| 1 | //===--- UnwrappedLineParser.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 UnwrappedLineParser, |
| 11 | /// which turns a stream of tokens into UnwrappedLines. |
| 12 | /// |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | #ifndef LLVM_CLANG_LIB_FORMAT_UNWRAPPEDLINEPARSER_H |
| 16 | #define LLVM_CLANG_LIB_FORMAT_UNWRAPPEDLINEPARSER_H |
| 17 | |
| 18 | #include "FormatToken.h" |
| 19 | #include "clang/Basic/IdentifierTable.h" |
| 20 | #include "clang/Format/Format.h" |
| 21 | #include "llvm/Support/Regex.h" |
| 22 | #include <list> |
| 23 | #include <stack> |
| 24 | |
| 25 | namespace clang { |
| 26 | namespace format { |
| 27 | |
| 28 | struct UnwrappedLineNode; |
| 29 | |
| 30 | /// An unwrapped line is a sequence of \c Token, that we would like to |
| 31 | /// put on a single line if there was no column limit. |
| 32 | /// |
| 33 | /// This is used as a main interface between the \c UnwrappedLineParser and the |
| 34 | /// \c UnwrappedLineFormatter. The key property is that changing the formatting |
| 35 | /// within an unwrapped line does not affect any other unwrapped lines. |
| 36 | struct UnwrappedLine { |
| 37 | UnwrappedLine(); |
| 38 | |
| 39 | // FIXME: Don't use std::list here. |
| 40 | /// The \c Tokens comprising this \c UnwrappedLine. |
| 41 | std::list<UnwrappedLineNode> Tokens; |
| 42 | |
| 43 | /// The indent level of the \c UnwrappedLine. |
| 44 | unsigned Level; |
| 45 | |
| 46 | /// Whether this \c UnwrappedLine is part of a preprocessor directive. |
| 47 | bool InPPDirective; |
| 48 | |
| 49 | bool MustBeDeclaration; |
| 50 | |
| 51 | /// If this \c UnwrappedLine closes a block in a sequence of lines, |
| 52 | /// \c MatchingOpeningBlockLineIndex stores the index of the corresponding |
| 53 | /// opening line. Otherwise, \c MatchingOpeningBlockLineIndex must be |
| 54 | /// \c kInvalidIndex. |
| 55 | size_t MatchingOpeningBlockLineIndex = kInvalidIndex; |
| 56 | |
| 57 | /// If this \c UnwrappedLine opens a block, stores the index of the |
| 58 | /// line with the corresponding closing brace. |
| 59 | size_t MatchingClosingBlockLineIndex = kInvalidIndex; |
| 60 | |
| 61 | static const size_t kInvalidIndex = -1; |
| 62 | |
| 63 | unsigned FirstStartColumn = 0; |
| 64 | }; |
| 65 | |
| 66 | class UnwrappedLineConsumer { |
| 67 | public: |
| 68 | virtual ~UnwrappedLineConsumer() {} |
| 69 | virtual void consumeUnwrappedLine(const UnwrappedLine &Line) = 0; |
| 70 | virtual void finishRun() = 0; |
| 71 | }; |
| 72 | |
| 73 | class FormatTokenSource; |
| 74 | |
| 75 | class UnwrappedLineParser { |
| 76 | public: |
| 77 | UnwrappedLineParser(const FormatStyle &Style, |
| 78 | const AdditionalKeywords &Keywords, |
| 79 | unsigned FirstStartColumn, ArrayRef<FormatToken *> Tokens, |
| 80 | UnwrappedLineConsumer &Callback); |
| 81 | |
| 82 | void parse(); |
| 83 | |
| 84 | private: |
| 85 | void reset(); |
| 86 | void parseFile(); |
| 87 | void parseLevel(bool HasOpeningBrace); |
| 88 | void parseBlock(bool MustBeDeclaration, bool AddLevel = true, |
| 89 | bool MunchSemi = true); |
| 90 | void parseChildBlock(); |
| 91 | void parsePPDirective(); |
| 92 | void parsePPDefine(); |
| 93 | void parsePPIf(bool IfDef); |
| 94 | void parsePPElIf(); |
| 95 | void parsePPElse(); |
| 96 | void parsePPEndIf(); |
| 97 | void parsePPUnknown(); |
| 98 | void readTokenWithJavaScriptASI(); |
| 99 | void parseStructuralElement(); |
| 100 | bool tryToParseBracedList(); |
| 101 | bool parseBracedList(bool ContinueOnSemicolons = false, |
| 102 | tok::TokenKind ClosingBraceKind = tok::r_brace); |
| 103 | void parseParens(); |
| 104 | void parseSquare(bool LambdaIntroducer = false); |
| 105 | void parseIfThenElse(); |
| 106 | void parseTryCatch(); |
| 107 | void parseForOrWhileLoop(); |
| 108 | void parseDoWhile(); |
| 109 | void parseLabel(); |
| 110 | void parseCaseLabel(); |
| 111 | void parseSwitch(); |
| 112 | void parseNamespace(); |
| 113 | void parseNew(); |
| 114 | void parseAccessSpecifier(); |
| 115 | bool parseEnum(); |
| 116 | void parseJavaEnumBody(); |
| 117 | // Parses a record (aka class) as a top level element. If ParseAsExpr is true, |
| 118 | // parses the record as a child block, i.e. if the class declaration is an |
| 119 | // expression. |
