clang  11.0.0git
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:
78  const AdditionalKeywords &Keywords,
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, bool IsEnum = 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(bool LeftAlignLabel = false);
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  void parseCSharpAttribute();
129  // Parse a C# generic type constraint: `where T : IComparable<T>`.
130  // See:
131  // https://docs.microsoft.com/en-us/dotnet/csharp/language-reference/keywords/where-generic-type-constraint
132  void parseCSharpGenericTypeConstraint();
133  bool tryToParseLambda();
134  bool tryToParseLambdaIntroducer();
135  bool tryToParsePropertyAccessor();
136  void tryToParseJSFunction();
137  bool tryToParseSimpleAttribute();
138  void addUnwrappedLine();
139  bool eof() const;
140  // LevelDifference is the difference of levels after and before the current
141  // token. For example:
142  // - if the token is '{' and opens a block, LevelDifference is 1.
143  // - if the token is '}' and closes a block, LevelDifference is -1.
144  void nextToken(int LevelDifference = 0);
145  void readToken(int LevelDifference = 0);
146 
147  // Decides which comment tokens should be added to the current line and which
148  // should be added as comments before the next token.
149  //
150  // Comments specifies the sequence of comment tokens to analyze. They get
151  // either pushed to the current line or added to the comments before the next
152  // token.
153  //
154  // NextTok specifies the next token. A null pointer NextTok is supported, and
155  // signifies either the absence of a next token, or that the next token
156  // shouldn't be taken into accunt for the analysis.
157  void distributeComments(const SmallVectorImpl<FormatToken *> &Comments,
158  const FormatToken *NextTok);
159 
160  // Adds the comment preceding the next token to unwrapped lines.
161  void flushComments(bool NewlineBeforeNext);
162  void pushToken(FormatToken *Tok);
163  void calculateBraceTypes(bool ExpectClassBody = false);
164 
165  // Marks a conditional compilation edge (for example, an '#if', '#ifdef',
166  // '#else' or merge conflict marker). If 'Unreachable' is true, assumes
167  // this branch either cannot be taken (for example '#if false'), or should
168  // not be taken in this round.
169  void conditionalCompilationCondition(bool Unreachable);
170  void conditionalCompilationStart(bool Unreachable);
171  void conditionalCompilationAlternative();
172  void conditionalCompilationEnd();
173 
174  bool isOnNewLine(const FormatToken &FormatTok);
175 
176  // Compute hash of the current preprocessor branch.
177  // This is used to identify the different branches, and thus track if block
178  // open and close in the same branch.
179  size_t computePPHash() const;
180 
181  // FIXME: We are constantly running into bugs where Line.Level is incorrectly
182  // subtracted from beyond 0. Introduce a method to subtract from Line.Level
183  // and use that everywhere in the Parser.
184  std::unique_ptr<UnwrappedLine> Line;
185 
186  // Comments are sorted into unwrapped lines by whether they are in the same
187  // line as the previous token, or not. If not, they belong to the next token.
188  // Since the next token might already be in a new unwrapped line, we need to
189  // store the comments belonging to that token.
190  SmallVector<FormatToken *, 1> CommentsBeforeNextToken;
191  FormatToken *FormatTok;
192  bool MustBreakBeforeNextToken;
193 
194  // The parsed lines. Only added to through \c CurrentLines.
196 
197  // Preprocessor directives are parsed out-of-order from other unwrapped lines.
198  // Thus, we need to keep a list of preprocessor directives to be reported
199  // after an unwrapped line that has been started was finished.
200  SmallVector<UnwrappedLine, 4> PreprocessorDirectives;
201 
202  // New unwrapped lines are added via CurrentLines.
203  // Usually points to \c &Lines. While parsing a preprocessor directive when
204  // there is an unfinished previous unwrapped line, will point to
205  // \c &PreprocessorDirectives.
206  SmallVectorImpl<UnwrappedLine> *CurrentLines;
207 
208  // We store for each line whether it must be a declaration depending on
209  // whether we are in a compound statement or not.
210  std::vector<bool> DeclarationScopeStack;
211 
212  const FormatStyle &Style;
213  const AdditionalKeywords &Keywords;
214 
215  llvm::Regex CommentPragmasRegex;
216 
218  UnwrappedLineConsumer &Callback;
219 
220  // FIXME: This is a temporary measure until we have reworked the ownership
221  // of the format tokens. The goal is to have the actual tokens created and
222  // owned outside of and handed into the UnwrappedLineParser.
