Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 1 | //===--- Format.cpp - Format C++ code -------------------------------------===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | /// |
| 10 | /// \file |
| 11 | /// \brief This file implements functions declared in Format.h. This will be |
| 12 | /// split into separate files as we go. |
| 13 | /// |
| 14 | /// This is EXPERIMENTAL code under heavy development. It is not in a state yet, |
| 15 | /// where it can be used to format real code. |
| 16 | /// |
| 17 | //===----------------------------------------------------------------------===// |
| 18 | |
| 19 | #include "clang/Format/Format.h" |
| 20 | |
| 21 | #include "clang/Basic/SourceManager.h" |
| 22 | #include "clang/Lex/Lexer.h" |
| 23 | |
| 24 | #include "UnwrappedLineParser.h" |
| 25 | |
| 26 | namespace clang { |
| 27 | namespace format { |
| 28 | |
| 29 | // FIXME: Move somewhere sane. |
| 30 | struct TokenAnnotation { |
| 31 | enum TokenType { TT_Unknown, TT_TemplateOpener, TT_BinaryOperator, |
| 32 | TT_UnaryOperator, TT_OverloadedOperator, TT_PointerOrReference, |
| 33 | TT_ConditionalExpr, TT_LineComment, TT_BlockComment }; |
| 34 | |
| 35 | TokenType Type; |
| 36 | |
| 37 | /// \brief The current parenthesis level, i.e. the number of opening minus |
| 38 | /// the number of closing parenthesis left of the current position. |
| 39 | unsigned ParenLevel; |
| 40 | |
| 41 | bool SpaceRequiredBefore; |
| 42 | bool CanBreakBefore; |
| 43 | bool MustBreakBefore; |
| 44 | }; |
| 45 | |
| 46 | using llvm::MutableArrayRef; |
| 47 | |
| 48 | FormatStyle getLLVMStyle() { |
| 49 | FormatStyle LLVMStyle; |
| 50 | LLVMStyle.ColumnLimit = 80; |
| 51 | LLVMStyle.MaxEmptyLinesToKeep = 1; |
| 52 | LLVMStyle.PointerAndReferenceBindToType = false; |
| 53 | LLVMStyle.AccessModifierOffset = -2; |
| 54 | LLVMStyle.SplitTemplateClosingGreater = true; |
| 55 | return LLVMStyle; |
| 56 | } |
| 57 | |
| 58 | FormatStyle getGoogleStyle() { |
| 59 | FormatStyle GoogleStyle; |
| 60 | GoogleStyle.ColumnLimit = 80; |
| 61 | GoogleStyle.MaxEmptyLinesToKeep = 1; |
| 62 | GoogleStyle.PointerAndReferenceBindToType = true; |
| 63 | GoogleStyle.AccessModifierOffset = -1; |
| 64 | GoogleStyle.SplitTemplateClosingGreater = false; |
| 65 | return GoogleStyle; |
| 66 | } |
| 67 | |
| 68 | struct OptimizationParameters { |
| 69 | unsigned PenaltyExtraLine; |
| 70 | unsigned PenaltyIndentLevel; |
| 71 | }; |
| 72 | |
| 73 | class UnwrappedLineFormatter { |
| 74 | public: |
| 75 | UnwrappedLineFormatter(const FormatStyle &Style, SourceManager &SourceMgr, |
| 76 | const UnwrappedLine &Line, |
| 77 | const std::vector<TokenAnnotation> &Annotations, |
| 78 | tooling::Replacements &Replaces) |
| 79 | : Style(Style), |
| 80 | SourceMgr(SourceMgr), |
| 81 | Line(Line), |
| 82 | Annotations(Annotations), |
| 83 | Replaces(Replaces) { |
| 84 | Parameters.PenaltyExtraLine = 100; |
| 85 | Parameters.PenaltyIndentLevel = 5; |
| 86 | } |
| 87 | |
| 88 | void format() { |
| 89 | formatFirstToken(); |
| 90 | count = 0; |
| 91 | IndentState State; |
| 92 | State.Column = Line.Level * 2 + Line.Tokens[0].Tok.getLength(); |
| 93 | State.CtorInitializerOnNewLine = false; |
| 94 | State.InCtorInitializer = false; |
| 95 | State.ConsumedTokens = 1; |
| 96 | |
| 97 | //State.UsedIndent.push_back(Line.Level * 2); |
| 98 | State.Indent.push_back(Line.Level * 2 + 4); |
| 99 | State.LastSpace.push_back(Line.Level * 2); |
| 100 | |