| 120 | void parseRecord(bool ParseAsExpr = false); |
| 121 | void parseObjCMethod(); |
| 122 | void parseObjCProtocolList(); |
| 123 | void parseObjCUntilAtEnd(); |
| 124 | void parseObjCInterfaceOrImplementation(); |
| 125 | bool parseObjCProtocol(); |
| 126 | void parseJavaScriptEs6ImportExport(); |
| 127 | void parseStatementMacro(); |
| 128 | bool tryToParseLambda(); |
| 129 | bool tryToParseLambdaIntroducer(); |
| 130 | void tryToParseJSFunction(); |
| 131 | void addUnwrappedLine(); |
| 132 | bool eof() const; |
| 133 | // LevelDifference is the difference of levels after and before the current |
| 134 | // token. For example: |
| 135 | // - if the token is '{' and opens a block, LevelDifference is 1. |
| 136 | // - if the token is '}' and closes a block, LevelDifference is -1. |
| 137 | void nextToken(int LevelDifference = 0); |
| 138 | void readToken(int LevelDifference = 0); |
| 139 | |
| 140 | // Decides which comment tokens should be added to the current line and which |
| 141 | // should be added as comments before the next token. |
| 142 | // |
| 143 | // Comments specifies the sequence of comment tokens to analyze. They get |
| 144 | // either pushed to the current line or added to the comments before the next |
| 145 | // token. |
| 146 | // |
| 147 | // NextTok specifies the next token. A null pointer NextTok is supported, and |
| 148 | // signifies either the absence of a next token, or that the next token |
| 149 | // shouldn't be taken into accunt for the analysis. |
| 150 | void distributeComments(const SmallVectorImpl<FormatToken *> &Comments, |
| 151 | const FormatToken *NextTok); |
| 152 | |
| 153 | // Adds the comment preceding the next token to unwrapped lines. |
| 154 | void flushComments(bool NewlineBeforeNext); |
| 155 | void pushToken(FormatToken *Tok); |
| 156 | void calculateBraceTypes(bool ExpectClassBody = false); |
| 157 | |
| 158 | // Marks a conditional compilation edge (for example, an '#if', '#ifdef', |
| 159 | // '#else' or merge conflict marker). If 'Unreachable' is true, assumes |
| 160 | // this branch either cannot be taken (for example '#if false'), or should |
| 161 | // not be taken in this round. |
| 162 | void conditionalCompilationCondition(bool Unreachable); |
| 163 | void conditionalCompilationStart(bool Unreachable); |
| 164 | void conditionalCompilationAlternative(); |
| 165 | void conditionalCompilationEnd(); |
| 166 | |
| 167 | bool isOnNewLine(const FormatToken &FormatTok); |
| 168 | |
| 169 | // Compute hash of the current preprocessor branch. |
| 170 | // This is used to identify the different branches, and thus track if block |
| 171 | // open and close in the same branch. |
| 172 | size_t computePPHash() const; |
| 173 | |
| 174 | // FIXME: We are constantly running into bugs where Line.Level is incorrectly |
| 175 | // subtracted from beyond 0. Introduce a method to subtract from Line.Level |
| 176 | // and use that everywhere in the Parser. |
| 177 | std::unique_ptr<UnwrappedLine> Line; |
| 178 | |
| 179 | // Comments are sorted into unwrapped lines by whether they are in the same |
| 180 | // line as the previous token, or not. If not, they belong to the next token. |
| 181 | // Since the next token might already be in a new unwrapped line, we need to |
| 182 | // store the comments belonging to that token. |
| 183 | SmallVector<FormatToken *, 1> CommentsBeforeNextToken; |
| 184 | FormatToken *FormatTok; |
| 185 | bool MustBreakBeforeNextToken; |
| 186 | |
| 187 | // The parsed lines. Only added to through \c CurrentLines. |
| 188 | SmallVector<UnwrappedLine, 8> Lines; |
| 189 | |
| 190 | // Preprocessor directives are parsed out-of-order from other unwrapped lines. |
| 191 | // Thus, we need to keep a list of preprocessor directives to be reported |
| 192 | // after an unwrapped line that has been started was finished. |
| 193 | SmallVector<UnwrappedLine, 4> PreprocessorDirectives; |
| 194 | |
| 195 | // New unwrapped lines are added via CurrentLines. |
| 196 | // Usually points to \c &Lines. While parsing a preprocessor directive when |
| 197 | // there is an unfinished previous unwrapped line, will point to |
| 198 | // \c &PreprocessorDirectives. |
| 199 | SmallVectorImpl<UnwrappedLine> *CurrentLines; |
