Coverage Report

Created: 2022-07-16 07:03

/Users/buildslave/jenkins/workspace/coverage/llvm-project/clang/lib/Format/UnwrappedLineParser.h
Line
Count
Source
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/ADT/BitVector.h"
22
#include "llvm/Support/Regex.h"
23
#include <list>
24
#include <stack>
25
#include <vector>
26
27
namespace clang {
28
namespace format {
29
30
struct UnwrappedLineNode;
31
32
/// An unwrapped line is a sequence of \c Token, that we would like to
33
/// put on a single line if there was no column limit.
34
///
35
/// This is used as a main interface between the \c UnwrappedLineParser and the
36
/// \c UnwrappedLineFormatter. The key property is that changing the formatting
37
/// within an unwrapped line does not affect any other unwrapped lines.
38
struct UnwrappedLine {
39
  UnwrappedLine();
40
41
  /// The \c Tokens comprising this \c UnwrappedLine.
42
  std::list<UnwrappedLineNode> Tokens;
43
44
  /// The indent level of the \c UnwrappedLine.
45
  unsigned Level;
46
47
  /// Whether this \c UnwrappedLine is part of a preprocessor directive.
48
  bool InPPDirective;
49
50
  bool MustBeDeclaration;
51
52
  /// If this \c UnwrappedLine closes a block in a sequence of lines,
53
  /// \c MatchingOpeningBlockLineIndex stores the index of the corresponding
54
  /// opening line. Otherwise, \c MatchingOpeningBlockLineIndex must be
55
  /// \c kInvalidIndex.
56
  size_t MatchingOpeningBlockLineIndex = kInvalidIndex;
57
58
  /// If this \c UnwrappedLine opens a block, stores the index of the
59
  /// line with the corresponding closing brace.
60
  size_t MatchingClosingBlockLineIndex = kInvalidIndex;
61
62
  static const size_t kInvalidIndex = -1;
63
64
  unsigned FirstStartColumn = 0;
65
};
66
67
class UnwrappedLineConsumer {
68
public:
69
65.2k
  virtual ~UnwrappedLineConsumer() {}
70
  virtual void consumeUnwrappedLine(const UnwrappedLine &Line) = 0;
71
  virtual void finishRun() = 0;
72
};
73
74
class FormatTokenSource;
75
76
class UnwrappedLineParser {
77
public:
78
  UnwrappedLineParser(const FormatStyle &Style,
79
                      const AdditionalKeywords &Keywords,
80
                      unsigned FirstStartColumn, ArrayRef<FormatToken *> Tokens,
81
                      UnwrappedLineConsumer &Callback);
82
83
  void parse();
84
85
private:
86
  enum class IfStmtKind {
87
    NotIf,   // Not an if statement.
88
    IfOnly,  // An if statement without the else clause.
89
    IfElse,  // An if statement followed by else but not else if.
90
    IfElseIf // An if statement followed by else if.
91
  };
92
93
  void reset();
94
  void parseFile();
95
  bool precededByCommentOrPPDirective() const;
96
  bool parseLevel(const FormatToken *OpeningBrace = nullptr,
97
                  bool CanContainBracedList = true,
98
                  TokenType NextLBracesType = TT_Unknown,
99
                  IfStmtKind *IfKind = nullptr,
100
                  FormatToken **IfLeftBrace = nullptr);
101
  bool mightFitOnOneLine(UnwrappedLine &Line,
102
                         const FormatToken *OpeningBrace = nullptr) const;
103
  FormatToken *parseBlock(bool MustBeDeclaration = false,
104
                          unsigned AddLevels = 1u, bool MunchSemi = true,
105
                          bool KeepBraces = true, IfStmtKind *IfKind = nullptr,
106
                          bool UnindentWhitesmithsBraces = false,
107
                          bool CanContainBracedList = true,
108
                          TokenType NextLBracesType = TT_Unknown);
109
  void parseChildBlock(bool CanContainBracedList = true,
110
                       TokenType NextLBracesType = TT_Unknown);
111
  void parsePPDirective();
112
  void parsePPDefine();
113
  void parsePPIf(bool IfDef);
114
  void parsePPElIf();
115
  void parsePPElse();
116
  void parsePPEndIf();
117
  void parsePPUnknown();
118
  void readTokenWithJavaScriptASI();
119
  void parseStructuralElement(bool IsTopLevel = false,
120
                              TokenType NextLBracesType = TT_Unknown,
121
                              IfStmtKind *IfKind = nullptr,
122
                              FormatToken **IfLeftBrace = nullptr,
123
                              bool *HasDoWhile = nullptr,
124
                              bool *HasLabel = nullptr);
125
  bool tryToParseBracedList();
126
  bool parseBracedList(bool ContinueOnSemicolons = false, bool IsEnum = false,
127
                       tok::TokenKind ClosingBraceKind = tok::r_brace);
128
  void parseParens(TokenType AmpAmpTokenType = TT_Unknown);
129
  void parseSquare(bool LambdaIntroducer = false);
130
  void keepAncestorBraces();
131
  void parseUnbracedBody(bool CheckEOF = false);
132
  void handleAttributes();
133
  bool handleCppAttributes();
134
  FormatToken *parseIfThenElse(IfStmtKind *IfKind, bool KeepBraces = false);
135
  void parseTryCatch();
136
  void parseLoopBody(bool KeepBraces, bool WrapRightBrace);
137
  void parseForOrWhileLoop();
138
  void parseDoWhile();
139
  void parseLabel(bool LeftAlignLabel = false);
140
  void parseCaseLabel();
141
  void parseSwitch();
142
  void parseNamespace();
143
  void parseModuleImport();
144
  void parseNew();
145
  void parseAccessSpecifier();
146
  bool parseEnum();
147
  bool parseStructLike();
148
  void parseConcept();
149
  bool parseRequires();
150
  void parseRequiresClause(FormatToken *RequiresToken);
151
  void parseRequiresExpression(FormatToken *RequiresToken);
152
  void parseConstraintExpression();
153
  void parseJavaEnumBody();
154
  // Parses a record (aka class) as a top level element. If ParseAsExpr is true,
155
  // parses the record as a child block, i.e. if the class declaration is an
156
  // expression.