| 101 | // Start iterating at 1 as we have correctly formatted of Token #0 above. |
| 102 | for (unsigned i = 1, n = Line.Tokens.size(); i != n; ++i) { |
| 103 | unsigned NoBreak = calcPenalty(State, false, UINT_MAX); |
| 104 | unsigned Break = calcPenalty(State, true, NoBreak); |
| 105 | addToken(Break < NoBreak, false, State); |
| 106 | } |
| 107 | } |
| 108 | |
| 109 | private: |
| 110 | /// \brief The current state when indenting a unwrapped line. |
| 111 | /// |
| 112 | /// As the indenting tries different combinations this is copied by value. |
| 113 | struct IndentState { |
| 114 | /// \brief The number of used columns in the current line. |
| 115 | unsigned Column; |
| 116 | |
| 117 | /// \brief The number of tokens already consumed. |
| 118 | unsigned ConsumedTokens; |
| 119 | |
| 120 | /// \brief The position to which a specific parenthesis level needs to be |
| 121 | /// indented. |
| 122 | std::vector<unsigned> Indent; |
| 123 | |
| 124 | std::vector<unsigned> LastSpace; |
| 125 | |
| 126 | bool CtorInitializerOnNewLine; |
| 127 | bool InCtorInitializer; |
| 128 | |
| 129 | /// \brief Comparison operator to be able to used \c IndentState in \c map. |
| 130 | bool operator<(const IndentState &Other) const { |
| 131 | if (Other.ConsumedTokens != ConsumedTokens) |
| 132 | return Other.ConsumedTokens > ConsumedTokens; |
| 133 | if (Other.Column != Column) |
| 134 | return Other.Column > Column; |
| 135 | if (Other.Indent.size() != Indent.size()) |
| 136 | return Other.Indent.size() > Indent.size(); |
| 137 | for (int i = 0, e = Indent.size(); i != e; ++i) { |
| 138 | if (Other.Indent[i] != Indent[i]) |
| 139 | return Other.Indent[i] > Indent[i]; |
| 140 | } |
| 141 | if (Other.LastSpace.size() != LastSpace.size()) |
| 142 | return Other.LastSpace.size() > LastSpace.size(); |
| 143 | for (int i = 0, e = LastSpace.size(); i != e; ++i) { |
| 144 | if (Other.LastSpace[i] != LastSpace[i]) |
| 145 | return Other.LastSpace[i] > LastSpace[i]; |
| 146 | } |
| 147 | return false; |
| 148 | } |
| 149 | }; |
| 150 | |
| 151 | /// Append the next token to \p State. |
| 152 | void addToken(bool Newline, bool DryRun, IndentState &State) { |
| 153 | unsigned Index = State.ConsumedTokens; |
| 154 | const FormatToken &Current = Line.Tokens[Index]; |
| 155 | const FormatToken &Previous = Line.Tokens[Index - 1]; |
| 156 | unsigned ParenLevel = Annotations[Index].ParenLevel; |
| 157 | |
| 158 | if (Current.Tok.is(tok::l_paren) || Current.Tok.is(tok::l_square) || |
| 159 | Annotations[Index].Type == TokenAnnotation::TT_TemplateOpener) { |
| 160 | State.Indent.push_back(4 + State.LastSpace.back()); |
| 161 | State.LastSpace.push_back(State.LastSpace.back()); |
| 162 | } |
| 163 | |
| 164 | if (Newline) { |
| 165 | if (Current.Tok.is(tok::string_literal) && |
| 166 | Previous.Tok.is(tok::string_literal)) |
| 167 | State.Column = State.Column - Previous.Tok.getLength(); |
| 168 | else if (Previous.Tok.is(tok::equal) && ParenLevel != 0) |
| 169 | State.Column = State.Indent[ParenLevel] + 4; |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 170 | else if (ParenLevel < State.Indent.size()) |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 171 | State.Column = State.Indent[ParenLevel]; |
| 172 | if (!DryRun) |
| 173 | replaceWhitespace(Current, 1, State.Column); |
| 174 | |
| 175 | State.Column += Current.Tok.getLength(); |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 176 | if (ParenLevel < State.LastSpace.size()) |
