clang  14.0.0git
UnwrappedLineParser.h
Go to the documentation of this file.
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"
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.
48 
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.
56 
57  /// If this \c UnwrappedLine opens a block, stores the index of the
58  /// line with the corresponding closing brace.
60 
61  static const size_t kInvalidIndex = -1;
62 
63  unsigned FirstStartColumn = 0;
64 };
65 
67 public:
69  virtual void consumeUnwrappedLine(const UnwrappedLine &Line) = 0;
70  virtual void finishRun() = 0;
71 };
72 
73 class FormatTokenSource;
74 
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, unsigned AddLevels = 1u,
89  bool MunchSemi = true,
90  bool UnindentWhitesmithsBraces = false);
91  void parseChildBlock();
92  void parsePPDirective();
93  void parsePPDefine();
94  void parsePPIf(bool IfDef);
95  void parsePPElIf();
96  void parsePPElse();
97  void parsePPEndIf();
98  void parsePPUnknown();
99  void readTokenWithJavaScriptASI();
100  void parseStructuralElement(bool IsTopLevel = false);
101  bool tryToParseBracedList();
102  bool parseBracedList(bool ContinueOnSemicolons = false, bool IsEnum = false,
103  tok::TokenKind ClosingBraceKind = tok::r_brace);
104  void parseParens();
105  void parseSquare(bool LambdaIntroducer = false);
106  void parseIfThenElse();
107  void parseTryCatch();
108  void parseForOrWhileLoop();
109  void parseDoWhile();
110  void parseLabel(bool LeftAlignLabel = false);
111  void parseCaseLabel();
112  void parseSwitch();
113  void parseNamespace();
114  void parseNew();
115  void parseAccessSpecifier();
116  bool parseEnum();
117  bool parseStructLike();
118  void parseConcept();
119  void parseRequires();
120  void parseRequiresExpression(unsigned int OriginalLevel);
121  void parseConstraintExpression(unsigned int OriginalLevel);
122  void parseJavaEnumBody();
123  // Parses a record (aka class) as a top level element. If ParseAsExpr is true,
124  // parses the record as a child block, i.e. if the class declaration is an
125  // expression.
126  void parseRecord(bool ParseAsExpr = false);
127  void parseObjCLightweightGenerics();
128  void parseObjCMethod();
129  void parseObjCProtocolList();
130  void parseObjCUntilAtEnd();
131  void parseObjCInterfaceOrImplementation();
132  bool parseObjCProtocol();
133  void parseJavaScriptEs6ImportExport();
134  void parseStatementMacro();
135  void parseCSharpAttribute();
136  // Parse a C# generic type constraint: `where T : IComparable<T>`.
137  // See:
138  // https://docs.microsoft.com/en-us/dotnet/csharp/language-reference/keywords/where-generic-type-constraint
139  void parseCSharpGenericTypeConstraint();
140  bool tryToParseLambda();
141  bool tryToParseLambdaIntroducer();
142  bool tryToParsePropertyAccessor();
143  void tryToParseJSFunction();
144  bool tryToParseSimpleAttribute();
145 
146  // Used by addUnwrappedLine to denote whether to keep or remove a level
147  // when resetting the line state.
148  enum class LineLevel { Remove, Keep };
149 
150  void addUnwrappedLine(LineLevel AdjustLevel = LineLevel::Remove);
151  bool eof() const;
152  // LevelDifference is the difference of levels after and before the current
153  // token. For example:
154  // - if the token is '{' and opens a block, LevelDifference is 1.
155  // - if the token is '}' and closes a block, LevelDifference is -1.
156  void nextToken(int LevelDifference = 0);
157  void readToken(int LevelDifference = 0);
158 
159  // Decides which comment tokens should be added to the current line and which
160  // should be added as comments before the next token.
161  //
162  // Comments specifies the sequence of comment tokens to analyze. They get
163  // either pushed to the current line or added to the comments before the next
164  // token.
165  //
166  // NextTok specifies the next token. A null pointer NextTok is supported, and
167  // signifies either the absence of a next token, or that the next token
168  // shouldn't be taken into accunt for the analysis.
169  void distributeComments(const SmallVectorImpl<FormatToken *> &Comments,
170  const FormatToken *NextTok);
171 
172  // Adds the comment preceding the next token to unwrapped lines.
173  void flushComments(bool NewlineBeforeNext);
174  void pushToken(FormatToken *Tok);
175  void calculateBraceTypes(bool ExpectClassBody = false);
176 
177  // Marks a conditional compilation edge (for example, an '#if', '#ifdef',
178  // '#else' or merge conflict marker). If 'Unreachable' is true, assumes
179  // this branch either cannot be taken (for example '#if false'), or should
180  // not be taken in this round.
181  void conditionalCompilationCondition(bool Unreachable);
182  void conditionalCompilationStart(bool Unreachable);
183  void conditionalCompilationAlternative();
184  void conditionalCompilationEnd();
185 
186  bool isOnNewLine(const FormatToken &FormatTok);
187 
188  // Compute hash of the current preprocessor branch.
