||
| 2176 | TryReflow = true; | |||
| 2177 | if (!DryRun) | |||
| 2178 | Token->compressWhitespace(LineIndex, TailOffset, Split, | |||
| 2179 | Whitespaces); | |||
| 2180 | // When we continue on the same line, leave one space between content. | |||
| 2181 | ContentStartColumn += ToSplitColumns + 1; | |||
| 2182 | Penalty += ExcessCharactersPenalty; | |||
| 2183 | TailOffset += Split.first + Split.second; | |||
| 2184 | RemainingTokenColumns = Token->getRemainingLength( | |||
| 2185 | LineIndex, TailOffset, ContentStartColumn); | |||
| 2186 | continue; | |||
| 2187 | } | |||
| 2188 | } | |||
| 2189 | LLVM_DEBUG(llvm::dbgs() << " Breaking...\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Breaking...\n" ; } } while (false); | |||
| 2190 | // Update the ContentIndent only if the current line was not reflown with | |||
| 2191 | // the previous line, since in that case the previous line should still | |||
| 2192 | // determine the ContentIndent. Also never intent the last line. | |||
| 2193 | if (!Reflow) | |||
| 2194 | ContentIndent = Token->getContentIndent(LineIndex); | |||
| 2195 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ContentIndent: " << ContentIndent << "\n"; } } while (false) | |||
| 2196 | << " ContentIndent: " << ContentIndent << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " ContentIndent: " << ContentIndent << "\n"; } } while (false); | |||
| 2197 | ContentStartColumn = ContentIndent + Token->getContentStartColumn( | |||
| 2198 | LineIndex, /*Break=*/true); | |||
| 2199 | ||||
| 2200 | unsigned NewRemainingTokenColumns = Token->getRemainingLength( | |||
| 2201 | LineIndex, TailOffset + Split.first + Split.second, | |||
| 2202 | ContentStartColumn); | |||
| 2203 | if (NewRemainingTokenColumns == 0) { | |||
| 2204 | // No content to indent. | |||
| 2205 | ContentIndent = 0; | |||
| 2206 | ContentStartColumn = | |||
| 2207 | Token->getContentStartColumn(LineIndex, /*Break=*/true); | |||
| 2208 | NewRemainingTokenColumns = Token->getRemainingLength( | |||
| 2209 | LineIndex, TailOffset + Split.first + Split.second, | |||
| 2210 | ContentStartColumn); | |||
| 2211 | } | |||
| 2212 | ||||
| 2213 | // When breaking before a tab character, it may be moved by a few columns, | |||
| 2214 | // but will still be expanded to the next tab stop, so we don't save any | |||
| 2215 | // columns. | |||
| 2216 | if (NewRemainingTokenColumns >= RemainingTokenColumns) { | |||
| 2217 | // FIXME: Do we need to adjust the penalty? | |||
| 2218 | break; | |||
| 2219 | } | |||
| 2220 | ||||
| 2221 | LLVM_DEBUG(llvm::dbgs() << " Breaking at: " << TailOffset + Split.firstdo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Breaking at: " << TailOffset + Split.first << ", " << Split .second << "\n"; } } while (false) | |||
| 2222 | << ", " << Split.second << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Breaking at: " << TailOffset + Split.first << ", " << Split .second << "\n"; } } while (false); | |||
| 2223 | if (!DryRun) | |||
| 2224 | Token->insertBreak(LineIndex, TailOffset, Split, ContentIndent, | |||
| 2225 | Whitespaces); | |||
| 2226 | ||||
| 2227 | Penalty += NewBreakPenalty; | |||
| 2228 | TailOffset += Split.first + Split.second; | |||
| 2229 | RemainingTokenColumns = NewRemainingTokenColumns; | |||
| 2230 | BreakInserted = true; | |||
| 2231 | NewBreakBefore = true; | |||
| 2232 | } | |||
| 2233 | // In case there's another line, prepare the state for the start of the next | |||
| 2234 | // line. | |||
| 2235 | if (LineIndex + 1 != EndIndex) { | |||
| 2236 | unsigned NextLineIndex = LineIndex + 1; | |||
| 2237 | if (NewBreakBefore) | |||
| 2238 | // After breaking a line, try to reflow the next line into the current | |||
| 2239 | // one once RemainingTokenColumns fits. | |||
| 2240 | TryReflow = true; | |||
| 2241 | if (TryReflow) { | |||
| 2242 | // We decided that we want to try reflowing the next line into the | |||
| 2243 | // current one. | |||
| 2244 | // We will now adjust the state as if the reflow is successful (in | |||
| 2245 | // preparation for the next line), and see whether that works. If we | |||
| 2246 | // decide that we cannot reflow, we will later reset the state to the | |||
| 2247 | // start of the next line. | |||
| 2248 | Reflow = false; | |||
| 2249 | // As we did not continue breaking the line, RemainingTokenColumns is | |||
| 2250 | // known to fit after ContentStartColumn. Adapt ContentStartColumn to | |||
| 2251 | // the position at which we want to format the next line if we do | |||
| 2252 | // actually reflow. | |||
| 2253 | // When we reflow, we need to add a space between the end of the current | |||
| 2254 | // line and the next line's start column. | |||
| 2255 | ContentStartColumn += RemainingTokenColumns + 1; | |||
| 2256 | // Get the split that we need to reflow next logical line into the end | |||
| 2257 | // of the current one; the split will include any leading whitespace of | |||
| 2258 | // the next logical line. | |||
| 2259 | BreakableToken::Split SplitBeforeNext = | |||
| 2260 | Token->getReflowSplit(NextLineIndex, CommentPragmasRegex); | |||
| 2261 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Size of reflown text: " << ContentStartColumn << "\n Potential reflow split: " ; } } while (false) | |||
| 2262 | << " Size of reflown text: " << ContentStartColumndo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Size of reflown text: " << ContentStartColumn << "\n Potential reflow split: " ; } } while (false) | |||
| 2263 | << "\n Potential reflow split: ")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Size of reflown text: " << ContentStartColumn << "\n Potential reflow split: " ; } } while (false); | |||
| 2264 | if (SplitBeforeNext.first != StringRef::npos) { | |||
| 2265 | LLVM_DEBUG(llvm::dbgs() << SplitBeforeNext.first << ", "do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << SplitBeforeNext. first << ", " << SplitBeforeNext.second << "\n" ; } } while (false) | |||
| 2266 | << SplitBeforeNext.second << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << SplitBeforeNext. first << ", " << SplitBeforeNext.second << "\n" ; } } while (false); | |||
| 2267 | TailOffset = SplitBeforeNext.first + SplitBeforeNext.second; | |||
| 2268 | // If the rest of the next line fits into the current line below the | |||
| 2269 | // column limit, we can safely reflow. | |||
| 2270 | RemainingTokenColumns = Token->getRemainingLength( | |||
| 2271 | NextLineIndex, TailOffset, ContentStartColumn); | |||
| 2272 | Reflow = true; | |||
| 2273 | if (ContentStartColumn + RemainingTokenColumns > ColumnLimit) { | |||
| 2274 | LLVM_DEBUG(llvm::dbgs()do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit after reflow, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2275 | << " Over limit after reflow, need: "do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit after reflow, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2276 | << (ContentStartColumn + RemainingTokenColumns)do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit after reflow, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2277 | << ", space: " << ColumnLimitdo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit after reflow, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2278 | << ", reflown prefix: " << ContentStartColumndo { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit after reflow, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false) | |||
| 2279 | << ", offset in line: " << TailOffset << "\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Over limit after reflow, need: " << (ContentStartColumn + RemainingTokenColumns) << ", space: " << ColumnLimit << ", reflown prefix: " << ContentStartColumn << ", offset in line: " << TailOffset << "\n"; } } while (false); | |||
| 2280 | // If the whole next line does not fit, try to find a point in | |||
| 2281 | // the next line at which we can break so that attaching the part | |||
| 2282 | // of the next line to that break point onto the current line is | |||
| 2283 | // below the column limit. | |||
| 2284 | BreakableToken::Split Split = | |||
| 2285 | Token->getSplit(NextLineIndex, TailOffset, ColumnLimit, | |||
| 2286 | ContentStartColumn, CommentPragmasRegex); | |||
| 2287 | if (Split.first == StringRef::npos) { | |||
| 2288 | LLVM_DEBUG(llvm::dbgs() << " Did not find later break\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Did not find later break\n" ; } } while (false); | |||
| 2289 | Reflow = false; | |||
| 2290 | } else { | |||
| 2291 | // Check whether the first split point gets us below the column | |||
| 2292 | // limit. Note that we will execute this split below as part of | |||
| 2293 | // the normal token breaking and reflow logic within the line. | |||
| 2294 | unsigned ToSplitColumns = Token->getRangeLength( | |||
| 2295 | NextLineIndex, TailOffset, Split.first, ContentStartColumn); | |||
| 2296 | if (ContentStartColumn + ToSplitColumns > ColumnLimit) { | |||
| 2297 | LLVM_DEBUG(llvm::dbgs() << " Next split protrudes, need: "do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Next split protrudes, need: " << (ContentStartColumn + ToSplitColumns) << ", space: " << ColumnLimit; } } while (false) | |||
| 2298 | << (ContentStartColumn + ToSplitColumns)do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Next split protrudes, need: " << (ContentStartColumn + ToSplitColumns) << ", space: " << ColumnLimit; } } while (false) | |||
| 2299 | << ", space: " << ColumnLimit)do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << " Next split protrudes, need: " << (ContentStartColumn + ToSplitColumns) << ", space: " << ColumnLimit; } } while (false); | |||
| 2300 | unsigned ExcessCharactersPenalty = | |||
| 2301 | (ContentStartColumn + ToSplitColumns - ColumnLimit) * | |||
| 2302 | Style.PenaltyExcessCharacter; | |||
| 2303 | if (NewBreakPenalty < ExcessCharactersPenalty) { | |||
| 2304 | Reflow = false; | |||
| 2305 | } | |||
| 2306 | } | |||
| 2307 | } | |||
| 2308 | } | |||
| 2309 | } else { | |||
| 2310 | LLVM_DEBUG(llvm::dbgs() << "not found.\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << "not found.\n"; } } while (false); | |||
| 2311 | } | |||
| 2312 | } | |||
| 2313 | if (!Reflow) { | |||
| 2314 | // If we didn't reflow into the next line, the only space to consider is | |||
| 2315 | // the next logical line. Reset our state to match the start of the next | |||
| 2316 | // line. | |||
| 2317 | TailOffset = 0; | |||
| 2318 | ContentStartColumn = | |||
| 2319 | Token->getContentStartColumn(NextLineIndex, /*Break=*/false); | |||
| 2320 | RemainingTokenColumns = Token->getRemainingLength( | |||
| 2321 | NextLineIndex, TailOffset, ContentStartColumn); | |||
| 2322 | // Adapt the start of the token, for example indent. | |||
| 2323 | if (!DryRun) | |||
| 2324 | Token->adaptStartOfLine(NextLineIndex, Whitespaces); | |||
| 2325 | } else { | |||