| 177 | State.LastSpace[ParenLevel] = State.Indent[ParenLevel]; |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 178 | if (Current.Tok.is(tok::colon) && |
| 179 | Annotations[Index].Type != TokenAnnotation::TT_ConditionalExpr) { |
| 180 | State.Indent[ParenLevel] += 2; |
| 181 | State.CtorInitializerOnNewLine = true; |
| 182 | State.InCtorInitializer = true; |
| 183 | } |
| 184 | } else { |
| 185 | unsigned Spaces = Annotations[Index].SpaceRequiredBefore ? 1 : 0; |
| 186 | if (Annotations[Index].Type == TokenAnnotation::TT_LineComment) |
| 187 | Spaces = 2; |
| 188 | if (!DryRun) |
| 189 | replaceWhitespace(Current, 0, Spaces); |
| 190 | if (Previous.Tok.is(tok::l_paren)) |
| 191 | State.Indent[ParenLevel] = State.Column; |
| 192 | if (Previous.Tok.is(tok::less) && |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 193 | Annotations[Index - 1].Type == TokenAnnotation::TT_TemplateOpener && |
| 194 | ParenLevel < State.Indent.size()) |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 195 | State.Indent[ParenLevel] = State.Column; |
| 196 | if (Current.Tok.is(tok::colon)) { |
| 197 | State.Indent[ParenLevel] = State.Column + 3; |
| 198 | State.InCtorInitializer = true; |
| 199 | } |
| 200 | // Top-level spaces are exempt as that mostly leads to better results. |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 201 | if (Spaces > 0 && ParenLevel != 0 && |
| 202 | ParenLevel < State.LastSpace.size()) |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 203 | State.LastSpace[ParenLevel] = State.Column + Spaces; |
| 204 | State.Column += Current.Tok.getLength() + Spaces; |
| 205 | } |
| 206 | |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 207 | if (!DryRun && |
| 208 | (Current.Tok.is(tok::r_paren) || Current.Tok.is(tok::r_square) || |
| 209 | Annotations[Index].Type == TokenAnnotation::TT_TemplateOpener)) { |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 210 | State.Indent.pop_back(); |
| 211 | State.LastSpace.pop_back(); |
| 212 | } |
| 213 | |
| 214 | ++State.ConsumedTokens; |
| 215 | } |
| 216 | |
| 217 | typedef std::map<IndentState, unsigned> StateMap; |
| 218 | StateMap Memory; |
| 219 | |
| 220 | unsigned splitPenalty(const FormatToken &Token) { |
| 221 | if (Token.Tok.is(tok::semi)) |
| 222 | return 0; |
| 223 | if (Token.Tok.is(tok::comma)) |
| 224 | return 1; |
| 225 | if (Token.Tok.is(tok::equal) || Token.Tok.is(tok::l_paren) || |
| 226 | Token.Tok.is(tok::pipepipe) || Token.Tok.is(tok::ampamp)) |
| 227 | return 2; |
| 228 | return 3; |
| 229 | } |
| 230 | |
| 231 | /// \brief Calculate the number of lines needed to format the remaining part |
| 232 | /// of the unwrapped line. |
| 233 | /// |
| 234 | /// Assumes the formatting so far has led to |
| 235 | /// the \c IndentState \p State. If \p NewLine is set, a new line will be |
| 236 | /// added after the previous token. |
| 237 | /// |
| 238 | /// \param StopAt is used for optimization. If we can determine that we'll |
| 239 | /// definitely need at least \p StopAt additional lines, we already know of a |
| 240 | /// better solution. |
| 241 | unsigned calcPenalty(IndentState State, bool NewLine, unsigned StopAt) { |
| 242 | // We are at the end of the unwrapped line, so we don't need any more lines. |
| 243 | if (State.ConsumedTokens >= Line.Tokens.size()) |
| 244 | return 0; |
| 245 | |
| 246 | if (!NewLine && Annotations[State.ConsumedTokens].MustBreakBefore) |
| 247 | return UINT_MAX; |
| 248 | if (NewLine && !Annotations[State.ConsumedTokens].CanBreakBefore) |