157
  void parseRecord(bool ParseAsExpr = false);
158
  void parseObjCLightweightGenerics();
159
  void parseObjCMethod();
160
  void parseObjCProtocolList();
161
  void parseObjCUntilAtEnd();
162
  void parseObjCInterfaceOrImplementation();
163
  bool parseObjCProtocol();
164
  void parseJavaScriptEs6ImportExport();
165
  void parseStatementMacro();
166
  void parseCSharpAttribute();
167
  // Parse a C# generic type constraint: `where T : IComparable<T>`.
168
  // See:
169
  // https://docs.microsoft.com/en-us/dotnet/csharp/language-reference/keywords/where-generic-type-constraint
170
  void parseCSharpGenericTypeConstraint();
171
  bool tryToParseLambda();
172
  bool tryToParseChildBlock();
173
  bool tryToParseLambdaIntroducer();
174
  bool tryToParsePropertyAccessor();
175
  void tryToParseJSFunction();
176
  bool tryToParseSimpleAttribute();
177
178
  // Used by addUnwrappedLine to denote whether to keep or remove a level
179
  // when resetting the line state.
180
  enum class LineLevel { Remove, Keep };
181
182
  void addUnwrappedLine(LineLevel AdjustLevel = LineLevel::Remove);
183
  bool eof() const;
184
  // LevelDifference is the difference of levels after and before the current
185
  // token. For example:
186
  // - if the token is '{' and opens a block, LevelDifference is 1.
187
  // - if the token is '}' and closes a block, LevelDifference is -1.
188
  void nextToken(int LevelDifference = 0);
189
  void readToken(int LevelDifference = 0);
190
191
  // Decides which comment tokens should be added to the current line and which
192
  // should be added as comments before the next token.
193
  //
194
  // Comments specifies the sequence of comment tokens to analyze. They get
195
  // either pushed to the current line or added to the comments before the next
196
  // token.
197
  //
198
  // NextTok specifies the next token. A null pointer NextTok is supported, and
199
  // signifies either the absence of a next token, or that the next token
200
  // shouldn't be taken into accunt for the analysis.
201
  void distributeComments(const SmallVectorImpl<FormatToken *> &Comments,
202
                          const FormatToken *NextTok);
203
204
  // Adds the comment preceding the next token to unwrapped lines.
205
  void flushComments(bool NewlineBeforeNext);
206
  void pushToken(FormatToken *Tok);
207
  void calculateBraceTypes(bool ExpectClassBody = false);
208
209
  // Marks a conditional compilation edge (for example, an '#if', '#ifdef',
210
  // '#else' or merge conflict marker). If 'Unreachable' is true, assumes
211
  // this branch either cannot be taken (for example '#if false'), or should
212
  // not be taken in this round.
213
  void conditionalCompilationCondition(bool Unreachable);
214
  void conditionalCompilationStart(bool Unreachable);
215
  void conditionalCompilationAlternative();
216
  void conditionalCompilationEnd();
217
218
  bool isOnNewLine(const FormatToken &FormatTok);
219
220
  // Compute hash of the current preprocessor branch.
221
  // This is used to identify the different branches, and thus track if block
222
  // open and close in the same branch.
223
  size_t computePPHash() const;
224
225
  // FIXME: We are constantly running into bugs where Line.Level is incorrectly
226
  // subtracted from beyond 0. Introduce a method to subtract from Line.Level
227
  // and use that everywhere in the Parser.
228
  std::unique_ptr<UnwrappedLine> Line;
229
230
  // Comments are sorted into unwrapped lines by whether they are in the same
231
  // line as the previous token, or not. If not, they belong to the next token.
232
  // Since the next token might already be in a new unwrapped line, we need to
233
  // store the comments belonging to that token.