| 2326 | // If we found a reflow split and have added a new break before the next | |||
| 2327 | // line, we are going to remove the line break at the start of the next | |||
| 2328 | // logical line. For example, here we'll add a new line break after | |||
| 2329 | // 'text', and subsequently delete the line break between 'that' and | |||
| 2330 | // 'reflows'. | |||
| 2331 | // // some text that | |||
| 2332 | // // reflows | |||
| 2333 | // -> | |||
| 2334 | // // some text | |||
| 2335 | // // that reflows | |||
| 2336 | // When adding the line break, we also added the penalty for it, so we | |||
| 2337 | // need to subtract that penalty again when we remove the line break due | |||
| 2338 | // to reflowing. | |||
| 2339 | if (NewBreakBefore) { | |||
| 2340 | assert(Penalty >= NewBreakPenalty)(static_cast <bool> (Penalty >= NewBreakPenalty) ? void (0) : __assert_fail ("Penalty >= NewBreakPenalty", "clang/lib/Format/ContinuationIndenter.cpp" , 2340, __extension__ __PRETTY_FUNCTION__)); | |||
| 2341 | Penalty -= NewBreakPenalty; | |||
| 2342 | } | |||
| 2343 | if (!DryRun) | |||
| 2344 | Token->reflow(NextLineIndex, Whitespaces); | |||
| 2345 | } | |||
| 2346 | } | |||
| 2347 | } | |||
| 2348 | ||||
| 2349 | BreakableToken::Split SplitAfterLastLine = | |||
| 2350 | Token->getSplitAfterLastLine(TailOffset); | |||
| 2351 | if (SplitAfterLastLine.first != StringRef::npos) { | |||
| 2352 | LLVM_DEBUG(llvm::dbgs() << "Replacing whitespace after last line.\n")do { if (::llvm::DebugFlag && ::llvm::isCurrentDebugType ("format-indenter")) { llvm::dbgs() << "Replacing whitespace after last line.\n" ; } } while (false); | |||
| 2353 | ||||
| 2354 | // We add the last line's penalty here, since that line is going to be split | |||
| 2355 | // now. | |||
| 2356 | Penalty += Style.PenaltyExcessCharacter * | |||
| 2357 | (ContentStartColumn + RemainingTokenColumns - ColumnLimit); | |||
| 2358 | ||||
| 2359 | if (!DryRun) | |||
| 2360 | Token->replaceWhitespaceAfterLastLine(TailOffset, SplitAfterLastLine, | |||
| 2361 | Whitespaces); | |||
| 2362 | ContentStartColumn = | |||
| 2363 | Token->getContentStartColumn(Token->getLineCount() - 1, /*Break=*/true); | |||
| 2364 | RemainingTokenColumns = Token->getRemainingLength( | |||
| 2365 | Token->getLineCount() - 1, | |||
| 2366 | TailOffset + SplitAfterLastLine.first + SplitAfterLastLine.second, | |||
| 2367 | ContentStartColumn); | |||
| 2368 | } | |||
| 2369 | ||||
| 2370 | State.Column = ContentStartColumn + RemainingTokenColumns - | |||
| 2371 | Current.UnbreakableTailLength; | |||
| 2372 | ||||
| 2373 | if (BreakInserted) { | |||
| 2374 | // If we break the token inside a parameter list, we need to break before | |||
| 2375 | // the next parameter on all levels, so that the next parameter is clearly | |||
| 2376 | // visible. Line comments already introduce a break. | |||
| 2377 | if (Current.isNot(TT_LineComment)) { | |||
| 2378 | for (unsigned i = 0, e = State.Stack.size(); i != e; ++i) | |||
| 2379 | State.Stack[i].BreakBeforeParameter = true; | |||
| 2380 | } | |||
| 2381 | ||||
| 2382 | if (Current.is(TT_BlockComment)) | |||
| 2383 | State.NoContinuation = true; | |||
| 2384 | ||||
| 2385 | State.Stack.back().LastSpace = StartColumn; | |||
| 2386 | } | |||
| 2387 | ||||
| 2388 | Token->updateNextToken(State); | |||
| 2389 | ||||
| 2390 | return {Penalty, Exceeded}; | |||
| 2391 | } | |||
| 2392 | ||||
| 2393 | unsigned ContinuationIndenter::getColumnLimit(const LineState &State) const { | |||
| 2394 | // In preprocessor directives reserve two chars for trailing " \" | |||
| 2395 | return Style.ColumnLimit - (State.Line->InPPDirective ? 2 : 0); | |||
| 2396 | } | |||
| 2397 | ||||
| 2398 | bool ContinuationIndenter::nextIsMultilineString(const LineState &State) { | |||
| 2399 | const FormatToken &Current = *State.NextToken; | |||
| 2400 | if (!Current.isStringLiteral() || Current.is(TT_ImplicitStringLiteral)) | |||
| 2401 | return false; | |||
| 2402 | // We never consider raw string literals "multiline" for the purpose of | |||
| 2403 | // AlwaysBreakBeforeMultilineStrings implementation as they are special-cased | |||
| 2404 | // (see TokenAnnotator::mustBreakBefore(). | |||
| 2405 | if (Current.TokenText.startswith("R\"")) | |||
| 2406 | return false; | |||
| 2407 | if (Current.IsMultiline) | |||
| 2408 | return true; | |||
| 2409 | if (Current.getNextNonComment() && | |||
| 2410 | Current.getNextNonComment()->isStringLiteral()) | |||
| 2411 | return true; // Implicit concatenation. | |||
| 2412 | if (Style.ColumnLimit != 0 && Style.BreakStringLiterals && | |||
| 2413 | State.Column + Current.ColumnWidth + Current.UnbreakableTailLength > | |||
| 2414 | Style.ColumnLimit) | |||
| 2415 | return true; // String will be split. | |||
| 2416 | return false; | |||
| 2417 | } | |||
| 2418 | ||||
| 2419 | } // namespace format | |||
| 2420 | } // namespace clang |
| 1 | //===--- FormatToken.h - Format C++ code ------------------------*- C++ -*-===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | /// |
| 9 | /// \file |
| 10 | /// This file contains the declaration of the FormatToken, a wrapper |
| 11 | /// around Token with additional information related to formatting. |
| 12 | /// |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | #ifndef LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H |
| 16 | #define LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H |
| 17 | |
| 18 | #include "clang/Basic/IdentifierTable.h" |
| 19 | #include "clang/Basic/OperatorPrecedence.h" |
| 20 | #include "clang/Format/Format.h" |
| 21 | #include "clang/Lex/Lexer.h" |
| 22 | #include <memory> |
| 23 | #include <unordered_set> |
| 24 | |
| 25 | namespace clang { |
| 26 | namespace format { |
| 27 | |
| 28 | #define LIST_TOKEN_TYPESTYPE(ArrayInitializerLSquare) TYPE(ArraySubscriptLSquare) TYPE (AttributeColon) TYPE(AttributeMacro) TYPE(AttributeParen) TYPE (AttributeSquare) TYPE(BinaryOperator) TYPE(BitFieldColon) TYPE (BlockComment) TYPE(CastRParen) TYPE(ConditionalExpr) TYPE(ConflictAlternative ) TYPE(ConflictEnd) TYPE(ConflictStart) TYPE(ConstraintJunctions ) TYPE(CtorInitializerColon) TYPE(CtorInitializerComma) TYPE( DesignatedInitializerLSquare) TYPE(DesignatedInitializerPeriod ) TYPE(DictLiteral) TYPE(FatArrow) TYPE(ForEachMacro) TYPE(FunctionAnnotationRParen ) TYPE(FunctionDeclarationName) TYPE(FunctionLBrace) TYPE(FunctionTypeLParen ) TYPE(IfMacro) TYPE(ImplicitStringLiteral) TYPE(InheritanceColon ) TYPE(InheritanceComma) TYPE(InlineASMBrace) TYPE(InlineASMColon ) TYPE(InlineASMSymbolicNameLSquare) TYPE(JavaAnnotation) TYPE (JsComputedPropertyName) TYPE(JsExponentiation) TYPE(JsExponentiationEqual ) TYPE(JsPipePipeEqual) TYPE(JsPrivateIdentifier) TYPE(JsTypeColon ) TYPE(JsTypeOperator) TYPE(JsTypeOptionalQuestion) TYPE(JsAndAndEqual ) TYPE(LambdaArrow) TYPE(LambdaLBrace) TYPE(LambdaLSquare) TYPE (LeadingJavaAnnotation) TYPE(LineComment) TYPE(MacroBlockBegin ) TYPE(MacroBlockEnd) TYPE(ModulePartitionColon) TYPE(NamespaceMacro ) TYPE(NonNullAssertion) TYPE(NullCoalescingEqual) TYPE(NullCoalescingOperator ) TYPE(NullPropagatingOperator) TYPE(ObjCBlockLBrace) TYPE(ObjCBlockLParen ) TYPE(ObjCDecl) TYPE(ObjCForIn) TYPE(ObjCMethodExpr) TYPE(ObjCMethodSpecifier ) TYPE(ObjCProperty) TYPE(ObjCStringLiteral) TYPE(OverloadedOperator ) TYPE(OverloadedOperatorLParen) TYPE(PointerOrReference) TYPE (PureVirtualSpecifier) TYPE(RangeBasedForLoopColon) TYPE(RegexLiteral ) TYPE(SelectorName) TYPE(StartOfName) TYPE(StatementAttributeLikeMacro ) TYPE(StatementMacro) TYPE(StructuredBindingLSquare) TYPE(TemplateCloser ) TYPE(TemplateOpener) TYPE(TemplateString) TYPE(ProtoExtensionLSquare ) TYPE(TrailingAnnotation) TYPE(TrailingReturnArrow) TYPE(TrailingUnaryOperator ) TYPE(TypeDeclarationParen) TYPE(TypenameMacro) TYPE(UnaryOperator ) TYPE(UntouchableMacroFunc) TYPE(CSharpStringLiteral) TYPE(CSharpNamedArgumentColon ) TYPE(CSharpNullable) TYPE(CSharpNullConditionalLSquare) TYPE (CSharpGenericTypeConstraint) TYPE(CSharpGenericTypeConstraintColon ) TYPE(CSharpGenericTypeConstraintComma) TYPE(Unknown) \ |
| 29 | TYPE(ArrayInitializerLSquare) \ |
| 30 | TYPE(ArraySubscriptLSquare) \ |
| 31 | TYPE(AttributeColon) \ |
| 32 | TYPE(AttributeMacro) \ |
| 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(ConstraintJunctions) \ |
| 44 | TYPE(CtorInitializerColon) \ |
| 45 | TYPE(CtorInitializerComma) \ |
| 46 | TYPE(DesignatedInitializerLSquare) \ |
| 47 | TYPE(DesignatedInitializerPeriod) \ |
| 48 | TYPE(DictLiteral) \ |
| 49 | TYPE(FatArrow) \ |
| 50 | TYPE(ForEachMacro) \ |
| 51 | TYPE(FunctionAnnotationRParen) \ |
| 52 | TYPE(FunctionDeclarationName) \ |
| 53 | TYPE(FunctionLBrace) \ |
| 54 | TYPE(FunctionTypeLParen) \ |
| 55 | TYPE(IfMacro) \ |
| 56 | TYPE(ImplicitStringLiteral) \ |
| 57 | TYPE(InheritanceColon) \ |
| 58 | TYPE(InheritanceComma) \ |
| 59 | TYPE(InlineASMBrace) \ |
| 60 | TYPE(InlineASMColon) \ |
| 61 | TYPE(InlineASMSymbolicNameLSquare) \ |
| 62 | TYPE(JavaAnnotation) \ |
| 63 | TYPE(JsComputedPropertyName) \ |
| 64 | TYPE(JsExponentiation) \ |
| 65 | TYPE(JsExponentiationEqual) \ |
| 66 | TYPE(JsPipePipeEqual) \ |
| 67 | TYPE(JsPrivateIdentifier) \ |
| 68 | TYPE(JsTypeColon) \ |
| 69 | TYPE(JsTypeOperator) \ |
| 70 | TYPE(JsTypeOptionalQuestion) \ |
| 71 | TYPE(JsAndAndEqual) \ |
| 72 | TYPE(LambdaArrow) \ |
| 73 | TYPE(LambdaLBrace) \ |
| 74 | TYPE(LambdaLSquare) \ |
| 75 | TYPE(LeadingJavaAnnotation) \ |
| 76 | TYPE(LineComment) \ |
| 77 | TYPE(MacroBlockBegin) \ |
| 78 | TYPE(MacroBlockEnd) \ |
| 79 | TYPE(ModulePartitionColon) \ |
| 80 | TYPE(NamespaceMacro) \ |
| 81 | TYPE(NonNullAssertion) \ |
| 82 | TYPE(NullCoalescingEqual) \ |
| 83 | TYPE(NullCoalescingOperator) \ |
| 84 | TYPE(NullPropagatingOperator) \ |
| 85 | TYPE(ObjCBlockLBrace) \ |
| 86 | TYPE(ObjCBlockLParen) \ |
| 87 | TYPE(ObjCDecl) \ |
| 88 | TYPE(ObjCForIn) \ |
| 89 | TYPE(ObjCMethodExpr) \ |
| 90 | TYPE(ObjCMethodSpecifier) \ |
| 91 | TYPE(ObjCProperty) \ |
| 92 | TYPE(ObjCStringLiteral) \ |
| 93 | TYPE(OverloadedOperator) \ |
| 94 | TYPE(OverloadedOperatorLParen) \ |
| 95 | TYPE(PointerOrReference) \ |
| 96 | TYPE(PureVirtualSpecifier) \ |
| 97 | TYPE(RangeBasedForLoopColon) \ |
| 98 | TYPE(RegexLiteral) \ |
| 99 | TYPE(SelectorName) \ |
| 100 | TYPE(StartOfName) \ |
| 101 | TYPE(StatementAttributeLikeMacro) \ |
| 102 | TYPE(StatementMacro) \ |
| 103 | TYPE(StructuredBindingLSquare) \ |
| 104 | TYPE(TemplateCloser) \ |
| 105 | TYPE(TemplateOpener) \ |
| 106 | TYPE(TemplateString) \ |
| 107 | TYPE(ProtoExtensionLSquare) \ |
| 108 | TYPE(TrailingAnnotation) \ |
| 109 | TYPE(TrailingReturnArrow) \ |
| 110 | TYPE(TrailingUnaryOperator) \ |
| 111 | TYPE(TypeDeclarationParen) \ |
| 112 | TYPE(TypenameMacro) \ |
| 113 | TYPE(UnaryOperator) \ |
| 114 | TYPE(UntouchableMacroFunc) \ |
| 115 | TYPE(CSharpStringLiteral) \ |
| 116 | TYPE(CSharpNamedArgumentColon) \ |