| 249 | return UINT_MAX; |
| 250 | |
| 251 | if (State.ConsumedTokens > 0 && !NewLine && |
| 252 | State.CtorInitializerOnNewLine && |
| 253 | Line.Tokens[State.ConsumedTokens - 1].Tok.is(tok::comma)) |
| 254 | return UINT_MAX; |
| 255 | |
| 256 | if (NewLine && State.InCtorInitializer && !State.CtorInitializerOnNewLine) |
| 257 | return UINT_MAX; |
| 258 | |
| 259 | addToken(NewLine, true, State); |
| 260 | |
| 261 | // Exceeding column limit is bad. |
| 262 | if (State.Column > Style.ColumnLimit) |
| 263 | return UINT_MAX; |
| 264 | |
| 265 | unsigned CurrentPenalty = 0; |
| 266 | if (NewLine) { |
| 267 | CurrentPenalty += Parameters.PenaltyIndentLevel * |
| 268 | Annotations[State.ConsumedTokens - 1].ParenLevel + |
| 269 | Parameters.PenaltyExtraLine + |
| 270 | splitPenalty(Line.Tokens[State.ConsumedTokens - 2]); |
| 271 | } |
| 272 | |
| 273 | if (StopAt <= CurrentPenalty) |
| 274 | return UINT_MAX; |
| 275 | StopAt -= CurrentPenalty; |
| 276 | |
| 277 | // Has this state already been examined? |
| 278 | StateMap::iterator I = Memory.find(State); |
| 279 | if (I != Memory.end()) |
| 280 | return I->second; |
| 281 | ++count; |
| 282 | |
| 283 | unsigned NoBreak = calcPenalty(State, false, StopAt); |
| 284 | unsigned WithBreak = calcPenalty(State, true, std::min(StopAt, NoBreak)); |
| 285 | unsigned Result = std::min(NoBreak, WithBreak); |
| 286 | if (Result != UINT_MAX) |
| 287 | Result += CurrentPenalty; |
| 288 | Memory[State] = Result; |
| 289 | assert(Memory.find(State) != Memory.end()); |
| 290 | return Result; |
| 291 | } |
| 292 | |
| 293 | /// \brief Replaces the whitespace in front of \p Tok. Only call once for |
| 294 | /// each \c FormatToken. |
| 295 | void replaceWhitespace(const FormatToken &Tok, unsigned NewLines, |
| 296 | unsigned Spaces) { |
| 297 | Replaces.insert(tooling::Replacement( |
| 298 | SourceMgr, Tok.WhiteSpaceStart, Tok.WhiteSpaceLength, |
| 299 | std::string(NewLines, '\n') + std::string(Spaces, ' '))); |
| 300 | } |
| 301 | |
| 302 | /// \brief Add a new line and the required indent before the first Token |
| 303 | /// of the \c UnwrappedLine. |
| 304 | void formatFirstToken() { |
| 305 | const FormatToken &Token = Line.Tokens[0]; |
| 306 | if (Token.WhiteSpaceStart.isValid()) { |
| 307 | unsigned Newlines = |
| 308 | std::min(Token.NewlinesBefore, Style.MaxEmptyLinesToKeep + 1); |
| 309 | unsigned Offset = SourceMgr.getFileOffset(Token.WhiteSpaceStart); |
| 310 | if (Newlines == 0 && Offset != 0) |
| 311 | Newlines = 1; |
| 312 | unsigned Indent = Line.Level * 2; |
| 313 | if (Token.Tok.is(tok::kw_public) || Token.Tok.is(tok::kw_protected) || |
| 314 | Token.Tok.is(tok::kw_private)) |
| 315 | Indent += Style.AccessModifierOffset; |
| 316 | replaceWhitespace(Token, Newlines, Indent); |
| 317 | } |
| 318 | } |
| 319 | |
| 320 | FormatStyle Style; |
| 321 | SourceManager &SourceMgr; |
| 322 | const UnwrappedLine &Line; |
| 323 | const std::vector<TokenAnnotation> &Annotations; |
| 324 | tooling::Replacements &Replaces; |
| 325 | unsigned int count; |
| 326 | |
| 327 | OptimizationParameters Parameters; |
| 328 | }; |
| 329 | |
| 330 | /// \brief Determines extra information about the tokens comprising an |
| 331 | /// \c UnwrappedLine. |
| 332 | class TokenAnnotator { |
| 333 | public: |
| 334 | TokenAnnotator(const UnwrappedLine &Line, const FormatStyle &Style, |
| 335 | SourceManager &SourceMgr) |
| 336 | : Line(Line), |
| 337 | Style(Style), |
| 338 | SourceMgr(SourceMgr) { |