223  ArrayRef<FormatToken *> AllTokens;
224 
225  // Represents preprocessor branch type, so we can find matching
226  // #if/#else/#endif directives.
227  enum PPBranchKind {
228  PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0
229  PP_Unreachable // #if 0 or a conditional preprocessor block inside #if 0
230  };
231 
232  struct PPBranch {
233  PPBranch(PPBranchKind Kind, size_t Line) : Kind(Kind), Line(Line) {}
234  PPBranchKind Kind;
235  size_t Line;
236  };
237 
238  // Keeps a stack of currently active preprocessor branching directives.
240 
241  // The \c UnwrappedLineParser re-parses the code for each combination
242  // of preprocessor branches that can be taken.
243  // To that end, we take the same branch (#if, #else, or one of the #elif
244  // branches) for each nesting level of preprocessor branches.
245  // \c PPBranchLevel stores the current nesting level of preprocessor
246  // branches during one pass over the code.
247  int PPBranchLevel;
248 
249  // Contains the current branch (#if, #else or one of the #elif branches)
250  // for each nesting level.
251  SmallVector<int, 8> PPLevelBranchIndex;
252 
253  // Contains the maximum number of branches at each nesting level.
254  SmallVector<int, 8> PPLevelBranchCount;
255 
256  // Contains the number of branches per nesting level we are currently
257  // in while parsing a preprocessor branch sequence.
258  // This is used to update PPLevelBranchCount at the end of a branch
259  // sequence.
260  std::stack<int> PPChainBranchIndex;
261 
262  // Include guard search state. Used to fixup preprocessor indent levels
263  // so that include guards do not participate in indentation.
264  enum IncludeGuardState {
265  IG_Inited, // Search started, looking for #ifndef.
266  IG_IfNdefed, // #ifndef found, IncludeGuardToken points to condition.
267  IG_Defined, // Matching #define found, checking other requirements.
268  IG_Found, // All requirements met, need to fix indents.
269  IG_Rejected, // Search failed or never started.
270  };
271 
272  // Current state of include guard search.
273  IncludeGuardState IncludeGuard;
274 
275  // Points to the #ifndef condition for a potential include guard. Null unless
276  // IncludeGuardState == IG_IfNdefed.
277  FormatToken *IncludeGuardToken;
278 
279  // Contains the first start column where the source begins. This is zero for
280  // normal source code and may be nonzero when formatting a code fragment that
281  // does not start at the beginning of the file.
282  unsigned FirstStartColumn;
283 
284  friend class ScopedLineState;
286 };
287 
289  UnwrappedLineNode() : Tok(nullptr) {}
291 
294 };
295 
299 
300 } // end namespace format
301 } // end namespace clang
302 
303 #endif
size_t MatchingClosingBlockLineIndex
If this UnwrappedLine opens a block, stores the index of the line with the corresponding closing brac...
SmallVector< UnwrappedLine, 0 > Children
unsigned Level
The indent level of the UnwrappedLine.
const FormatToken & Tok
An unwrapped line is a sequence of Token, that we would like to put on a single line if there was no ...
std::list< UnwrappedLineNode > Tokens
The Tokens comprising this UnwrappedLine.
const AnnotatedLine * Line
static const size_t kInvalidIndex
bool InPPDirective
Whether this UnwrappedLine is part of a preprocessor directive.
A wrapper around a Token storing information about the whitespace characters preceding it...
Definition: FormatToken.h:142
Defines the clang::IdentifierInfo, clang::IdentifierTable, and clang::Selector interfaces.
#define false
Definition: stdbool.h:17
Kind
Various functions to configurably format source code.
size_t MatchingOpeningBlockLineIndex
If this UnwrappedLine closes a block in a sequence of lines, MatchingOpeningBlockLineIndex stores the...
Encapsulates keywords that are context sensitive or for languages not properly supported by Clang&#39;s l...
Definition: FormatToken.h:710
TokenKind
Provides a simple uniform namespace for tokens from all C languages.
Definition: TokenKinds.h:25
The FormatStyle is used to configure the formatting to follow specific guidelines.
Definition: Format.h:54
Dataflow Directional Tag Classes.
Represents a complete lambda introducer.
Definition: DeclSpec.h:2635
This file contains the declaration of the FormatToken, a wrapper around Token with additional informa...
const FormatStyle & Style