| 117 | TYPE(CSharpNullable) \ |
| 118 | TYPE(CSharpNullConditionalLSquare) \ |
| 119 | TYPE(CSharpGenericTypeConstraint) \ |
| 120 | TYPE(CSharpGenericTypeConstraintColon) \ |
| 121 | TYPE(CSharpGenericTypeConstraintComma) \ |
| 122 | TYPE(Unknown) |
| 123 | |
| 124 | /// Determines the semantic type of a syntactic token, e.g. whether "<" is a |
| 125 | /// template opener or binary operator. |
| 126 | enum TokenType : uint8_t { |
| 127 | #define TYPE(X) TT_##X, |
| 128 | LIST_TOKEN_TYPESTYPE(ArrayInitializerLSquare) TYPE(ArraySubscriptLSquare) TYPE (AttributeColon) TYPE(AttributeMacro) TYPE(AttributeParen) TYPE (AttributeSquare) TYPE(BinaryOperator) TYPE(BitFieldColon) TYPE (BlockComment) TYPE(CastRParen) TYPE(ConditionalExpr) TYPE(ConflictAlternative ) TYPE(ConflictEnd) TYPE(ConflictStart) TYPE(ConstraintJunctions ) TYPE(CtorInitializerColon) TYPE(CtorInitializerComma) TYPE( DesignatedInitializerLSquare) TYPE(DesignatedInitializerPeriod ) TYPE(DictLiteral) TYPE(FatArrow) TYPE(ForEachMacro) TYPE(FunctionAnnotationRParen ) TYPE(FunctionDeclarationName) TYPE(FunctionLBrace) TYPE(FunctionTypeLParen ) TYPE(IfMacro) TYPE(ImplicitStringLiteral) TYPE(InheritanceColon ) TYPE(InheritanceComma) TYPE(InlineASMBrace) TYPE(InlineASMColon ) TYPE(InlineASMSymbolicNameLSquare) TYPE(JavaAnnotation) TYPE (JsComputedPropertyName) TYPE(JsExponentiation) TYPE(JsExponentiationEqual ) TYPE(JsPipePipeEqual) TYPE(JsPrivateIdentifier) TYPE(JsTypeColon ) TYPE(JsTypeOperator) TYPE(JsTypeOptionalQuestion) TYPE(JsAndAndEqual ) TYPE(LambdaArrow) TYPE(LambdaLBrace) TYPE(LambdaLSquare) TYPE (LeadingJavaAnnotation) TYPE(LineComment) TYPE(MacroBlockBegin ) TYPE(MacroBlockEnd) TYPE(ModulePartitionColon) TYPE(NamespaceMacro ) TYPE(NonNullAssertion) TYPE(NullCoalescingEqual) TYPE(NullCoalescingOperator ) TYPE(NullPropagatingOperator) TYPE(ObjCBlockLBrace) TYPE(ObjCBlockLParen ) TYPE(ObjCDecl) TYPE(ObjCForIn) TYPE(ObjCMethodExpr) TYPE(ObjCMethodSpecifier ) TYPE(ObjCProperty) TYPE(ObjCStringLiteral) TYPE(OverloadedOperator ) TYPE(OverloadedOperatorLParen) TYPE(PointerOrReference) TYPE (PureVirtualSpecifier) TYPE(RangeBasedForLoopColon) TYPE(RegexLiteral ) TYPE(SelectorName) TYPE(StartOfName) TYPE(StatementAttributeLikeMacro ) TYPE(StatementMacro) TYPE(StructuredBindingLSquare) TYPE(TemplateCloser ) TYPE(TemplateOpener) TYPE(TemplateString) TYPE(ProtoExtensionLSquare ) TYPE(TrailingAnnotation) TYPE(TrailingReturnArrow) TYPE(TrailingUnaryOperator ) TYPE(TypeDeclarationParen) TYPE(TypenameMacro) TYPE(UnaryOperator ) TYPE(UntouchableMacroFunc) TYPE(CSharpStringLiteral) TYPE(CSharpNamedArgumentColon ) TYPE(CSharpNullable) TYPE(CSharpNullConditionalLSquare) TYPE (CSharpGenericTypeConstraint) TYPE(CSharpGenericTypeConstraintColon ) TYPE(CSharpGenericTypeConstraintComma) TYPE(Unknown) |
| 129 | #undef TYPE |
| 130 | NUM_TOKEN_TYPES |
| 131 | }; |
| 132 | |
| 133 | /// Determines the name of a token type. |
| 134 | const char *getTokenTypeName(TokenType Type); |
| 135 | |
| 136 | // Represents what type of block a set of braces open. |
| 137 | enum BraceBlockKind { BK_Unknown, BK_Block, BK_BracedInit }; |
| 138 | |
| 139 | // The packing kind of a function's parameters. |
| 140 | enum ParameterPackingKind { PPK_BinPacked, PPK_OnePerLine, PPK_Inconclusive }; |
| 141 | |
| 142 | enum FormatDecision { FD_Unformatted, FD_Continue, FD_Break }; |
| 143 | |
| 144 | /// Roles a token can take in a configured macro expansion. |
| 145 | enum MacroRole { |
| 146 | /// The token was expanded from a macro argument when formatting the expanded |
| 147 | /// token sequence. |
| 148 | MR_ExpandedArg, |
| 149 | /// The token is part of a macro argument that was previously formatted as |
| 150 | /// expansion when formatting the unexpanded macro call. |
| 151 | MR_UnexpandedArg, |
| 152 | /// The token was expanded from a macro definition, and is not visible as part |
| 153 | /// of the macro call. |
| 154 | MR_Hidden, |
| 155 | }; |
| 156 | |
| 157 | struct FormatToken; |
| 158 | |
| 159 | /// Contains information on the token's role in a macro expansion. |
| 160 | /// |
| 161 | /// Given the following definitions: |
| 162 | /// A(X) = [ X ] |
| 163 | /// B(X) = < X > |
| 164 | /// C(X) = X |
| 165 | /// |
| 166 | /// Consider the macro call: |
| 167 | /// A({B(C(C(x)))}) -> [{<x>}] |
| 168 | /// |
| 169 | /// In this case, the tokens of the unexpanded macro call will have the |
| 170 | /// following relevant entries in their macro context (note that formatting |
| 171 | /// the unexpanded macro call happens *after* formatting the expanded macro |
| 172 | /// call): |
| 173 | /// A( { B( C( C(x) ) ) } ) |
| 174 | /// Role: NN U NN NN NNUN N N U N (N=None, U=UnexpandedArg) |
| 175 | /// |
| 176 | /// [ { < x > } ] |
| 177 | /// Role: H E H E H E H (H=Hidden, E=ExpandedArg) |
| 178 | /// ExpandedFrom[0]: A A A A A A A |
| 179 | /// ExpandedFrom[1]: B B B |
| 180 | /// ExpandedFrom[2]: C |
| 181 | /// ExpandedFrom[3]: C |
| 182 | /// StartOfExpansion: 1 0 1 2 0 0 0 |
| 183 | /// EndOfExpansion: 0 0 0 2 1 0 1 |
| 184 | struct MacroExpansion { |
| 185 | MacroExpansion(MacroRole Role) : Role(Role) {} |
| 186 | |
| 187 | /// The token's role in the macro expansion. |
| 188 | /// When formatting an expanded macro, all tokens that are part of macro |
| 189 | /// arguments will be MR_ExpandedArg, while all tokens that are not visible in |
| 190 | /// the macro call will be MR_Hidden. |
| 191 | /// When formatting an unexpanded macro call, all tokens that are part of |
| 192 | /// macro arguments will be MR_UnexpandedArg. |
| 193 | MacroRole Role; |
| 194 | |
| 195 | /// The stack of macro call identifier tokens this token was expanded from. |
| 196 | llvm::SmallVector<FormatToken *, 1> ExpandedFrom; |
| 197 | |
| 198 | /// The number of expansions of which this macro is the first entry. |
| 199 | unsigned StartOfExpansion = 0; |
| 200 | |
| 201 | /// The number of currently open expansions in \c ExpandedFrom this macro is |
| 202 | /// the last token in. |
| 203 | unsigned EndOfExpansion = 0; |
| 204 | }; |
| 205 | |
| 206 | class TokenRole; |
| 207 | class AnnotatedLine; |
| 208 | |
| 209 | /// A wrapper around a \c Token storing information about the |
| 210 | /// whitespace characters preceding it. |
| 211 | struct FormatToken { |
| 212 | FormatToken() |
| 213 | : HasUnescapedNewline(false), IsMultiline(false), IsFirst(false), |
| 214 | MustBreakBefore(false), IsUnterminatedLiteral(false), |
| 215 | CanBreakBefore(false), ClosesTemplateDeclaration(false), |
| 216 | StartsBinaryExpression(false), EndsBinaryExpression(false), |
| 217 | PartOfMultiVariableDeclStmt(false), ContinuesLineCommentSection(false), |
| 218 | Finalized(false), BlockKind(BK_Unknown), Decision(FD_Unformatted), |
| 219 | PackingKind(PPK_Inconclusive), Type(TT_Unknown) {} |
| 220 | |
| 221 | /// The \c Token. |
| 222 | Token Tok; |
| 223 | |
| 224 | /// The raw text of the token. |
| 225 | /// |
| 226 | /// Contains the raw token text without leading whitespace and without leading |
| 227 | /// escaped newlines. |
| 228 | StringRef TokenText; |
| 229 | |
| 230 | /// A token can have a special role that can carry extra information |
| 231 | /// about the token's formatting. |
| 232 | /// FIXME: Make FormatToken for parsing and AnnotatedToken two different |
| 233 | /// classes and make this a unique_ptr in the AnnotatedToken class. |
| 234 | std::shared_ptr<TokenRole> Role; |
| 235 | |
| 236 | /// The range of the whitespace immediately preceding the \c Token. |
| 237 | SourceRange WhitespaceRange; |
| 238 | |
| 239 | /// Whether there is at least one unescaped newline before the \c |
| 240 | /// Token. |
| 241 | unsigned HasUnescapedNewline : 1; |
| 242 | |
| 243 | /// Whether the token text contains newlines (escaped or not). |
| 244 | unsigned IsMultiline : 1; |
| 245 | |
| 246 | /// Indicates that this is the first token of the file. |
| 247 | unsigned IsFirst : 1; |
| 248 | |
| 249 | /// Whether there must be a line break before this token. |
| 250 | /// |
| 251 | /// This happens for example when a preprocessor directive ended directly |
| 252 | /// before the token. |
| 253 | unsigned MustBreakBefore : 1; |
| 254 | |
| 255 | /// Set to \c true if this token is an unterminated literal. |
| 256 | unsigned IsUnterminatedLiteral : 1; |
| 257 | |
| 258 | /// \c true if it is allowed to break before this token. |
| 259 | unsigned CanBreakBefore : 1; |
| 260 | |
| 261 | /// \c true if this is the ">" of "template<..>". |
| 262 | unsigned ClosesTemplateDeclaration : 1; |
| 263 | |
| 264 | /// \c true if this token starts a binary expression, i.e. has at least |
| 265 | /// one fake l_paren with a precedence greater than prec::Unknown. |
| 266 | unsigned StartsBinaryExpression : 1; |
| 267 | /// \c true if this token ends a binary expression. |
| 268 | unsigned EndsBinaryExpression : 1; |
| 269 | |
| 270 | /// Is this token part of a \c DeclStmt defining multiple variables? |
| 271 | /// |
| 272 | /// Only set if \c Type == \c TT_StartOfName. |
| 273 | unsigned PartOfMultiVariableDeclStmt : 1; |
| 274 | |
| 275 | /// Does this line comment continue a line comment section? |
| 276 | /// |
| 277 | /// Only set to true if \c Type == \c TT_LineComment. |
| 278 | unsigned ContinuesLineCommentSection : 1; |
| 279 | |
| 280 | /// If \c true, this token has been fully formatted (indented and |
| 281 | /// potentially re-formatted inside), and we do not allow further formatting |
| 282 | /// changes. |
| 283 | unsigned Finalized : 1; |
| 284 | |
| 285 | private: |
| 286 | /// Contains the kind of block if this token is a brace. |
| 287 | unsigned BlockKind : 2; |
| 288 | |
| 289 | public: |
| 290 | BraceBlockKind getBlockKind() const { |
| 291 | return static_cast<BraceBlockKind>(BlockKind); |
| 292 | } |
| 293 | void setBlockKind(BraceBlockKind BBK) { |
| 294 | BlockKind = BBK; |
| 295 | assert(getBlockKind() == BBK && "BraceBlockKind overflow!")(static_cast <bool> (getBlockKind() == BBK && "BraceBlockKind overflow!" ) ? void (0) : __assert_fail ("getBlockKind() == BBK && \"BraceBlockKind overflow!\"" , "clang/lib/Format/FormatToken.h", 295, __extension__ __PRETTY_FUNCTION__ )); |
| 296 | } |
| 297 | |
| 298 | private: |
| 299 | /// Stores the formatting decision for the token once it was made. |
| 300 | unsigned Decision : 2; |
| 301 | |
| 302 | public: |
| 303 | FormatDecision getDecision() const { |
| 304 | return static_cast<FormatDecision>(Decision); |
| 305 | } |
| 306 | void setDecision(FormatDecision D) { |
| 307 | Decision = D; |
| 308 | assert(getDecision() == D && "FormatDecision overflow!")(static_cast <bool> (getDecision() == D && "FormatDecision overflow!" ) ? void (0) : __assert_fail ("getDecision() == D && \"FormatDecision overflow!\"" , "clang/lib/Format/FormatToken.h", 308, __extension__ __PRETTY_FUNCTION__ )); |
| 309 | } |
| 310 | |
| 311 | private: |