| 339 | } |
| 340 | |
| 341 | /// \brief A parser that gathers additional information about tokens. |
| 342 | /// |
| 343 | /// The \c TokenAnnotator tries to matches parenthesis and square brakets and |
| 344 | /// store a parenthesis levels. It also tries to resolve matching "<" and ">" |
| 345 | /// into template parameter lists. |
| 346 | class AnnotatingParser { |
| 347 | public: |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 348 | AnnotatingParser(const SmallVector<FormatToken, 16> &Tokens, |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 349 | std::vector<TokenAnnotation> &Annotations) |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 350 | : Tokens(Tokens), |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 351 | Annotations(Annotations), |
| 352 | Index(0) { |
| 353 | } |
| 354 | |
| 355 | bool parseAngle(unsigned Level) { |
| 356 | while (Index < Tokens.size()) { |
| 357 | if (Tokens[Index].Tok.is(tok::greater)) { |
| 358 | Annotations[Index].Type = TokenAnnotation::TT_TemplateOpener; |
| 359 | Annotations[Index].ParenLevel = Level; |
| 360 | next(); |
| 361 | return true; |
| 362 | } |
| 363 | if (Tokens[Index].Tok.is(tok::r_paren) || |
| 364 | Tokens[Index].Tok.is(tok::r_square)) |
| 365 | return false; |
| 366 | if (Tokens[Index].Tok.is(tok::pipepipe) || |
| 367 | Tokens[Index].Tok.is(tok::ampamp) || |
| 368 | Tokens[Index].Tok.is(tok::question) || |
| 369 | Tokens[Index].Tok.is(tok::colon)) |
| 370 | return false; |
| 371 | consumeToken(Level); |
| 372 | } |
| 373 | return false; |
| 374 | } |
| 375 | |
| 376 | bool parseParens(unsigned Level) { |
| 377 | while (Index < Tokens.size()) { |
| 378 | if (Tokens[Index].Tok.is(tok::r_paren)) { |
| 379 | Annotations[Index].ParenLevel = Level; |
| 380 | next(); |
| 381 | return true; |
| 382 | } |
| 383 | if (Tokens[Index].Tok.is(tok::r_square)) |
| 384 | return false; |
| 385 | consumeToken(Level); |
| 386 | } |
| 387 | return false; |
| 388 | } |
| 389 | |
| 390 | bool parseSquare(unsigned Level) { |
| 391 | while (Index < Tokens.size()) { |
| 392 | if (Tokens[Index].Tok.is(tok::r_square)) { |
| 393 | Annotations[Index].ParenLevel = Level; |
| 394 | next(); |
| 395 | return true; |
| 396 | } |
| 397 | if (Tokens[Index].Tok.is(tok::r_paren)) |
| 398 | return false; |
| 399 | consumeToken(Level); |
| 400 | } |
| 401 | return false; |
| 402 | } |
| 403 | |
| 404 | bool parseConditional(unsigned Level) { |
| 405 | while (Index < Tokens.size()) { |
| 406 | if (Tokens[Index].Tok.is(tok::colon)) { |
| 407 | Annotations[Index].Type = TokenAnnotation::TT_ConditionalExpr; |
| 408 | next(); |
| 409 | return true; |
| 410 | } |
| 411 | consumeToken(Level); |
| 412 | } |
| 413 | return false; |
| 414 | } |
| 415 | |
| 416 | void consumeToken(unsigned Level) { |
| 417 | Annotations[Index].ParenLevel = Level; |
| 418 | unsigned CurrentIndex = Index; |
| 419 | next(); |
| 420 | switch (Tokens[CurrentIndex].Tok.getKind()) { |
| 421 | case tok::l_paren: |
| 422 | parseParens(Level + 1); |
| 423 | break; |
| 424 | case tok::l_square: |
| 425 | parseSquare(Level + 1); |
| 426 | break; |
| 427 | case tok::less: |
| 428 | if (parseAngle(Level + 1)) |
| 429 | Annotations[CurrentIndex].Type = TokenAnnotation::TT_TemplateOpener; |
| 430 | else { |
| 431 | Annotations[CurrentIndex].Type = TokenAnnotation::TT_BinaryOperator; |
| 432 | Index = CurrentIndex + 1; |
| 433 | } |
| 434 | break; |
| 435 | case tok::greater: |