189  // This is used to identify the different branches, and thus track if block
190  // open and close in the same branch.
191  size_t computePPHash() const;
192 
193  // FIXME: We are constantly running into bugs where Line.Level is incorrectly
194  // subtracted from beyond 0. Introduce a method to subtract from Line.Level
195  // and use that everywhere in the Parser.
196  std::unique_ptr<UnwrappedLine> Line;
197 
198  // Comments are sorted into unwrapped lines by whether they are in the same
199  // line as the previous token, or not. If not, they belong to the next token.
200  // Since the next token might already be in a new unwrapped line, we need to
201  // store the comments belonging to that token.
202  SmallVector<FormatToken *, 1> CommentsBeforeNextToken;
203  FormatToken *FormatTok;
204  bool MustBreakBeforeNextToken;
205 
206  // The parsed lines. Only added to through \c CurrentLines.
208 
209  // Preprocessor directives are parsed out-of-order from other unwrapped lines.
210  // Thus, we need to keep a list of preprocessor directives to be reported
211  // after an unwrapped line that has been started was finished.
212  SmallVector<UnwrappedLine, 4> PreprocessorDirectives;
213 
214  // New unwrapped lines are added via CurrentLines.
215  // Usually points to \c &Lines. While parsing a preprocessor directive when
216  // there is an unfinished previous unwrapped line, will point to
217  // \c &PreprocessorDirectives.
218  SmallVectorImpl<UnwrappedLine> *CurrentLines;
219 
220  // We store for each line whether it must be a declaration depending on
221  // whether we are in a compound statement or not.
222  std::vector<bool> DeclarationScopeStack;
223 
224  const FormatStyle &Style;
225  const AdditionalKeywords &Keywords;
226 
227  llvm::Regex CommentPragmasRegex;
228 
229  FormatTokenSource *Tokens;
230  UnwrappedLineConsumer &Callback;
231 
232  // FIXME: This is a temporary measure until we have reworked the ownership
233  // of the format tokens. The goal is to have the actual tokens created and
234  // owned outside of and handed into the UnwrappedLineParser.
235  ArrayRef<FormatToken *> AllTokens;
236 
237  // Represents preprocessor branch type, so we can find matching
238  // #if/#else/#endif directives.
239  enum PPBranchKind {
240  PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0
241  PP_Unreachable // #if 0 or a conditional preprocessor block inside #if 0
242  };
243 
244  struct PPBranch {
245  PPBranch(PPBranchKind Kind, size_t Line) : Kind(Kind), Line(Line) {}
246  PPBranchKind Kind;
247  size_t Line;
248  };
249 
250  // Keeps a stack of currently active preprocessor branching directives.
252 
253  // The \c UnwrappedLineParser re-parses the code for each combination
254  // of preprocessor branches that can be taken.
255  // To that end, we take the same branch (#if, #else, or one of the #elif
256  // branches) for each nesting level of preprocessor branches.
257  // \c PPBranchLevel stores the current nesting level of preprocessor
258  // branches during one pass over the code.
259  int PPBranchLevel;
260 
261  // Contains the current branch (#if, #else or one of the #elif branches)
262  // for each nesting level.
263  SmallVector<int, 8> PPLevelBranchIndex;
264 
265  // Contains the maximum number of branches at each nesting level.
266  SmallVector<int, 8> PPLevelBranchCount;
267 
268  // Contains the number of branches per nesting level we are currently
269  // in while parsing a preprocessor branch sequence.
270  // This is used to update PPLevelBranchCount at the end of a branch
271  // sequence.
272  std::stack<int> PPChainBranchIndex;
273 
274  // Include guard search state. Used to fixup preprocessor indent levels
275  // so that include guards do not participate in indentation.
276  enum IncludeGuardState {
277  IG_Inited, // Search started, looking for #ifndef.
278  IG_IfNdefed, // #ifndef found, IncludeGuardToken points to condition.
279  IG_Defined, // Matching #define found, checking other requirements.
280  IG_Found, // All requirements met, need to fix indents.
281  IG_Rejected, // Search failed or never started.
282  };
283 
284  // Current state of include guard search.
285  IncludeGuardState IncludeGuard;
286 
287  // Points to the #ifndef condition for a potential include guard. Null unless
288  // IncludeGuardState == IG_IfNdefed.
289  FormatToken *IncludeGuardToken;
290 
291  // Contains the first start column where the source begins. This is zero for
292  // normal source code and may be nonzero when formatting a code fragment that
293  // does not start at the beginning of the file.
294  unsigned FirstStartColumn;
295 
296  friend class ScopedLineState;
298 };
299 
301  UnwrappedLineNode() : Tok(nullptr) {}
303 
306 };
307 
309  : Level(0), InPPDirective(false), MustBeDeclaration(false),
310  MatchingOpeningBlockLineIndex(kInvalidIndex) {}
311 
312 } // end namespace format
313 } // end namespace clang
314 
315 #endif
clang::format::CompoundStatementIndenter
Definition: UnwrappedLineParser.cpp:172
clang::format::ScopedLineState
Definition: UnwrappedLineParser.cpp:139
clang::format::UnwrappedLineNode::Children
SmallVector< UnwrappedLine, 0 > Children
Definition: UnwrappedLineParser.h:305
clang::LambdaIntroducer
Represents a complete lambda introducer.