| 312 | /// If this is an opening parenthesis, how are the parameters packed? |
| 313 | unsigned PackingKind : 2; |
| 314 | |
| 315 | public: |
| 316 | ParameterPackingKind getPackingKind() const { |
| 317 | return static_cast<ParameterPackingKind>(PackingKind); |
| 318 | } |
| 319 | void setPackingKind(ParameterPackingKind K) { |
| 320 | PackingKind = K; |
| 321 | assert(getPackingKind() == K && "ParameterPackingKind overflow!")(static_cast <bool> (getPackingKind() == K && "ParameterPackingKind overflow!" ) ? void (0) : __assert_fail ("getPackingKind() == K && \"ParameterPackingKind overflow!\"" , "clang/lib/Format/FormatToken.h", 321, __extension__ __PRETTY_FUNCTION__ )); |
| 322 | } |
| 323 | |
| 324 | private: |
| 325 | TokenType Type; |
| 326 | |
| 327 | public: |
| 328 | /// Returns the token's type, e.g. whether "<" is a template opener or |
| 329 | /// binary operator. |
| 330 | TokenType getType() const { return Type; } |
| 331 | void setType(TokenType T) { Type = T; } |
| 332 | |
| 333 | /// The number of newlines immediately before the \c Token. |
| 334 | /// |
| 335 | /// This can be used to determine what the user wrote in the original code |
| 336 | /// and thereby e.g. leave an empty line between two function definitions. |
| 337 | unsigned NewlinesBefore = 0; |
| 338 | |
| 339 | /// The offset just past the last '\n' in this token's leading |
| 340 | /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'. |
| 341 | unsigned LastNewlineOffset = 0; |
| 342 | |
| 343 | /// The width of the non-whitespace parts of the token (or its first |
| 344 | /// line for multi-line tokens) in columns. |
| 345 | /// We need this to correctly measure number of columns a token spans. |
| 346 | unsigned ColumnWidth = 0; |
| 347 | |
| 348 | /// Contains the width in columns of the last line of a multi-line |
| 349 | /// token. |
| 350 | unsigned LastLineColumnWidth = 0; |
| 351 | |
| 352 | /// The number of spaces that should be inserted before this token. |
| 353 | unsigned SpacesRequiredBefore = 0; |
| 354 | |
| 355 | /// Number of parameters, if this is "(", "[" or "<". |
| 356 | unsigned ParameterCount = 0; |
| 357 | |
| 358 | /// Number of parameters that are nested blocks, |
| 359 | /// if this is "(", "[" or "<". |
| 360 | unsigned BlockParameterCount = 0; |
| 361 | |
| 362 | /// If this is a bracket ("<", "(", "[" or "{"), contains the kind of |
| 363 | /// the surrounding bracket. |
| 364 | tok::TokenKind ParentBracket = tok::unknown; |
| 365 | |
| 366 | /// The total length of the unwrapped line up to and including this |
| 367 | /// token. |
| 368 | unsigned TotalLength = 0; |
| 369 | |
| 370 | /// The original 0-based column of this token, including expanded tabs. |
| 371 | /// The configured TabWidth is used as tab width. |
| 372 | unsigned OriginalColumn = 0; |
| 373 | |
| 374 | /// The length of following tokens until the next natural split point, |
| 375 | /// or the next token that can be broken. |
| 376 | unsigned UnbreakableTailLength = 0; |
| 377 | |
| 378 | // FIXME: Come up with a 'cleaner' concept. |
| 379 | /// The binding strength of a token. This is a combined value of |
| 380 | /// operator precedence, parenthesis nesting, etc. |
| 381 | unsigned BindingStrength = 0; |
| 382 | |
| 383 | /// The nesting level of this token, i.e. the number of surrounding (), |
| 384 | /// [], {} or <>. |
| 385 | unsigned NestingLevel = 0; |
| 386 | |
| 387 | /// The indent level of this token. Copied from the surrounding line. |
| 388 | unsigned IndentLevel = 0; |
| 389 | |
| 390 | /// Penalty for inserting a line break before this token. |
| 391 | unsigned SplitPenalty = 0; |
| 392 | |
| 393 | /// If this is the first ObjC selector name in an ObjC method |
| 394 | /// definition or call, this contains the length of the longest name. |
| 395 | /// |
| 396 | /// This being set to 0 means that the selectors should not be colon-aligned, |
| 397 | /// e.g. because several of them are block-type. |
| 398 | unsigned LongestObjCSelectorName = 0; |
| 399 | |
| 400 | /// If this is the first ObjC selector name in an ObjC method |
| 401 | /// definition or call, this contains the number of parts that the whole |
| 402 | /// selector consist of. |
| 403 | unsigned ObjCSelectorNameParts = 0; |
| 404 | |
| 405 | /// The 0-based index of the parameter/argument. For ObjC it is set |
| 406 | /// for the selector name token. |
| 407 | /// For now calculated only for ObjC. |
| 408 | unsigned ParameterIndex = 0; |
| 409 | |
| 410 | /// Stores the number of required fake parentheses and the |
| 411 | /// corresponding operator precedence. |
| 412 | /// |
| 413 | /// If multiple fake parentheses start at a token, this vector stores them in |
| 414 | /// reverse order, i.e. inner fake parenthesis first. |
| 415 | SmallVector<prec::Level, 4> FakeLParens; |
| 416 | /// Insert this many fake ) after this token for correct indentation. |
| 417 | unsigned FakeRParens = 0; |
| 418 | |
| 419 | /// If this is an operator (or "."/"->") in a sequence of operators |
| 420 | /// with the same precedence, contains the 0-based operator index. |
| 421 | unsigned OperatorIndex = 0; |
| 422 | |
| 423 | /// If this is an operator (or "."/"->") in a sequence of operators |
| 424 | /// with the same precedence, points to the next operator. |
| 425 | FormatToken *NextOperator = nullptr; |
| 426 | |
| 427 | /// If this is a bracket, this points to the matching one. |
| 428 | FormatToken *MatchingParen = nullptr; |
| 429 | |
| 430 | /// The previous token in the unwrapped line. |
| 431 | FormatToken *Previous = nullptr; |
| 432 | |
| 433 | /// The next token in the unwrapped line. |
| 434 | FormatToken *Next = nullptr; |
| 435 | |
| 436 | /// The first token in set of column elements. |
| 437 | bool StartsColumn = false; |
| 438 | |
| 439 | /// This notes the start of the line of an array initializer. |
| 440 | bool ArrayInitializerLineStart = false; |
| 441 | |
| 442 | /// This starts an array initializer. |
| 443 | bool IsArrayInitializer = false; |
| 444 | |
| 445 | /// Is optional and can be removed. |
| 446 | bool Optional = false; |
| 447 | |
| 448 | /// If this token starts a block, this contains all the unwrapped lines |
| 449 | /// in it. |
| 450 | SmallVector<AnnotatedLine *, 1> Children; |
| 451 | |
| 452 | // Contains all attributes related to how this token takes part |
| 453 | // in a configured macro expansion. |
| 454 | llvm::Optional<MacroExpansion> MacroCtx; |
| 455 | |
| 456 | bool is(tok::TokenKind Kind) const { return Tok.is(Kind); } |
| 457 | bool is(TokenType TT) const { return getType() == TT; } |
| 458 | bool is(const IdentifierInfo *II) const { |
| 459 | return II && II == Tok.getIdentifierInfo(); |
| 460 | } |
| 461 | bool is(tok::PPKeywordKind Kind) const { |
| 462 | return Tok.getIdentifierInfo() && |
| 463 | Tok.getIdentifierInfo()->getPPKeywordID() == Kind; |
| 464 | } |
| 465 | bool is(BraceBlockKind BBK) const { return getBlockKind() == BBK; } |
| 466 | bool is(ParameterPackingKind PPK) const { return getPackingKind() == PPK; } |
| 467 | |
| 468 | template <typename A, typename B> bool isOneOf(A K1, B K2) const { |
| 469 | return is(K1) || is(K2); |
| 470 | } |
| 471 | template <typename A, typename B, typename... Ts> |
| 472 | bool isOneOf(A K1, B K2, Ts... Ks) const { |
| 473 | return is(K1) || isOneOf(K2, Ks...); |
| 474 | } |
| 475 | template <typename T> bool isNot(T Kind) const { return !is(Kind); } |
| 476 | |
| 477 | bool isIf(bool AllowConstexprMacro = true) const { |
| 478 | return is(tok::kw_if) || endsSequence(tok::kw_constexpr, tok::kw_if) || |
| 479 | (endsSequence(tok::identifier, tok::kw_if) && AllowConstexprMacro); |
| 480 | } |
| 481 | |
| 482 | bool closesScopeAfterBlock() const { |
| 483 | if (getBlockKind() == BK_Block) |
| 484 | return true; |
| 485 | if (closesScope()) |
| 486 | return Previous->closesScopeAfterBlock(); |
| 487 | return false; |
| 488 | } |
| 489 | |
| 490 | /// \c true if this token starts a sequence with the given tokens in order, |
| 491 | /// following the ``Next`` pointers, ignoring comments. |
| 492 | template <typename A, typename... Ts> |
| 493 | bool startsSequence(A K1, Ts... Tokens) const { |
| 494 | return startsSequenceInternal(K1, Tokens...); |
| 495 | } |
| 496 | |
| 497 | /// \c true if this token ends a sequence with the given tokens in order, |
| 498 | /// following the ``Previous`` pointers, ignoring comments. |
| 499 | /// For example, given tokens [T1, T2, T3], the function returns true if |
| 500 | /// 3 tokens ending at this (ignoring comments) are [T3, T2, T1]. In other |
| 501 | /// words, the tokens passed to this function need to the reverse of the |
| 502 | /// order the tokens appear in code. |
| 503 | template <typename A, typename... Ts> |
| 504 | bool endsSequence(A K1, Ts... Tokens) const { |
| 505 | return endsSequenceInternal(K1, Tokens...); |
| 506 | } |
| 507 | |
| 508 | bool isStringLiteral() const { return tok::isStringLiteral(Tok.getKind()); } |
| 509 | |
| 510 | bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const { |
| 511 | return Tok.isObjCAtKeyword(Kind); |
| 512 | } |
| 513 | |
| 514 | bool isAccessSpecifier(bool ColonRequired = true) const { |
| 515 | return isOneOf(tok::kw_public, tok::kw_protected, tok::kw_private) && |
| 516 | (!ColonRequired || (Next && Next->is(tok::colon))); |
| 517 | } |
| 518 | |
| 519 | bool canBePointerOrReferenceQualifier() const { |
| 520 | return isOneOf(tok::kw_const, tok::kw_restrict, tok::kw_volatile, |
| 521 | tok::kw___attribute, tok::kw__Nonnull, tok::kw__Nullable, |
| 522 | tok::kw__Null_unspecified, tok::kw___ptr32, tok::kw___ptr64, |
| 523 | TT_AttributeMacro); |
| 524 | } |
| 525 | |
| 526 | /// Determine whether the token is a simple-type-specifier. |
| 527 | LLVM_NODISCARD[[clang::warn_unused_result]] bool isSimpleTypeSpecifier() const; |
| 528 | |
| 529 | LLVM_NODISCARD[[clang::warn_unused_result]] bool isTypeOrIdentifier() const; |
| 530 | |
| 531 | bool isObjCAccessSpecifier() const { |
| 532 | return is(tok::at) && Next && |
| 533 | (Next->isObjCAtKeyword(tok::objc_public) || |
| 534 | Next->isObjCAtKeyword(tok::objc_protected) || |
| 535 | Next->isObjCAtKeyword(tok::objc_package) || |
| 536 | Next->isObjCAtKeyword(tok::objc_private)); |
| 537 | } |
| 538 | |
| 539 | /// Returns whether \p Tok is ([{ or an opening < of a template or in |
| 540 | /// protos. |
| 541 | bool opensScope() const { |
| 542 | if (is(TT_TemplateString) && TokenText.endswith("${")) |
| 543 | return true; |
| 544 | if (is(TT_DictLiteral) && is(tok::less)) |
| 545 | return true; |
| 546 | return isOneOf(tok::l_paren, tok::l_brace, tok::l_square, |
| 547 | TT_TemplateOpener); |
| 548 | } |
| 549 | /// Returns whether \p Tok is )]} or a closing > of a template or in |
| 550 | /// protos. |
| 551 | bool closesScope() const { |
| 552 | if (is(TT_TemplateString) && TokenText.startswith("}")) |
| 553 | return true; |