| 436 | Annotations[CurrentIndex].Type = TokenAnnotation::TT_BinaryOperator; |
| 437 | break; |
| 438 | case tok::kw_operator: |
| 439 | if (!Tokens[Index].Tok.is(tok::l_paren)) |
| 440 | Annotations[Index].Type = TokenAnnotation::TT_OverloadedOperator; |
| 441 | next(); |
| 442 | break; |
| 443 | case tok::question: |
| 444 | parseConditional(Level); |
| 445 | break; |
| 446 | default: |
| 447 | break; |
| 448 | } |
| 449 | } |
| 450 | |
| 451 | void parseLine() { |
| 452 | while (Index < Tokens.size()) { |
| 453 | consumeToken(0); |
| 454 | } |
| 455 | } |
| 456 | |
| 457 | void next() { |
| 458 | ++Index; |
| 459 | } |
| 460 | |
| 461 | private: |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 462 | const SmallVector<FormatToken, 16> &Tokens; |
| 463 | std::vector<TokenAnnotation> &Annotations; |
| 464 | unsigned Index; |
| 465 | }; |
| 466 | |
| 467 | void annotate() { |
| 468 | Annotations.clear(); |
| 469 | for (int i = 0, e = Line.Tokens.size(); i != e; ++i) { |
| 470 | Annotations.push_back(TokenAnnotation()); |
| 471 | } |
| 472 | |
Manuel Klimek | 0be4b36 | 2012-12-03 20:55:42 +0000 | [diff] [blame^] | 473 | AnnotatingParser Parser(Line.Tokens, Annotations); |
Daniel Jasper | bac016b | 2012-12-03 18:12:45 +0000 | [diff] [blame] | 474 | Parser.parseLine(); |
| 475 | |
| 476 | determineTokenTypes(); |
| 477 | |
| 478 | for (int i = 1, e = Line.Tokens.size(); i != e; ++i) { |
| 479 | TokenAnnotation &Annotation = Annotations[i]; |
| 480 | |
| 481 | Annotation.CanBreakBefore = |
| 482 | canBreakBetween(Line.Tokens[i - 1], Line.Tokens[i]); |
| 483 | |
| 484 | if (Line.Tokens[i].Tok.is(tok::colon)) { |
| 485 | if (Line.Tokens[0].Tok.is(tok::kw_case) || i == e - 1) { |
| 486 | Annotation.SpaceRequiredBefore = false; |
| 487 | } else { |
| 488 | Annotation.SpaceRequiredBefore = TokenAnnotation::TT_ConditionalExpr; |
| 489 | } |
| 490 | } else if (Annotations[i - 1].Type == TokenAnnotation::TT_UnaryOperator) { |
| 491 | Annotation.SpaceRequiredBefore = false; |
| 492 | } else if (Annotation.Type == TokenAnnotation::TT_UnaryOperator) { |
| 493 | Annotation.SpaceRequiredBefore = |
| 494 | Line.Tokens[i - 1].Tok.isNot(tok::l_paren); |
| 495 | } else if (Line.Tokens[i - 1].Tok.is(tok::greater) && |
| 496 | Line.Tokens[i].Tok.is(tok::greater)) { |
| 497 | if (Annotation.Type == TokenAnnotation::TT_TemplateOpener && |
| 498 | Annotations[i - 1].Type == TokenAnnotation::TT_TemplateOpener) |
| 499 | Annotation.SpaceRequiredBefore = Style.SplitTemplateClosingGreater; |
| 500 | else |
| 501 | Annotation.SpaceRequiredBefore = false; |
| 502 | } else if ( |
| 503 | Annotation.Type == TokenAnnotation::TT_BinaryOperator || |
| 504 | Annotations[i - 1].Type == TokenAnnotation::TT_BinaryOperator) { |
| 505 | Annotation.SpaceRequiredBefore = true; |
| 506 | } else if ( |
| 507 | Annotations[i - 1].Type == TokenAnnotation::TT_TemplateOpener && |
| 508 | Line.Tokens[i].Tok.is(tok::l_paren)) { |
| 509 | Annotation.SpaceRequiredBefore = false; |
| 510 | } else { |
| 511 | Annotation.SpaceRequiredBefore = |
| 512 | spaceRequiredBetween(Line.Tokens[i - 1].Tok, Line.Tokens[i].Tok); |
| 513 | } |
| 514 | |
| 515 | if (Annotations[i - 1].Type == TokenAnnotation::TT_LineComment || |
| 516 | (Line.Tokens[i].Tok.is(tok::string_literal) && |
| 517 | Line.Tokens[i - 1].Tok.is(tok::string_literal))) { |
| 518 | Annotation.MustBreakBefore = true; |
| 519 | } |
| 520 | |
| 521 | if (Annotation.MustBreakBefore) |
| 522 | Annotation.CanBreakBefore = true; |
| 523 | } |
| 524 | } |