234
  SmallVector<FormatToken *, 1> CommentsBeforeNextToken;
235
  FormatToken *FormatTok;
236
  bool MustBreakBeforeNextToken;
237
238
  // The parsed lines. Only added to through \c CurrentLines.
239
  SmallVector<UnwrappedLine, 8> Lines;
240
241
  // Preprocessor directives are parsed out-of-order from other unwrapped lines.
242
  // Thus, we need to keep a list of preprocessor directives to be reported
243
  // after an unwrapped line that has been started was finished.
244
  SmallVector<UnwrappedLine, 4> PreprocessorDirectives;
245
246
  // New unwrapped lines are added via CurrentLines.
247
  // Usually points to \c &Lines. While parsing a preprocessor directive when
248
  // there is an unfinished previous unwrapped line, will point to
249
  // \c &PreprocessorDirectives.
250
  SmallVectorImpl<UnwrappedLine> *CurrentLines;
251
252
  // We store for each line whether it must be a declaration depending on
253
  // whether we are in a compound statement or not.
254
  llvm::BitVector DeclarationScopeStack;
255
256
  const FormatStyle &Style;
257
  const AdditionalKeywords &Keywords;
258
259
  llvm::Regex CommentPragmasRegex;
260
261
  FormatTokenSource *Tokens;
262
  UnwrappedLineConsumer &Callback;
263
264
  // FIXME: This is a temporary measure until we have reworked the ownership
265
  // of the format tokens. The goal is to have the actual tokens created and
266
  // owned outside of and handed into the UnwrappedLineParser.
267
  ArrayRef<FormatToken *> AllTokens;
268
269
  // Keeps a stack of the states of nested control statements (true if the
270
  // statement contains more than some predefined number of nested statements).
271
  SmallVector<bool, 8> NestedTooDeep;
272
273
  // Represents preprocessor branch type, so we can find matching
274
  // #if/#else/#endif directives.
275
  enum PPBranchKind {
276
    PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0
277
    PP_Unreachable  // #if 0 or a conditional preprocessor block inside #if 0
278
  };
279
280
  struct PPBranch {
281
3.98k
    PPBranch(PPBranchKind Kind, size_t Line) : Kind(Kind), Line(Line) {}
282
    PPBranchKind Kind;
283
    size_t Line;
284
  };
285
286
  // Keeps a stack of currently active preprocessor branching directives.
287
  SmallVector<PPBranch, 16> PPStack;
288
289
  // The \c UnwrappedLineParser re-parses the code for each combination
290
  // of preprocessor branches that can be taken.
291
  // To that end, we take the same branch (#if, #else, or one of the #elif
292
  // branches) for each nesting level of preprocessor branches.
293
  // \c PPBranchLevel stores the current nesting level of preprocessor
294
  // branches during one pass over the code.
295
  int PPBranchLevel;
296
297
  // Contains the current branch (#if, #else or one of the #elif branches)
298
  // for each nesting level.
299
  SmallVector<int, 8> PPLevelBranchIndex;
300
301
  // Contains the maximum number of branches at each nesting level.
302
  SmallVector<int, 8> PPLevelBranchCount;
303
304
  // Contains the number of branches per nesting level we are currently
305
  // in while parsing a preprocessor branch sequence.
306
  // This is used to update PPLevelBranchCount at the end of a branch
307
  // sequence.
308
  std::stack<int> PPChainBranchIndex;
309
310
  // Include guard search state. Used to fixup preprocessor indent levels
311
  // so that include guards do not participate in indentation.
312
  enum IncludeGuardState {
313
    IG_Inited,   // Search started, looking for #ifndef.
314
    IG_IfNdefed, // #ifndef found, IncludeGuardToken points to condition.
315
    IG_Defined,  // Matching #define found, checking other requirements.
316
    IG_Found,    // All requirements met, need to fix indents.
317
    IG_Rejected, // Search failed or never started.
318
  };
319
320
  // Current state of include guard search.
321
  IncludeGuardState IncludeGuard;
322
323
  // Points to the #ifndef condition for a potential include guard. Null unless
324
  // IncludeGuardState == IG_IfNdefed.
325
  FormatToken *IncludeGuardToken;
326
327
  // Contains the first start column where the source begins. This is zero for
328
  // normal source code and may be nonzero when formatting a code fragment that
329
  // does not start at the beginning of the file.
330
  unsigned FirstStartColumn;
331
332
  friend class ScopedLineState;
333
  friend class CompoundStatementIndenter;
334
};
335
336
struct UnwrappedLineNode {
337
762
  UnwrappedLineNode() : Tok(nullptr) {}
338
1.11M
  UnwrappedLineNode(FormatToken *Tok) : Tok(Tok) {}
339
340
  FormatToken *Tok;
341
  SmallVector<UnwrappedLine, 0> Children;
342
};
343
344
inline UnwrappedLine::UnwrappedLine()
345
    : Level(0), InPPDirective(false), MustBeDeclaration(false),
346
149k
      MatchingOpeningBlockLineIndex(kInvalidIndex) {}
347
348
} // end namespace format
349
} // end namespace clang
350
351
#endif