| 554 | if (is(TT_DictLiteral) && is(tok::greater)) |
| 555 | return true; |
| 556 | return isOneOf(tok::r_paren, tok::r_brace, tok::r_square, |
| 557 | TT_TemplateCloser); |
| 558 | } |
| 559 | |
| 560 | /// Returns \c true if this is a "." or "->" accessing a member. |
| 561 | bool isMemberAccess() const { |
| 562 | return isOneOf(tok::arrow, tok::period, tok::arrowstar) && |
| 563 | !isOneOf(TT_DesignatedInitializerPeriod, TT_TrailingReturnArrow, |
| 564 | TT_LambdaArrow, TT_LeadingJavaAnnotation); |
| 565 | } |
| 566 | |
| 567 | bool isUnaryOperator() const { |
| 568 | switch (Tok.getKind()) { |
| 569 | case tok::plus: |
| 570 | case tok::plusplus: |
| 571 | case tok::minus: |
| 572 | case tok::minusminus: |
| 573 | case tok::exclaim: |
| 574 | case tok::tilde: |
| 575 | case tok::kw_sizeof: |
| 576 | case tok::kw_alignof: |
| 577 | return true; |
| 578 | default: |
| 579 | return false; |
| 580 | } |
| 581 | } |
| 582 | |
| 583 | bool isBinaryOperator() const { |
| 584 | // Comma is a binary operator, but does not behave as such wrt. formatting. |
| 585 | return getPrecedence() > prec::Comma; |
| 586 | } |
| 587 | |
| 588 | bool isTrailingComment() const { |
| 589 | return is(tok::comment) && |
| 590 | (is(TT_LineComment) || !Next || Next->NewlinesBefore > 0); |
| 591 | } |
| 592 | |
| 593 | /// Returns \c true if this is a keyword that can be used |
| 594 | /// like a function call (e.g. sizeof, typeid, ...). |
| 595 | bool isFunctionLikeKeyword() const { |
| 596 | switch (Tok.getKind()) { |
| 597 | case tok::kw_throw: |
| 598 | case tok::kw_typeid: |
| 599 | case tok::kw_return: |
| 600 | case tok::kw_sizeof: |
| 601 | case tok::kw_alignof: |
| 602 | case tok::kw_alignas: |
| 603 | case tok::kw_decltype: |
| 604 | case tok::kw_noexcept: |
| 605 | case tok::kw_static_assert: |
| 606 | case tok::kw__Atomic: |
| 607 | case tok::kw___attribute: |
| 608 | case tok::kw___underlying_type: |
| 609 | case tok::kw_requires: |
| 610 | return true; |
| 611 | default: |
| 612 | return false; |
| 613 | } |
| 614 | } |
| 615 | |
| 616 | /// Returns \c true if this is a string literal that's like a label, |
| 617 | /// e.g. ends with "=" or ":". |
| 618 | bool isLabelString() const { |
| 619 | if (!is(tok::string_literal)) |
| 620 | return false; |
| 621 | StringRef Content = TokenText; |
| 622 | if (Content.startswith("\"") || Content.startswith("'")) |
| 623 | Content = Content.drop_front(1); |
| 624 | if (Content.endswith("\"") || Content.endswith("'")) |
| 625 | Content = Content.drop_back(1); |
| 626 | Content = Content.trim(); |
| 627 | return Content.size() > 1 && |
| 628 | (Content.back() == ':' || Content.back() == '='); |
| 629 | } |
| 630 | |
| 631 | /// Returns actual token start location without leading escaped |
| 632 | /// newlines and whitespace. |
| 633 | /// |
| 634 | /// This can be different to Tok.getLocation(), which includes leading escaped |
| 635 | /// newlines. |
| 636 | SourceLocation getStartOfNonWhitespace() const { |
| 637 | return WhitespaceRange.getEnd(); |
| 638 | } |
| 639 | |
| 640 | prec::Level getPrecedence() const { |
| 641 | return getBinOpPrecedence(Tok.getKind(), /*GreaterThanIsOperator=*/true, |
| 642 | /*CPlusPlus11=*/true); |
| 643 | } |
| 644 | |
| 645 | /// Returns the previous token ignoring comments. |
| 646 | FormatToken *getPreviousNonComment() const { |
| 647 | FormatToken *Tok = Previous; |
| 648 | while (Tok && Tok->is(tok::comment)) |
| 649 | Tok = Tok->Previous; |
| 650 | return Tok; |
| 651 | } |
| 652 | |
| 653 | /// Returns the next token ignoring comments. |
| 654 | const FormatToken *getNextNonComment() const { |
| 655 | const FormatToken *Tok = Next; |
| 656 | while (Tok && Tok->is(tok::comment)) |
| 657 | Tok = Tok->Next; |
| 658 | return Tok; |
| 659 | } |
| 660 | |
| 661 | /// Returns \c true if this tokens starts a block-type list, i.e. a |
| 662 | /// list that should be indented with a block indent. |
| 663 | bool opensBlockOrBlockTypeList(const FormatStyle &Style) const { |
| 664 | // C# Does not indent object initialisers as continuations. |
| 665 | if (is(tok::l_brace) && getBlockKind() == BK_BracedInit && Style.isCSharp()) |
| 666 | return true; |
| 667 | if (is(TT_TemplateString) && opensScope()) |
| 668 | return true; |
| 669 | return is(TT_ArrayInitializerLSquare) || is(TT_ProtoExtensionLSquare) || |
| 670 | (is(tok::l_brace) && |
| 671 | (getBlockKind() == BK_Block || is(TT_DictLiteral) || |
| 672 | (!Style.Cpp11BracedListStyle && NestingLevel == 0))) || |
| 673 | (is(tok::less) && (Style.Language == FormatStyle::LK_Proto || |
| 674 | Style.Language == FormatStyle::LK_TextProto)); |
| 675 | } |
| 676 | |
| 677 | /// Returns whether the token is the left square bracket of a C++ |
| 678 | /// structured binding declaration. |
| 679 | bool isCppStructuredBinding(const FormatStyle &Style) const { |
| 680 | if (!Style.isCpp() || isNot(tok::l_square)) |
| 681 | return false; |
| 682 | const FormatToken *T = this; |
| 683 | do { |
| 684 | T = T->getPreviousNonComment(); |
| 685 | } while (T && T->isOneOf(tok::kw_const, tok::kw_volatile, tok::amp, |
| 686 | tok::ampamp)); |
| 687 | return T && T->is(tok::kw_auto); |
| 688 | } |
| 689 | |
| 690 | /// Same as opensBlockOrBlockTypeList, but for the closing token. |
| 691 | bool closesBlockOrBlockTypeList(const FormatStyle &Style) const { |
| 692 | if (is(TT_TemplateString) && closesScope()) |
| 693 | return true; |
| 694 | return MatchingParen && MatchingParen->opensBlockOrBlockTypeList(Style); |
| 695 | } |
| 696 | |
| 697 | /// Return the actual namespace token, if this token starts a namespace |
| 698 | /// block. |
| 699 | const FormatToken *getNamespaceToken() const { |
| 700 | const FormatToken *NamespaceTok = this; |
| 701 | if (is(tok::comment)) |
| 702 | NamespaceTok = NamespaceTok->getNextNonComment(); |
| 703 | // Detect "(inline|export)? namespace" in the beginning of a line. |
| 704 | if (NamespaceTok && NamespaceTok->isOneOf(tok::kw_inline, tok::kw_export)) |
| 705 | NamespaceTok = NamespaceTok->getNextNonComment(); |
| 706 | return NamespaceTok && |
| 707 | NamespaceTok->isOneOf(tok::kw_namespace, TT_NamespaceMacro) |
| 708 | ? NamespaceTok |
| 709 | : nullptr; |
| 710 | } |
| 711 | |
| 712 | void copyFrom(const FormatToken &Tok) { *this = Tok; } |
| 713 | |
| 714 | private: |
| 715 | // Only allow copying via the explicit copyFrom method. |
| 716 | FormatToken(const FormatToken &) = delete; |
| 717 | FormatToken &operator=(const FormatToken &) = default; |
| 718 | |
| 719 | template <typename A, typename... Ts> |
| 720 | bool startsSequenceInternal(A K1, Ts... Tokens) const { |
| 721 | if (is(tok::comment) && Next) |
| 722 | return Next->startsSequenceInternal(K1, Tokens...); |
| 723 | return is(K1) && Next && Next->startsSequenceInternal(Tokens...); |
| 724 | } |
| 725 | |
| 726 | template <typename A> bool startsSequenceInternal(A K1) const { |
| 727 | if (is(tok::comment) && Next) |
| 728 | return Next->startsSequenceInternal(K1); |
| 729 | return is(K1); |
| 730 | } |
| 731 | |
| 732 | template <typename A, typename... Ts> bool endsSequenceInternal(A K1) const { |
| 733 | if (is(tok::comment) && Previous) |
| 734 | return Previous->endsSequenceInternal(K1); |
| 735 | return is(K1); |
| 736 | } |
| 737 | |
| 738 | template <typename A, typename... Ts> |
| 739 | bool endsSequenceInternal(A K1, Ts... Tokens) const { |
| 740 | if (is(tok::comment) && Previous) |
| 741 | return Previous->endsSequenceInternal(K1, Tokens...); |
| 742 | return is(K1) && Previous && Previous->endsSequenceInternal(Tokens...); |
| 743 | } |
| 744 | }; |
| 745 | |
| 746 | class ContinuationIndenter; |
| 747 | struct LineState; |
| 748 | |
| 749 | class TokenRole { |
| 750 | public: |
| 751 | TokenRole(const FormatStyle &Style) : Style(Style) {} |
| 752 | virtual ~TokenRole(); |
| 753 | |
| 754 | /// After the \c TokenAnnotator has finished annotating all the tokens, |
| 755 | /// this function precomputes required information for formatting. |
| 756 | virtual void precomputeFormattingInfos(const FormatToken *Token); |
| 757 | |
| 758 | /// Apply the special formatting that the given role demands. |
| 759 | /// |
| 760 | /// Assumes that the token having this role is already formatted. |
| 761 | /// |
| 762 | /// Continues formatting from \p State leaving indentation to \p Indenter and |
| 763 | /// returns the total penalty that this formatting incurs. |
| 764 | virtual unsigned formatFromToken(LineState &State, |
| 765 | ContinuationIndenter *Indenter, |
| 766 | bool DryRun) { |
| 767 | return 0; |
| 768 | } |
| 769 | |
| 770 | /// Same as \c formatFromToken, but assumes that the first token has |
| 771 | /// already been set thereby deciding on the first line break. |
| 772 | virtual unsigned formatAfterToken(LineState &State, |
| 773 | ContinuationIndenter *Indenter, |
| 774 | bool DryRun) { |
| 775 | return 0; |
| 776 | } |
| 777 | |
| 778 | /// Notifies the \c Role that a comma was found. |
| 779 | virtual void CommaFound(const FormatToken *Token) {} |
| 780 | |
| 781 | virtual const FormatToken *lastComma() { return nullptr; } |
| 782 | |
| 783 | protected: |
| 784 | const FormatStyle &Style; |
| 785 | }; |
| 786 | |
| 787 | class CommaSeparatedList : public TokenRole { |
| 788 | public: |
| 789 | CommaSeparatedList(const FormatStyle &Style) |
| 790 | : TokenRole(Style), HasNestedBracedList(false) {} |
| 791 | |
| 792 | void precomputeFormattingInfos(const FormatToken *Token) override; |
| 793 | |
| 794 | unsigned formatAfterToken(LineState &State, ContinuationIndenter *Indenter, |
| 795 | bool DryRun) override; |
| 796 | |
| 797 | unsigned formatFromToken(LineState &State, ContinuationIndenter *Indenter, |
| 798 | bool DryRun) override; |
| 799 | |
| 800 | /// Adds \p Token as the next comma to the \c CommaSeparated list. |
| 801 | void CommaFound(const FormatToken *Token) override { |
| 802 | Commas.push_back(Token); |
| 803 | } |
| 804 | |
| 805 | const FormatToken *lastComma() override { |
| 806 | if (Commas.empty()) |
| 807 | return nullptr; |
| 808 | return Commas.back(); |
| 809 | } |
| 810 | |
| 811 | private: |
| 812 | /// A struct that holds information on how to format a given list with |
| 813 | /// a specific number of columns. |
| 814 | struct ColumnFormat { |
| 815 | /// The number of columns to use. |
| 816 | unsigned Columns; |
| 817 | |
| 818 | /// The total width in characters. |
| 819 | unsigned TotalWidth; |
| 820 | |
| 821 | /// The number of lines required for this format. |
| 822 | unsigned LineCount; |
| 823 | |
| 824 | /// The size of each column in characters. |
| 825 | SmallVector<unsigned, 8> ColumnSizes; |
| 826 | }; |
| 827 | |
| 828 | /// Calculate which \c ColumnFormat fits best into |
| 829 | /// \p RemainingCharacters. |
| 830 | const ColumnFormat *getColumnFormat(unsigned RemainingCharacters) const; |
| 831 | |