| 525 | |
| 526 | const std::vector<TokenAnnotation> &getAnnotations() { |
| 527 | return Annotations; |
| 528 | } |
| 529 | |
| 530 | private: |
| 531 | void determineTokenTypes() { |
| 532 | for (int i = 0, e = Line.Tokens.size(); i != e; ++i) { |
| 533 | TokenAnnotation &Annotation = Annotations[i]; |
| 534 | const FormatToken &Tok = Line.Tokens[i]; |
| 535 | |
| 536 | if (Tok.Tok.is(tok::star) || Tok.Tok.is(tok::amp)) |
| 537 | Annotation.Type = determineStarAmpUsage(i); |
| 538 | else if (Tok.Tok.is(tok::minus) && Line.Tokens[i - 1].Tok.is(tok::equal)) |
| 539 | Annotation.Type = TokenAnnotation::TT_UnaryOperator; |
| 540 | else if (isBinaryOperator(Line.Tokens[i])) |
| 541 | Annotation.Type = TokenAnnotation::TT_BinaryOperator; |
| 542 | else if (Tok.Tok.is(tok::comment)) { |
| 543 | StringRef Data(SourceMgr.getCharacterData(Tok.Tok.getLocation()), |
| 544 | Tok.Tok.getLength()); |
| 545 | if (Data.startswith("//")) |
| 546 | Annotation.Type = TokenAnnotation::TT_LineComment; |
| 547 | else |
| 548 | Annotation.Type = TokenAnnotation::TT_BlockComment; |
| 549 | } |
| 550 | } |
| 551 | } |
| 552 | |
| 553 | bool isBinaryOperator(const FormatToken &Tok) { |
| 554 | switch (Tok.Tok.getKind()) { |
| 555 | case tok::equal: |
| 556 | case tok::equalequal: |
| 557 | case tok::star: |
| 558 | //case tok::amp: |
| 559 | case tok::plus: |
| 560 | case tok::slash: |
| 561 | case tok::minus: |
| 562 | case tok::ampamp: |
| 563 | case tok::pipe: |
| 564 | case tok::pipepipe: |
| 565 | case tok::percent: |
| 566 | return true; |
| 567 | default: |
| 568 | return false; |
| 569 | } |
| 570 | } |
| 571 | |
| 572 | TokenAnnotation::TokenType determineStarAmpUsage(unsigned Index) { |
| 573 | if (Index == Annotations.size()) |
| 574 | return TokenAnnotation::TT_Unknown; |
| 575 | |
| 576 | if (Index == 0 || Line.Tokens[Index - 1].Tok.is(tok::l_paren) || |
| 577 | Line.Tokens[Index - 1].Tok.is(tok::comma) || |
| 578 | Annotations[Index - 1].Type == TokenAnnotation::TT_BinaryOperator) |
| 579 | return TokenAnnotation::TT_UnaryOperator; |
| 580 | |
| 581 | if (Line.Tokens[Index - 1].Tok.isLiteral() || |
| 582 | Line.Tokens[Index + 1].Tok.isLiteral()) |
| 583 | return TokenAnnotation::TT_BinaryOperator; |
| 584 | |
| 585 | return TokenAnnotation::TT_PointerOrReference; |
| 586 | } |
| 587 | |
| 588 | bool isIfForOrWhile(Token Tok) { |
| 589 | return Tok.is(tok::kw_if) || Tok.is(tok::kw_for) || Tok.is(tok::kw_while); |
| 590 | } |
| 591 | |
| 592 | bool spaceRequiredBetween(Token Left, Token Right) { |
| 593 | if (Left.is(tok::kw_template) && Right.is(tok::less)) |
| 594 | return true; |
| 595 | if (Left.is(tok::arrow) || Right.is(tok::arrow)) |
| 596 | return false; |
| 597 | if (Left.is(tok::exclaim) || Left.is(tok::tilde)) |
| 598 | return false; |
| 599 | if (Left.is(tok::less) || Right.is(tok::greater) || Right.is(tok::less)) |
| 600 | return false; |
| 601 | if (Left.is(tok::amp) || Left.is(tok::star)) |
| 602 | return Right.isLiteral() || Style.PointerAndReferenceBindToType; |
| 603 | if (Right.is(tok::star) && Left.is(tok::l_paren)) |
| 604 | return false; |
| 605 | if (Right.is(tok::amp) || Right.is(tok::star)) |
| 606 | return Left.isLiteral() || !Style.PointerAndReferenceBindToType; |
| 607 | if (Left.is(tok::l_square) || Right.is(tok::l_square) || |
| 608 | Right.is(tok::r_square)) |
| 609 | return false; |
| 610 | if (Left.is(tok::coloncolon) || Right.is(tok::coloncolon)) |
| 611 | return false; |