| 200 | |
| 201 | // We store for each line whether it must be a declaration depending on |
| 202 | // whether we are in a compound statement or not. |
| 203 | std::vector<bool> DeclarationScopeStack; |
| 204 | |
| 205 | const FormatStyle &Style; |
| 206 | const AdditionalKeywords &Keywords; |
| 207 | |
| 208 | llvm::Regex CommentPragmasRegex; |
| 209 | |
| 210 | FormatTokenSource *Tokens; |
| 211 | UnwrappedLineConsumer &Callback; |
| 212 | |
| 213 | // FIXME: This is a temporary measure until we have reworked the ownership |
| 214 | // of the format tokens. The goal is to have the actual tokens created and |
| 215 | // owned outside of and handed into the UnwrappedLineParser. |
| 216 | ArrayRef<FormatToken *> AllTokens; |
| 217 | |
| 218 | // Represents preprocessor branch type, so we can find matching |
| 219 | // #if/#else/#endif directives. |
| 220 | enum PPBranchKind { |
| 221 | PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0 |
| 222 | PP_Unreachable // #if 0 or a conditional preprocessor block inside #if 0 |
| 223 | }; |
| 224 | |
| 225 | struct PPBranch { |
| 226 | PPBranch(PPBranchKind Kind, size_t Line) : Kind(Kind), Line(Line) {} |
| 227 | PPBranchKind Kind; |
| 228 | size_t Line; |
| 229 | }; |
| 230 | |
| 231 | // Keeps a stack of currently active preprocessor branching directives. |
| 232 | SmallVector<PPBranch, 16> PPStack; |
| 233 | |
| 234 | // The \c UnwrappedLineParser re-parses the code for each combination |
| 235 | // of preprocessor branches that can be taken. |
| 236 | // To that end, we take the same branch (#if, #else, or one of the #elif |
| 237 | // branches) for each nesting level of preprocessor branches. |
| 238 | // \c PPBranchLevel stores the current nesting level of preprocessor |
| 239 | // branches during one pass over the code. |
| 240 | int PPBranchLevel; |
| 241 | |
| 242 | // Contains the current branch (#if, #else or one of the #elif branches) |
| 243 | // for each nesting level. |
| 244 | SmallVector<int, 8> PPLevelBranchIndex; |
| 245 | |
| 246 | // Contains the maximum number of branches at each nesting level. |
| 247 | SmallVector<int, 8> PPLevelBranchCount; |
| 248 | |
| 249 | // Contains the number of branches per nesting level we are currently |
| 250 | // in while parsing a preprocessor branch sequence. |
| 251 | // This is used to update PPLevelBranchCount at the end of a branch |
| 252 | // sequence. |
| 253 | std::stack<int> PPChainBranchIndex; |
| 254 | |
| 255 | // Include guard search state. Used to fixup preprocessor indent levels |
| 256 | // so that include guards do not participate in indentation. |
| 257 | enum IncludeGuardState { |
| 258 | IG_Inited, // Search started, looking for #ifndef. |
| 259 | IG_IfNdefed, // #ifndef found, IncludeGuardToken points to condition. |
| 260 | IG_Defined, // Matching #define found, checking other requirements. |
| 261 | IG_Found, // All requirements met, need to fix indents. |
| 262 | IG_Rejected, // Search failed or never started. |
| 263 | }; |
| 264 | |
| 265 | // Current state of include guard search. |
| 266 | IncludeGuardState IncludeGuard; |
| 267 | |
| 268 | // Points to the #ifndef condition for a potential include guard. Null unless |
| 269 | // IncludeGuardState == IG_IfNdefed. |
| 270 | FormatToken *IncludeGuardToken; |
| 271 | |
| 272 | // Contains the first start column where the source begins. This is zero for |
| 273 | // normal source code and may be nonzero when formatting a code fragment that |
| 274 | // does not start at the beginning of the file. |
| 275 | unsigned FirstStartColumn; |
| 276 | |
| 277 | friend class ScopedLineState; |
| 278 | friend class CompoundStatementIndenter; |
| 279 | }; |
| 280 | |
| 281 | struct UnwrappedLineNode { |
| 282 | UnwrappedLineNode() : Tok(nullptr) {} |
| 283 | UnwrappedLineNode(FormatToken *Tok) : Tok(Tok) {} |
| 284 | |
| 285 | FormatToken *Tok; |
| 286 | SmallVector<UnwrappedLine, 0> Children; |
| 287 | }; |
| 288 | |
| 289 | inline UnwrappedLine::UnwrappedLine() |
| 290 | : Level(0), InPPDirective(false), MustBeDeclaration(false), |
| 291 | MatchingOpeningBlockLineIndex(kInvalidIndex) {} |
| 292 | |
| 293 | } // end namespace format |
| 294 | } // end namespace clang |
| 295 | |
| 296 | #endif |
| 297 | |