| 832 | /// The ordered \c FormatTokens making up the commas of this list. |
| 833 | SmallVector<const FormatToken *, 8> Commas; |
| 834 | |
| 835 | /// The length of each of the list's items in characters including the |
| 836 | /// trailing comma. |
| 837 | SmallVector<unsigned, 8> ItemLengths; |
| 838 | |
| 839 | /// Precomputed formats that can be used for this list. |
| 840 | SmallVector<ColumnFormat, 4> Formats; |
| 841 | |
| 842 | bool HasNestedBracedList; |
| 843 | }; |
| 844 | |
| 845 | /// Encapsulates keywords that are context sensitive or for languages not |
| 846 | /// properly supported by Clang's lexer. |
| 847 | struct AdditionalKeywords { |
| 848 | AdditionalKeywords(IdentifierTable &IdentTable) { |
| 849 | kw_final = &IdentTable.get("final"); |
| 850 | kw_override = &IdentTable.get("override"); |
| 851 | kw_in = &IdentTable.get("in"); |
| 852 | kw_of = &IdentTable.get("of"); |
| 853 | kw_CF_CLOSED_ENUM = &IdentTable.get("CF_CLOSED_ENUM"); |
| 854 | kw_CF_ENUM = &IdentTable.get("CF_ENUM"); |
| 855 | kw_CF_OPTIONS = &IdentTable.get("CF_OPTIONS"); |
| 856 | kw_NS_CLOSED_ENUM = &IdentTable.get("NS_CLOSED_ENUM"); |
| 857 | kw_NS_ENUM = &IdentTable.get("NS_ENUM"); |
| 858 | kw_NS_OPTIONS = &IdentTable.get("NS_OPTIONS"); |
| 859 | |
| 860 | kw_as = &IdentTable.get("as"); |
| 861 | kw_async = &IdentTable.get("async"); |
| 862 | kw_await = &IdentTable.get("await"); |
| 863 | kw_declare = &IdentTable.get("declare"); |
| 864 | kw_finally = &IdentTable.get("finally"); |
| 865 | kw_from = &IdentTable.get("from"); |
| 866 | kw_function = &IdentTable.get("function"); |
| 867 | kw_get = &IdentTable.get("get"); |
| 868 | kw_import = &IdentTable.get("import"); |
| 869 | kw_infer = &IdentTable.get("infer"); |
| 870 | kw_is = &IdentTable.get("is"); |
| 871 | kw_let = &IdentTable.get("let"); |
| 872 | kw_module = &IdentTable.get("module"); |
| 873 | kw_readonly = &IdentTable.get("readonly"); |
| 874 | kw_set = &IdentTable.get("set"); |
| 875 | kw_type = &IdentTable.get("type"); |
| 876 | kw_typeof = &IdentTable.get("typeof"); |
| 877 | kw_var = &IdentTable.get("var"); |
| 878 | kw_yield = &IdentTable.get("yield"); |
| 879 | |
| 880 | kw_abstract = &IdentTable.get("abstract"); |
| 881 | kw_assert = &IdentTable.get("assert"); |
| 882 | kw_extends = &IdentTable.get("extends"); |
| 883 | kw_implements = &IdentTable.get("implements"); |
| 884 | kw_instanceof = &IdentTable.get("instanceof"); |
| 885 | kw_interface = &IdentTable.get("interface"); |
| 886 | kw_native = &IdentTable.get("native"); |
| 887 | kw_package = &IdentTable.get("package"); |
| 888 | kw_synchronized = &IdentTable.get("synchronized"); |
| 889 | kw_throws = &IdentTable.get("throws"); |
| 890 | kw___except = &IdentTable.get("__except"); |
| 891 | kw___has_include = &IdentTable.get("__has_include"); |
| 892 | kw___has_include_next = &IdentTable.get("__has_include_next"); |
| 893 | |
| 894 | kw_mark = &IdentTable.get("mark"); |
| 895 | |
| 896 | kw_extend = &IdentTable.get("extend"); |
| 897 | kw_option = &IdentTable.get("option"); |
| 898 | kw_optional = &IdentTable.get("optional"); |
| 899 | kw_repeated = &IdentTable.get("repeated"); |
| 900 | kw_required = &IdentTable.get("required"); |
| 901 | kw_returns = &IdentTable.get("returns"); |
| 902 | |
| 903 | kw_signals = &IdentTable.get("signals"); |
| 904 | kw_qsignals = &IdentTable.get("Q_SIGNALS"); |
| 905 | kw_slots = &IdentTable.get("slots"); |
| 906 | kw_qslots = &IdentTable.get("Q_SLOTS"); |
| 907 | |
| 908 | // C# keywords |
| 909 | kw_dollar = &IdentTable.get("dollar"); |
| 910 | kw_base = &IdentTable.get("base"); |
| 911 | kw_byte = &IdentTable.get("byte"); |
| 912 | kw_checked = &IdentTable.get("checked"); |
| 913 | kw_decimal = &IdentTable.get("decimal"); |
| 914 | kw_delegate = &IdentTable.get("delegate"); |
| 915 | kw_event = &IdentTable.get("event"); |
| 916 | kw_fixed = &IdentTable.get("fixed"); |
| 917 | kw_foreach = &IdentTable.get("foreach"); |
| 918 | kw_implicit = &IdentTable.get("implicit"); |
| 919 | kw_internal = &IdentTable.get("internal"); |
| 920 | kw_lock = &IdentTable.get("lock"); |
| 921 | kw_null = &IdentTable.get("null"); |
| 922 | kw_object = &IdentTable.get("object"); |
| 923 | kw_out = &IdentTable.get("out"); |
| 924 | kw_params = &IdentTable.get("params"); |
| 925 | kw_ref = &IdentTable.get("ref"); |
| 926 | kw_string = &IdentTable.get("string"); |
| 927 | kw_stackalloc = &IdentTable.get("stackalloc"); |
| 928 | kw_sbyte = &IdentTable.get("sbyte"); |
| 929 | kw_sealed = &IdentTable.get("sealed"); |
| 930 | kw_uint = &IdentTable.get("uint"); |
| 931 | kw_ulong = &IdentTable.get("ulong"); |
| 932 | kw_unchecked = &IdentTable.get("unchecked"); |
| 933 | kw_unsafe = &IdentTable.get("unsafe"); |
| 934 | kw_ushort = &IdentTable.get("ushort"); |
| 935 | kw_when = &IdentTable.get("when"); |
| 936 | kw_where = &IdentTable.get("where"); |
| 937 | |
| 938 | // Keep this at the end of the constructor to make sure everything here |
| 939 | // is |
| 940 | // already initialized. |
| 941 | JsExtraKeywords = std::unordered_set<IdentifierInfo *>( |
| 942 | {kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from, |
| 943 | kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_override, |
| 944 | kw_readonly, kw_set, kw_type, kw_typeof, kw_var, kw_yield, |
| 945 | // Keywords from the Java section. |
| 946 | kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface}); |
| 947 | |
| 948 | CSharpExtraKeywords = std::unordered_set<IdentifierInfo *>( |
| 949 | {kw_base, kw_byte, kw_checked, kw_decimal, kw_delegate, kw_event, |
| 950 | kw_fixed, kw_foreach, kw_implicit, kw_in, kw_interface, kw_internal, |
| 951 | kw_is, kw_lock, kw_null, kw_object, kw_out, kw_override, kw_params, |
| 952 | kw_readonly, kw_ref, kw_string, kw_stackalloc, kw_sbyte, kw_sealed, |
| 953 | kw_uint, kw_ulong, kw_unchecked, kw_unsafe, kw_ushort, kw_when, |
| 954 | kw_where, |
| 955 | // Keywords from the JavaScript section. |
| 956 | kw_as, kw_async, kw_await, kw_declare, kw_finally, kw_from, |
| 957 | kw_function, kw_get, kw_import, kw_is, kw_let, kw_module, kw_readonly, |
| 958 | kw_set, kw_type, kw_typeof, kw_var, kw_yield, |
| 959 | // Keywords from the Java section. |
| 960 | kw_abstract, kw_extends, kw_implements, kw_instanceof, kw_interface}); |
| 961 | } |
| 962 | |
| 963 | // Context sensitive keywords. |
| 964 | IdentifierInfo *kw_final; |
| 965 | IdentifierInfo *kw_override; |
| 966 | IdentifierInfo *kw_in; |
| 967 | IdentifierInfo *kw_of; |
| 968 | IdentifierInfo *kw_CF_CLOSED_ENUM; |
| 969 | IdentifierInfo *kw_CF_ENUM; |
| 970 | IdentifierInfo *kw_CF_OPTIONS; |
| 971 | IdentifierInfo *kw_NS_CLOSED_ENUM; |
| 972 | IdentifierInfo *kw_NS_ENUM; |
| 973 | IdentifierInfo *kw_NS_OPTIONS; |
| 974 | IdentifierInfo *kw___except; |
| 975 | IdentifierInfo *kw___has_include; |
| 976 | IdentifierInfo *kw___has_include_next; |
| 977 | |
| 978 | // JavaScript keywords. |
| 979 | IdentifierInfo *kw_as; |
| 980 | IdentifierInfo *kw_async; |
| 981 | IdentifierInfo *kw_await; |
| 982 | IdentifierInfo *kw_declare; |
| 983 | IdentifierInfo *kw_finally; |
| 984 | IdentifierInfo *kw_from; |
| 985 | IdentifierInfo *kw_function; |
| 986 | IdentifierInfo *kw_get; |
| 987 | IdentifierInfo *kw_import; |
| 988 | IdentifierInfo *kw_infer; |
| 989 | IdentifierInfo *kw_is; |
| 990 | IdentifierInfo *kw_let; |
| 991 | IdentifierInfo *kw_module; |
| 992 | IdentifierInfo *kw_readonly; |
| 993 | IdentifierInfo *kw_set; |
| 994 | IdentifierInfo *kw_type; |
| 995 | IdentifierInfo *kw_typeof; |
| 996 | IdentifierInfo *kw_var; |
| 997 | IdentifierInfo *kw_yield; |
| 998 | |
| 999 | // Java keywords. |
| 1000 | IdentifierInfo *kw_abstract; |
| 1001 | IdentifierInfo *kw_assert; |
| 1002 | IdentifierInfo *kw_extends; |
| 1003 | IdentifierInfo *kw_implements; |
| 1004 | IdentifierInfo *kw_instanceof; |
| 1005 | IdentifierInfo *kw_interface; |
| 1006 | IdentifierInfo *kw_native; |
| 1007 | IdentifierInfo *kw_package; |
| 1008 | IdentifierInfo *kw_synchronized; |
| 1009 | IdentifierInfo *kw_throws; |
| 1010 | |
| 1011 | // Pragma keywords. |
| 1012 | IdentifierInfo *kw_mark; |
| 1013 | |
| 1014 | // Proto keywords. |
| 1015 | IdentifierInfo *kw_extend; |
| 1016 | IdentifierInfo *kw_option; |
| 1017 | IdentifierInfo *kw_optional; |
| 1018 | IdentifierInfo *kw_repeated; |
| 1019 | IdentifierInfo *kw_required; |
| 1020 | IdentifierInfo *kw_returns; |
| 1021 | |
| 1022 | // QT keywords. |
| 1023 | IdentifierInfo *kw_signals; |
| 1024 | IdentifierInfo *kw_qsignals; |
| 1025 | IdentifierInfo *kw_slots; |
| 1026 | IdentifierInfo *kw_qslots; |
| 1027 | |
| 1028 | // C# keywords |
| 1029 | IdentifierInfo *kw_dollar; |
| 1030 | IdentifierInfo *kw_base; |
| 1031 | IdentifierInfo *kw_byte; |
| 1032 | IdentifierInfo *kw_checked; |
| 1033 | IdentifierInfo *kw_decimal; |
| 1034 | IdentifierInfo *kw_delegate; |
| 1035 | IdentifierInfo *kw_event; |
| 1036 | IdentifierInfo *kw_fixed; |
| 1037 | IdentifierInfo *kw_foreach; |
| 1038 | IdentifierInfo *kw_implicit; |
| 1039 | IdentifierInfo *kw_internal; |
| 1040 | |
| 1041 | IdentifierInfo *kw_lock; |
| 1042 | IdentifierInfo *kw_null; |
| 1043 | IdentifierInfo *kw_object; |
| 1044 | IdentifierInfo *kw_out; |
| 1045 | |
| 1046 | IdentifierInfo *kw_params; |
| 1047 | |
| 1048 | IdentifierInfo *kw_ref; |
| 1049 | IdentifierInfo *kw_string; |
| 1050 | IdentifierInfo *kw_stackalloc; |
| 1051 | IdentifierInfo *kw_sbyte; |
| 1052 | IdentifierInfo *kw_sealed; |
| 1053 | IdentifierInfo *kw_uint; |
| 1054 | IdentifierInfo *kw_ulong; |
| 1055 | IdentifierInfo *kw_unchecked; |
| 1056 | IdentifierInfo *kw_unsafe; |
| 1057 | IdentifierInfo *kw_ushort; |
| 1058 | IdentifierInfo *kw_when; |
| 1059 | IdentifierInfo *kw_where; |
| 1060 | |
| 1061 | /// Returns \c true if \p Tok is a true JavaScript identifier, returns |
| 1062 | /// \c false if it is a keyword or a pseudo keyword. |
| 1063 | /// If \c AcceptIdentifierName is true, returns true not only for keywords, |
| 1064 | // but also for IdentifierName tokens (aka pseudo-keywords), such as |
| 1065 | // ``yield``. |
| 1066 | bool IsJavaScriptIdentifier(const FormatToken &Tok, |
| 1067 | bool AcceptIdentifierName = true) const { |
| 1068 | // Based on the list of JavaScript & TypeScript keywords here: |
| 1069 | // https://github.com/microsoft/TypeScript/blob/main/src/compiler/scanner.ts#L74 |
| 1070 | switch (Tok.Tok.getKind()) { |
| 1071 | case tok::kw_break: |
| 1072 | case tok::kw_case: |
| 1073 | case tok::kw_catch: |
| 1074 | case tok::kw_class: |
| 1075 | case tok::kw_continue: |
| 1076 | case tok::kw_const: |
| 1077 | case tok::kw_default: |
| 1078 | case tok::kw_delete: |
| 1079 | case tok::kw_do: |
| 1080 | case tok::kw_else: |
| 1081 | case tok::kw_enum: |
| 1082 | case tok::kw_export: |
| 1083 | case tok::kw_false: |
| 1084 | case tok::kw_for: |
| 1085 | case tok::kw_if: |
| 1086 | case tok::kw_import: |
| 1087 | case tok::kw_module: |
| 1088 | case tok::kw_new: |