| 612 | if (Left.is(tok::period) || Right.is(tok::period)) |
| 613 | return false; |
| 614 | if (Left.is(tok::colon) || Right.is(tok::colon)) |
| 615 | return true; |
| 616 | if ((Left.is(tok::plusplus) && Right.isAnyIdentifier()) || |
| 617 | (Left.isAnyIdentifier() && Right.is(tok::plusplus)) || |
| 618 | (Left.is(tok::minusminus) && Right.isAnyIdentifier()) || |
| 619 | (Left.isAnyIdentifier() && Right.is(tok::minusminus))) |
| 620 | return false; |
| 621 | if (Left.is(tok::l_paren)) |
| 622 | return false; |
| 623 | if (Left.is(tok::hash)) |
| 624 | return false; |
| 625 | if (Right.is(tok::r_paren) || Right.is(tok::semi) || Right.is(tok::comma)) |
| 626 | return false; |
| 627 | if (Right.is(tok::l_paren)) { |
| 628 | return !Left.isAnyIdentifier() || isIfForOrWhile(Left); |
| 629 | } |
| 630 | return true; |
| 631 | } |
| 632 | |
| 633 | bool canBreakBetween(const FormatToken &Left, const FormatToken &Right) { |
| 634 | if (Right.Tok.is(tok::r_paren)) |
| 635 | return false; |
| 636 | if (isBinaryOperator(Left)) |
| 637 | return true; |
| 638 | return Right.Tok.is(tok::colon) || Left.Tok.is(tok::comma) || Left.Tok.is( |
| 639 | tok::semi) || Left.Tok.is(tok::equal) || Left.Tok.is(tok::ampamp) || |
| 640 | (Left.Tok.is(tok::l_paren) && !Right.Tok.is(tok::r_paren)); |
| 641 | } |
| 642 | |
| 643 | const UnwrappedLine &Line; |
| 644 | FormatStyle Style; |
| 645 | SourceManager &SourceMgr; |
| 646 | std::vector<TokenAnnotation> Annotations; |
| 647 | }; |
| 648 | |
| 649 | class Formatter : public UnwrappedLineConsumer { |
| 650 | public: |
| 651 | Formatter(const FormatStyle &Style, Lexer &Lex, SourceManager &SourceMgr, |
| 652 | const std::vector<CharSourceRange> &Ranges) |
| 653 | : Style(Style), |
| 654 | Lex(Lex), |
| 655 | SourceMgr(SourceMgr), |
| 656 | Ranges(Ranges) { |
| 657 | } |
| 658 | |
| 659 | tooling::Replacements format() { |
| 660 | UnwrappedLineParser Parser(Lex, SourceMgr, *this); |
| 661 | Parser.parse(); |
| 662 | return Replaces; |
| 663 | } |
| 664 | |
| 665 | private: |
| 666 | virtual void formatUnwrappedLine(const UnwrappedLine &TheLine) { |
| 667 | if (TheLine.Tokens.size() == 0) |
| 668 | return; |
| 669 | |
| 670 | CharSourceRange LineRange = |
| 671 | CharSourceRange::getTokenRange(TheLine.Tokens.front().Tok.getLocation(), |
| 672 | TheLine.Tokens.back().Tok.getLocation()); |
| 673 | |
| 674 | for (unsigned i = 0, e = Ranges.size(); i != e; ++i) { |
| 675 | if (SourceMgr.isBeforeInTranslationUnit(LineRange.getEnd(), |
| 676 | Ranges[i].getBegin()) || |
| 677 | SourceMgr.isBeforeInTranslationUnit(Ranges[i].getEnd(), |
| 678 | LineRange.getBegin())) |
| 679 | continue; |
| 680 | |
| 681 | TokenAnnotator Annotator(TheLine, Style, SourceMgr); |
| 682 | Annotator.annotate(); |
| 683 | UnwrappedLineFormatter Formatter(Style, SourceMgr, TheLine, |
| 684 | Annotator.getAnnotations(), Replaces); |
| 685 | Formatter.format(); |
| 686 | return; |
| 687 | } |
| 688 | } |
| 689 | |
| 690 | FormatStyle Style; |
| 691 | Lexer &Lex; |
| 692 | SourceManager &SourceMgr; |
| 693 | tooling::Replacements Replaces; |
| 694 | std::vector<CharSourceRange> Ranges; |
| 695 | }; |
| 696 | |
| 697 | tooling::Replacements reformat(const FormatStyle &Style, Lexer &Lex, |
| 698 | SourceManager &SourceMgr, |
| 699 | std::vector<CharSourceRange> Ranges) { |
| 700 | Formatter formatter(Style, Lex, SourceMgr, Ranges); |
| 701 | return formatter.format(); |
| 702 | } |
| 703 | |
| 704 | } // namespace format |
| 705 | } // namespace clang |