Definition: DeclSpec.h:2668
llvm::SmallVector
Definition: LLVM.h:38
clang::format::UnwrappedLineParser::parse
void parse()
Definition: UnwrappedLineParser.cpp:256
clang::format::UnwrappedLineConsumer::~UnwrappedLineConsumer
virtual ~UnwrappedLineConsumer()
Definition: UnwrappedLineParser.h:68
clang::format::UnwrappedLine::kInvalidIndex
static const size_t kInvalidIndex
Definition: UnwrappedLineParser.h:61
clang::format::FormatStyle
The FormatStyle is used to configure the formatting to follow specific guidelines.
Definition: Format.h:59
clang::format::FormatToken
A wrapper around a Token storing information about the whitespace characters preceding it.
Definition: FormatToken.h:210
clang::format::AdditionalKeywords
Encapsulates keywords that are context sensitive or for languages not properly supported by Clang's l...
Definition: FormatToken.h:841
clang::format::UnwrappedLine
An unwrapped line is a sequence of Token, that we would like to put on a single line if there was no ...
Definition: UnwrappedLineParser.h:36
Format.h
clang::format::UnwrappedLineParser
Definition: UnwrappedLineParser.h:75
clang::format::UnwrappedLineConsumer
Definition: UnwrappedLineParser.h:66
clang::format::UnwrappedLine::InPPDirective
bool InPPDirective
Whether this UnwrappedLine is part of a preprocessor directive.
Definition: UnwrappedLineParser.h:47
clang::format::UnwrappedLineNode::Tok
FormatToken * Tok
Definition: UnwrappedLineParser.h:304
IdentifierTable.h
Line
const AnnotatedLine * Line
Definition: UsingDeclarationsSorter.cpp:68
clang::format::UnwrappedLine::MustBeDeclaration
bool MustBeDeclaration
Definition: UnwrappedLineParser.h:49
clang::format::UnwrappedLine::MatchingClosingBlockLineIndex
size_t MatchingClosingBlockLineIndex
If this UnwrappedLine opens a block, stores the index of the line with the corresponding closing brac...
Definition: UnwrappedLineParser.h:59
clang::tok::TokenKind
TokenKind
Provides a simple uniform namespace for tokens from all C languages.
Definition: TokenKinds.h:25
false
#define false
Definition: stdbool.h:17
clang::format::UnwrappedLineNode::UnwrappedLineNode
UnwrappedLineNode(FormatToken *Tok)
Definition: UnwrappedLineParser.h:302
llvm::ArrayRef
Definition: LLVM.h:34
clang::format::UnwrappedLine::UnwrappedLine
UnwrappedLine()
Definition: UnwrappedLineParser.h:308
clang::ObjCPropertyAttribute::Kind
Kind
Definition: DeclObjCCommon.h:22
clang::format::UnwrappedLineConsumer::consumeUnwrappedLine
virtual void consumeUnwrappedLine(const UnwrappedLine &Line)=0
clang::format::UnwrappedLineNode
Definition: UnwrappedLineParser.h:300
clang
Definition: CalledOnceCheck.h:17
clang::format::UnwrappedLineConsumer::finishRun
virtual void finishRun()=0
clang::prec::Level
Level
Definition: OperatorPrecedence.h:26
clang::format::UnwrappedLineNode::UnwrappedLineNode
UnwrappedLineNode()
Definition: UnwrappedLineParser.h:301
FormatToken.h
clang::format::UnwrappedLine::Level
unsigned Level
The indent level of the UnwrappedLine.
Definition: UnwrappedLineParser.h:44
llvm::SmallVectorImpl
Definition: LLVM.h:39
clang::format::UnwrappedLine::MatchingOpeningBlockLineIndex
size_t MatchingOpeningBlockLineIndex
If this UnwrappedLine closes a block in a sequence of lines, MatchingOpeningBlockLineIndex stores the...
Definition: UnwrappedLineParser.h:55
clang::format::UnwrappedLine::Tokens
std::list< UnwrappedLineNode > Tokens
The Tokens comprising this UnwrappedLine.
Definition: UnwrappedLineParser.h:41
clang::format::UnwrappedLine::FirstStartColumn
unsigned FirstStartColumn
Definition: UnwrappedLineParser.h:63
clang::format::UnwrappedLineParser::UnwrappedLineParser
UnwrappedLineParser(const FormatStyle &Style, const AdditionalKeywords &Keywords, unsigned FirstStartColumn, ArrayRef< FormatToken * > Tokens, UnwrappedLineConsumer &Callback)
Definition: UnwrappedLineParser.cpp:225
clang::format::FormatTokenSource
Definition: UnwrappedLineParser.cpp:28