| 1089 | case tok::kw_private: |
| 1090 | case tok::kw_protected: |
| 1091 | case tok::kw_public: |
| 1092 | case tok::kw_return: |
| 1093 | case tok::kw_static: |
| 1094 | case tok::kw_switch: |
| 1095 | case tok::kw_this: |
| 1096 | case tok::kw_throw: |
| 1097 | case tok::kw_true: |
| 1098 | case tok::kw_try: |
| 1099 | case tok::kw_typeof: |
| 1100 | case tok::kw_void: |
| 1101 | case tok::kw_while: |
| 1102 | // These are JS keywords that are lexed by LLVM/clang as keywords. |
| 1103 | return false; |
| 1104 | case tok::identifier: { |
| 1105 | // For identifiers, make sure they are true identifiers, excluding the |
| 1106 | // JavaScript pseudo-keywords (not lexed by LLVM/clang as keywords). |
| 1107 | bool IsPseudoKeyword = |
| 1108 | JsExtraKeywords.find(Tok.Tok.getIdentifierInfo()) != |
| 1109 | JsExtraKeywords.end(); |
| 1110 | return AcceptIdentifierName || !IsPseudoKeyword; |
| 1111 | } |
| 1112 | default: |
| 1113 | // Other keywords are handled in the switch below, to avoid problems due |
| 1114 | // to duplicate case labels when using the #include trick. |
| 1115 | break; |
| 1116 | } |
| 1117 | |
| 1118 | switch (Tok.Tok.getKind()) { |
| 1119 | // Handle C++ keywords not included above: these are all JS identifiers. |
| 1120 | #define KEYWORD(X, Y) case tok::kw_##X: |
| 1121 | #include "clang/Basic/TokenKinds.def" |
| 1122 | // #undef KEYWORD is not needed -- it's #undef-ed at the end of |
| 1123 | // TokenKinds.def |
| 1124 | return true; |
| 1125 | default: |
| 1126 | // All other tokens (punctuation etc) are not JS identifiers. |
| 1127 | return false; |
| 1128 | } |
| 1129 | } |
| 1130 | |
| 1131 | /// Returns \c true if \p Tok is a C# keyword, returns |
| 1132 | /// \c false if it is a anything else. |
| 1133 | bool isCSharpKeyword(const FormatToken &Tok) const { |
| 1134 | switch (Tok.Tok.getKind()) { |
| 1135 | case tok::kw_bool: |
| 1136 | case tok::kw_break: |
| 1137 | case tok::kw_case: |
| 1138 | case tok::kw_catch: |
| 1139 | case tok::kw_char: |
| 1140 | case tok::kw_class: |
| 1141 | case tok::kw_const: |
| 1142 | case tok::kw_continue: |
| 1143 | case tok::kw_default: |
| 1144 | case tok::kw_do: |
| 1145 | case tok::kw_double: |
| 1146 | case tok::kw_else: |
| 1147 | case tok::kw_enum: |
| 1148 | case tok::kw_explicit: |
| 1149 | case tok::kw_extern: |
| 1150 | case tok::kw_false: |
| 1151 | case tok::kw_float: |
| 1152 | case tok::kw_for: |
| 1153 | case tok::kw_goto: |
| 1154 | case tok::kw_if: |
| 1155 | case tok::kw_int: |
| 1156 | case tok::kw_long: |
| 1157 | case tok::kw_namespace: |
| 1158 | case tok::kw_new: |
| 1159 | case tok::kw_operator: |
| 1160 | case tok::kw_private: |
| 1161 | case tok::kw_protected: |
| 1162 | case tok::kw_public: |
| 1163 | case tok::kw_return: |
| 1164 | case tok::kw_short: |
| 1165 | case tok::kw_sizeof: |
| 1166 | case tok::kw_static: |
| 1167 | case tok::kw_struct: |
| 1168 | case tok::kw_switch: |
| 1169 | case tok::kw_this: |
| 1170 | case tok::kw_throw: |
| 1171 | case tok::kw_true: |
| 1172 | case tok::kw_try: |
| 1173 | case tok::kw_typeof: |
| 1174 | case tok::kw_using: |
| 1175 | case tok::kw_virtual: |
| 1176 | case tok::kw_void: |
| 1177 | case tok::kw_volatile: |
| 1178 | case tok::kw_while: |
| 1179 | return true; |
| 1180 | default: |
| 1181 | return Tok.is(tok::identifier) && |
| 1182 | CSharpExtraKeywords.find(Tok.Tok.getIdentifierInfo()) == |
| 1183 | CSharpExtraKeywords.end(); |
| 1184 | } |
| 1185 | } |
| 1186 | |
| 1187 | private: |
| 1188 | /// The JavaScript keywords beyond the C++ keyword set. |
| 1189 | std::unordered_set<IdentifierInfo *> JsExtraKeywords; |
| 1190 | |
| 1191 | /// The C# keywords beyond the C++ keyword set |
| 1192 | std::unordered_set<IdentifierInfo *> CSharpExtraKeywords; |
| 1193 | }; |
| 1194 | |
| 1195 | } // namespace format |
| 1196 | } // namespace clang |
| 1197 | |
| 1198 | #endif |
| 1 | //===--- Token.h - Token interface ------------------------------*- 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 | // This file defines the Token interface. |
| 10 | // |
| 11 | //===----------------------------------------------------------------------===// |
| 12 | |
| 13 | #ifndef LLVM_CLANG_LEX_TOKEN_H |
| 14 | #define LLVM_CLANG_LEX_TOKEN_H |
| 15 | |
| 16 | #include "clang/Basic/SourceLocation.h" |
| 17 | #include "clang/Basic/TokenKinds.h" |
| 18 | #include "llvm/ADT/StringRef.h" |
| 19 | #include <cassert> |
| 20 | |
| 21 | namespace clang { |
| 22 | |
| 23 | class IdentifierInfo; |
| 24 | |
| 25 | /// Token - This structure provides full information about a lexed token. |
| 26 | /// It is not intended to be space efficient, it is intended to return as much |
| 27 | /// information as possible about each returned token. This is expected to be |
| 28 | /// compressed into a smaller form if memory footprint is important. |
| 29 | /// |
| 30 | /// The parser can create a special "annotation token" representing a stream of |
| 31 | /// tokens that were parsed and semantically resolved, e.g.: "foo::MyClass<int>" |
| 32 | /// can be represented by a single typename annotation token that carries |
| 33 | /// information about the SourceRange of the tokens and the type object. |
| 34 | class Token { |
| 35 | /// The location of the token. This is actually a SourceLocation. |
| 36 | SourceLocation::UIntTy Loc; |
| 37 | |
| 38 | // Conceptually these next two fields could be in a union. However, this |
| 39 | // causes gcc 4.2 to pessimize LexTokenInternal, a very performance critical |
| 40 | // routine. Keeping as separate members with casts until a more beautiful fix |
| 41 | // presents itself. |
| 42 | |
| 43 | /// UintData - This holds either the length of the token text, when |
| 44 | /// a normal token, or the end of the SourceRange when an annotation |
| 45 | /// token. |
| 46 | SourceLocation::UIntTy UintData; |
| 47 | |
| 48 | /// PtrData - This is a union of four different pointer types, which depends |
| 49 | /// on what type of token this is: |
| 50 | /// Identifiers, keywords, etc: |
| 51 | /// This is an IdentifierInfo*, which contains the uniqued identifier |
| 52 | /// spelling. |
| 53 | /// Literals: isLiteral() returns true. |
| 54 | /// This is a pointer to the start of the token in a text buffer, which |
| 55 | /// may be dirty (have trigraphs / escaped newlines). |
| 56 | /// Annotations (resolved type names, C++ scopes, etc): isAnnotation(). |
| 57 | /// This is a pointer to sema-specific data for the annotation token. |
| 58 | /// Eof: |
| 59 | // This is a pointer to a Decl. |
| 60 | /// Other: |
| 61 | /// This is null. |
| 62 | void *PtrData; |
| 63 | |
| 64 | /// Kind - The actual flavor of token this is. |
| 65 | tok::TokenKind Kind; |
| 66 | |
| 67 | /// Flags - Bits we track about this token, members of the TokenFlags enum. |
| 68 | unsigned short Flags; |
| 69 | |
| 70 | public: |
| 71 | // Various flags set per token: |
| 72 | enum TokenFlags { |
| 73 | StartOfLine = 0x01, // At start of line or only after whitespace |
| 74 | // (considering the line after macro expansion). |
| 75 | LeadingSpace = 0x02, // Whitespace exists before this token (considering |
| 76 | // whitespace after macro expansion). |
| 77 | DisableExpand = 0x04, // This identifier may never be macro expanded. |
| 78 | NeedsCleaning = 0x08, // Contained an escaped newline or trigraph. |
| 79 | LeadingEmptyMacro = 0x10, // Empty macro exists before this token. |
| 80 | HasUDSuffix = 0x20, // This string or character literal has a ud-suffix. |
| 81 | HasUCN = 0x40, // This identifier contains a UCN. |
| 82 | IgnoredComma = 0x80, // This comma is not a macro argument separator (MS). |
| 83 | StringifiedInMacro = 0x100, // This string or character literal is formed by |
| 84 | // macro stringizing or charizing operator. |
| 85 | CommaAfterElided = 0x200, // The comma following this token was elided (MS). |
| 86 | IsEditorPlaceholder = 0x400, // This identifier is a placeholder. |
| 87 | IsReinjected = 0x800, // A phase 4 token that was produced before and |
| 88 | // re-added, e.g. via EnterTokenStream. Annotation |
| 89 | // tokens are *not* reinjected. |
| 90 | }; |
| 91 | |
| 92 | tok::TokenKind getKind() const { return Kind; } |
| 93 | void setKind(tok::TokenKind K) { Kind = K; } |
| 94 | |
| 95 | /// is/isNot - Predicates to check if this token is a specific kind, as in |
| 96 | /// "if (Tok.is(tok::l_brace)) {...}". |
| 97 | bool is(tok::TokenKind K) const { return Kind == K; } |
| 98 | bool isNot(tok::TokenKind K) const { return Kind != K; } |
| 99 | bool isOneOf(tok::TokenKind K1, tok::TokenKind K2) const { |
| 100 | return is(K1) || is(K2); |
| 101 | } |
| 102 | template <typename... Ts> |
| 103 | bool isOneOf(tok::TokenKind K1, tok::TokenKind K2, Ts... Ks) const { |
| 104 | return is(K1) || isOneOf(K2, Ks...); |
| 105 | } |
| 106 | |
| 107 | /// Return true if this is a raw identifier (when lexing |
| 108 | /// in raw mode) or a non-keyword identifier (when lexing in non-raw mode). |
| 109 | bool isAnyIdentifier() const { |
| 110 | return tok::isAnyIdentifier(getKind()); |
| 111 | } |
| 112 | |
| 113 | /// Return true if this is a "literal", like a numeric |
| 114 | /// constant, string, etc. |
| 115 | bool isLiteral() const { |
| 116 | return tok::isLiteral(getKind()); |
| 117 | } |
| 118 | |
| 119 | /// Return true if this is any of tok::annot_* kind tokens. |
| 120 | bool isAnnotation() const { |
| 121 | return tok::isAnnotation(getKind()); |
| 122 | } |
| 123 | |
| 124 | /// Return a source location identifier for the specified |
| 125 | /// offset in the current file. |
| 126 | SourceLocation getLocation() const { |
| 127 | return SourceLocation::getFromRawEncoding(Loc); |
| 128 | } |
| 129 | unsigned getLength() const { |
| 130 | assert(!isAnnotation() && "Annotation tokens have no length field")(static_cast <bool> (!isAnnotation() && "Annotation tokens have no length field" ) ? void (0) : __assert_fail ("!isAnnotation() && \"Annotation tokens have no length field\"" , "clang/include/clang/Lex/Token.h", 130, __extension__ __PRETTY_FUNCTION__ )); |
| 131 | return UintData; |
| 132 | } |
| 133 | |
| 134 | void setLocation(SourceLocation L) { Loc = L.getRawEncoding(); } |
| 135 | void setLength(unsigned Len) { |
| 136 | assert(!isAnnotation() && "Annotation tokens have no length field")(static_cast <bool> (!isAnnotation() && "Annotation tokens have no length field" ) ? void (0) : __assert_fail ("!isAnnotation() && \"Annotation tokens have no length field\"" , "clang/include/clang/Lex/Token.h", 136, __extension__ __PRETTY_FUNCTION__ )); |
| 137 | UintData = Len; |
| 138 | } |
| 139 | |
| 140 | SourceLocation getAnnotationEndLoc() const { |
| 141 | assert(isAnnotation() && "Used AnnotEndLocID on non-annotation token")(static_cast <bool> (isAnnotation() && "Used AnnotEndLocID on non-annotation token" ) ? void (0) : __assert_fail ("isAnnotation() && \"Used AnnotEndLocID on non-annotation token\"" , "clang/include/clang/Lex/Token.h", 141, __extension__ __PRETTY_FUNCTION__ )); |
| 142 | return SourceLocation::getFromRawEncoding(UintData ? UintData : Loc); |
| 143 | } |
| 144 | void setAnnotationEndLoc(SourceLocation L) { |
| 145 | assert(isAnnotation() && "Used AnnotEndLocID on non-annotation token")(static_cast <bool> (isAnnotation() && "Used AnnotEndLocID on non-annotation token" ) ? void (0) : __assert_fail ("isAnnotation() && \"Used AnnotEndLocID on non-annotation token\"" , "clang/include/clang/Lex/Token.h", 145, __extension__ __PRETTY_FUNCTION__ )); |
| 146 | UintData = L.getRawEncoding(); |
| 147 | } |
| 148 | |
| 149 | SourceLocation getLastLoc() const { |
| 150 | return isAnnotation() ? getAnnotationEndLoc() : getLocation(); |
| 151 | } |
| 152 | |
| 153 | SourceLocation getEndLoc() const { |
| 154 | return isAnnotation() ? getAnnotationEndLoc() |
| 155 | : getLocation().getLocWithOffset(getLength()); |
| 156 | } |
| 157 | |
| 158 | /// SourceRange of the group of tokens that this annotation token |
| 159 | /// represents. |
| 160 | SourceRange getAnnotationRange() const { |
| 161 | return SourceRange(getLocation(), getAnnotationEndLoc()); |
| 162 | } |
| 163 | void setAnnotationRange(SourceRange R) { |
| 164 | setLocation(R.getBegin()); |
| 165 | setAnnotationEndLoc(R.getEnd()); |
| 166 | } |
| 167 | |
| 168 | const char *getName() const { return tok::getTokenName(Kind); } |
| 169 | |
| 170 | /// Reset all flags to cleared. |
| 171 | void startToken() { |
| 172 | Kind = tok::unknown; |
| 173 | Flags = 0; |
| 174 | PtrData = nullptr; |
| 175 | UintData = 0; |
| 176 | Loc = SourceLocation().getRawEncoding(); |
| 177 | } |
| 178 | |
| 179 | IdentifierInfo *getIdentifierInfo() const { |
| 180 | assert(isNot(tok::raw_identifier) &&(static_cast <bool> (isNot(tok::raw_identifier) && "getIdentifierInfo() on a tok::raw_identifier token!") ? void (0) : __assert_fail ("isNot(tok::raw_identifier) && \"getIdentifierInfo() on a tok::raw_identifier token!\"" , "clang/include/clang/Lex/Token.h", 181, __extension__ __PRETTY_FUNCTION__ )) |
| 181 | "getIdentifierInfo() on a tok::raw_identifier token!")(static_cast <bool> (isNot(tok::raw_identifier) && "getIdentifierInfo() on a tok::raw_identifier token!") ? void (0) : __assert_fail ("isNot(tok::raw_identifier) && \"getIdentifierInfo() on a tok::raw_identifier token!\"" , "clang/include/clang/Lex/Token.h", 181, __extension__ __PRETTY_FUNCTION__ )); |
| 182 | assert(!isAnnotation() &&(static_cast <bool> (!isAnnotation() && "getIdentifierInfo() on an annotation token!" ) ? void (0) : __assert_fail ("!isAnnotation() && \"getIdentifierInfo() on an annotation token!\"" , "clang/include/clang/Lex/Token.h", 183, __extension__ __PRETTY_FUNCTION__ )) |
| 183 | "getIdentifierInfo() on an annotation token!")(static_cast <bool> (!isAnnotation() && "getIdentifierInfo() on an annotation token!" ) ? void (0) : __assert_fail ("!isAnnotation() && \"getIdentifierInfo() on an annotation token!\"" , "clang/include/clang/Lex/Token.h", 183, __extension__ __PRETTY_FUNCTION__ )); |
| 184 | if (isLiteral()) return nullptr; |
| 185 | if (is(tok::eof)) return nullptr; |
| 186 | return (IdentifierInfo*) PtrData; |
| 187 | } |
| 188 | void setIdentifierInfo(IdentifierInfo *II) { |
| 189 | PtrData = (void*) II; |
| 190 | } |
| 191 | |
| 192 | const void *getEofData() const { |
| 193 | assert(is(tok::eof))(static_cast <bool> (is(tok::eof)) ? void (0) : __assert_fail ("is(tok::eof)", "clang/include/clang/Lex/Token.h", 193, __extension__ __PRETTY_FUNCTION__)); |
| 194 | return reinterpret_cast<const void *>(PtrData); |
| 195 | } |
| 196 | void setEofData(const void *D) { |
| 197 | assert(is(tok::eof))(static_cast <bool> (is(tok::eof)) ? void (0) : __assert_fail ("is(tok::eof)", "clang/include/clang/Lex/Token.h", 197, __extension__ __PRETTY_FUNCTION__)); |
| 198 | assert(!PtrData)(static_cast <bool> (!PtrData) ? void (0) : __assert_fail ("!PtrData", "clang/include/clang/Lex/Token.h", 198, __extension__ __PRETTY_FUNCTION__)); |
| 199 | PtrData = const_cast<void *>(D); |
| 200 | } |
| 201 | |
| 202 | /// getRawIdentifier - For a raw identifier token (i.e., an identifier |
| 203 | /// lexed in raw mode), returns a reference to the text substring in the |
| 204 | /// buffer if known. |
| 205 | StringRef getRawIdentifier() const { |
| 206 | assert(is(tok::raw_identifier))(static_cast <bool> (is(tok::raw_identifier)) ? void (0 ) : __assert_fail ("is(tok::raw_identifier)", "clang/include/clang/Lex/Token.h" , 206, __extension__ __PRETTY_FUNCTION__)); |
| 207 | return StringRef(reinterpret_cast<const char *>(PtrData), getLength()); |
| 208 | } |
| 209 | void setRawIdentifierData(const char *Ptr) { |
| 210 | assert(is(tok::raw_identifier))(static_cast <bool> (is(tok::raw_identifier)) ? void (0 ) : __assert_fail ("is(tok::raw_identifier)", "clang/include/clang/Lex/Token.h" , 210, __extension__ __PRETTY_FUNCTION__)); |
| 211 | PtrData = const_cast<char*>(Ptr); |
| 212 | } |
| 213 | |
| 214 | /// getLiteralData - For a literal token (numeric constant, string, etc), this |
| 215 | /// returns a pointer to the start of it in the text buffer if known, null |
| 216 | /// otherwise. |
| 217 | const char *getLiteralData() const { |
| 218 | assert(isLiteral() && "Cannot get literal data of non-literal")(static_cast <bool> (isLiteral() && "Cannot get literal data of non-literal" ) ? void (0) : __assert_fail ("isLiteral() && \"Cannot get literal data of non-literal\"" , "clang/include/clang/Lex/Token.h", 218, __extension__ __PRETTY_FUNCTION__ )); |
| 219 | return reinterpret_cast<const char*>(PtrData); |
| 220 | } |
| 221 | void setLiteralData(const char *Ptr) { |
| 222 | assert(isLiteral() && "Cannot set literal data of non-literal")(static_cast <bool> (isLiteral() && "Cannot set literal data of non-literal" ) ? void (0) : __assert_fail ("isLiteral() && \"Cannot set literal data of non-literal\"" , "clang/include/clang/Lex/Token.h", 222, __extension__ __PRETTY_FUNCTION__ )); |
| 223 | PtrData = const_cast<char*>(Ptr); |
| 224 | } |
| 225 | |
| 226 | void *getAnnotationValue() const { |
| 227 | assert(isAnnotation() && "Used AnnotVal on non-annotation token")(static_cast <bool> (isAnnotation() && "Used AnnotVal on non-annotation token" ) ? void (0) : __assert_fail ("isAnnotation() && \"Used AnnotVal on non-annotation token\"" , "clang/include/clang/Lex/Token.h", 227, __extension__ __PRETTY_FUNCTION__ )); |
| 228 | return PtrData; |
| 229 | } |
| 230 | void setAnnotationValue(void *val) { |
| 231 | assert(isAnnotation() && "Used AnnotVal on non-annotation token")(static_cast <bool> (isAnnotation() && "Used AnnotVal on non-annotation token" ) ? void (0) : __assert_fail ("isAnnotation() && \"Used AnnotVal on non-annotation token\"" , "clang/include/clang/Lex/Token.h", 231, __extension__ __PRETTY_FUNCTION__ )); |
| 232 | PtrData = val; |
| 233 | } |
| 234 | |
| 235 | /// Set the specified flag. |
| 236 | void setFlag(TokenFlags Flag) { |
| 237 | Flags |= Flag; |
| 238 | } |
| 239 | |
| 240 | /// Get the specified flag. |
| 241 | bool getFlag(TokenFlags Flag) const { |
| 242 | return (Flags & Flag) != 0; |
| 243 | } |
| 244 | |
| 245 | /// Unset the specified flag. |
| 246 | void clearFlag(TokenFlags Flag) { |
| 247 | Flags &= ~Flag; |
| 248 | } |
| 249 | |
| 250 | /// Return the internal represtation of the flags. |
| 251 | /// |
| 252 | /// This is only intended for low-level operations such as writing tokens to |
| 253 | /// disk. |
| 254 | unsigned getFlags() const { |
| 255 | return Flags; |
| 256 | } |
| 257 | |
| 258 | /// Set a flag to either true or false. |
| 259 | void setFlagValue(TokenFlags Flag, bool Val) { |
| 260 | if (Val) |
| 261 | setFlag(Flag); |
| 262 | else |
| 263 | clearFlag(Flag); |
| 264 | } |
| 265 | |
| 266 | /// isAtStartOfLine - Return true if this token is at the start of a line. |
| 267 | /// |
| 268 | bool isAtStartOfLine() const { return getFlag(StartOfLine); } |
| 269 | |
| 270 | /// Return true if this token has whitespace before it. |
| 271 | /// |
| 272 | bool hasLeadingSpace() const { return getFlag(LeadingSpace); } |
| 273 | |
| 274 | /// Return true if this identifier token should never |
| 275 | /// be expanded in the future, due to C99 6.10.3.4p2. |
| 276 | bool isExpandDisabled() const { return getFlag(DisableExpand); } |
| 277 | |
| 278 | /// Return true if we have an ObjC keyword identifier. |
| 279 | bool isObjCAtKeyword(tok::ObjCKeywordKind objcKey) const; |
| 280 | |
| 281 | /// Return the ObjC keyword kind. |
| 282 | tok::ObjCKeywordKind getObjCKeywordID() const; |
| 283 | |
| 284 | /// Return true if this token has trigraphs or escaped newlines in it. |
| 285 | bool needsCleaning() const { return getFlag(NeedsCleaning); } |
| 286 | |
| 287 | /// Return true if this token has an empty macro before it. |
| 288 | /// |
| 289 | bool hasLeadingEmptyMacro() const { return getFlag(LeadingEmptyMacro); } |
| 290 | |
| 291 | /// Return true if this token is a string or character literal which |
| 292 | /// has a ud-suffix. |
| 293 | bool hasUDSuffix() const { return getFlag(HasUDSuffix); } |
| 294 | |
| 295 | /// Returns true if this token contains a universal character name. |
| 296 | bool hasUCN() const { return getFlag(HasUCN); } |
| 297 | |
| 298 | /// Returns true if this token is formed by macro by stringizing or charizing |
| 299 | /// operator. |
| 300 | bool stringifiedInMacro() const { return getFlag(StringifiedInMacro); } |
| 301 | |
| 302 | /// Returns true if the comma after this token was elided. |
| 303 | bool commaAfterElided() const { return getFlag(CommaAfterElided); } |
| 304 | |
| 305 | /// Returns true if this token is an editor placeholder. |
| 306 | /// |
| 307 | /// Editor placeholders are produced by the code-completion engine and are |
| 308 | /// represented as characters between '<#' and '#>' in the source code. The |
| 309 | /// lexer uses identifier tokens to represent placeholders. |
| 310 | bool isEditorPlaceholder() const { return getFlag(IsEditorPlaceholder); } |
| 311 | }; |
| 312 | |
| 313 | /// Information about the conditional stack (\#if directives) |
| 314 | /// currently active. |
| 315 | struct PPConditionalInfo { |
| 316 | /// Location where the conditional started. |
| 317 | SourceLocation IfLoc; |
| 318 | |
| 319 | /// True if this was contained in a skipping directive, e.g., |
| 320 | /// in a "\#if 0" block. |
| 321 | bool WasSkipping; |
| 322 | |
| 323 | /// True if we have emitted tokens already, and now we're in |
| 324 | /// an \#else block or something. Only useful in Skipping blocks. |
| 325 | bool FoundNonSkip; |
| 326 | |
| 327 | /// True if we've seen a \#else in this block. If so, |
| 328 | /// \#elif/\#else directives are not allowed. |
| 329 | bool FoundElse; |
| 330 | }; |
| 331 | |
| 332 | } // end namespace clang |
| 333 | |
| 334 | #endif // LLVM_CLANG_LEX_TOKEN_H |