Coverage Report

Created: 2020-02-25 14:32

/Users/buildslave/jenkins/workspace/coverage/llvm-project/clang/lib/Format/UnwrappedLineParser.h
Line
Count
Source (jump to first uncovered line)
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
32.8k
  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(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
  bool tryToParseLambda();
130
  bool tryToParseLambdaIntroducer();
131
  void tryToParseJSFunction();
132
  void addUnwrappedLine();
133
  bool eof() const;
134
  // LevelDifference is the difference of levels after and before the current
135
  // token. For example:
136
  // - if the token is '{' and opens a block, LevelDifference is 1.
137
  // - if the token is '}' and closes a block, LevelDifference is -1.
138
  void nextToken(int LevelDifference = 0);
139
  void readToken(int LevelDifference = 0);
140
141
  // Decides which comment tokens should be added to the current line and which
142
  // should be added as comments before the next token.
143
  //
144
  // Comments specifies the sequence of comment tokens to analyze. They get
145
  // either pushed to the current line or added to the comments before the next
146
  // token.
147
  //
148
  // NextTok specifies the next token. A null pointer NextTok is supported, and
149
  // signifies either the absence of a next token, or that the next token
150
  // shouldn't be taken into accunt for the analysis.
151
  void distributeComments(const SmallVectorImpl<FormatToken *> &Comments,
152
                          const FormatToken *NextTok);
153
154
  // Adds the comment preceding the next token to unwrapped lines.
155
  void flushComments(bool NewlineBeforeNext);
156
  void pushToken(FormatToken *Tok);
157
  void calculateBraceTypes(bool ExpectClassBody = false);
158
159
  // Marks a conditional compilation edge (for example, an '#if', '#ifdef',
160
  // '#else' or merge conflict marker). If 'Unreachable' is true, assumes
161
  // this branch either cannot be taken (for example '#if false'), or should
162
  // not be taken in this round.
163
  void conditionalCompilationCondition(bool Unreachable);
164
  void conditionalCompilationStart(bool Unreachable);
165
  void conditionalCompilationAlternative();
166
  void conditionalCompilationEnd();
167
168
  bool isOnNewLine(const FormatToken &FormatTok);
169
170
  // Compute hash of the current preprocessor branch.
171
  // This is used to identify the different branches, and thus track if block
172
  // open and close in the same branch.
173
  size_t computePPHash() const;
174
175
  // FIXME: We are constantly running into bugs where Line.Level is incorrectly
176
  // subtracted from beyond 0. Introduce a method to subtract from Line.Level
177
  // and use that everywhere in the Parser.
178
  std::unique_ptr<UnwrappedLine> Line;
179
180
  // Comments are sorted into unwrapped lines by whether they are in the same
181
  // line as the previous token, or not. If not, they belong to the next token.
182
  // Since the next token might already be in a new unwrapped line, we need to
183
  // store the comments belonging to that token.
184
  SmallVector<FormatToken *, 1> CommentsBeforeNextToken;
185
  FormatToken *FormatTok;
186
  bool MustBreakBeforeNextToken;
187
188
  // The parsed lines. Only added to through \c CurrentLines.
189
  SmallVector<UnwrappedLine, 8> Lines;
190
191
  // Preprocessor directives are parsed out-of-order from other unwrapped lines.
192
  // Thus, we need to keep a list of preprocessor directives to be reported
193
  // after an unwrapped line that has been started was finished.
194
  SmallVector<UnwrappedLine, 4> PreprocessorDirectives;
195
196
  // New unwrapped lines are added via CurrentLines.
197
  // Usually points to \c &Lines. While parsing a preprocessor directive when
198
  // there is an unfinished previous unwrapped line, will point to
199
  // \c &PreprocessorDirectives.
200
  SmallVectorImpl<UnwrappedLine> *CurrentLines;
201
202
  // We store for each line whether it must be a declaration depending on
203
  // whether we are in a compound statement or not.
204
  std::vector<bool> DeclarationScopeStack;
205
206
  const FormatStyle &Style;
207
  const AdditionalKeywords &Keywords;
208
209
  llvm::Regex CommentPragmasRegex;
210
211
  FormatTokenSource *Tokens;
212
  UnwrappedLineConsumer &Callback;
213
214
  // FIXME: This is a temporary measure until we have reworked the ownership
215
  // of the format tokens. The goal is to have the actual tokens created and
216
  // owned outside of and handed into the UnwrappedLineParser.
217
  ArrayRef<FormatToken *> AllTokens;
218
219
  // Represents preprocessor branch type, so we can find matching
220
  // #if/#else/#endif directives.
221
  enum PPBranchKind {
222
    PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0
223
    PP_Unreachable  // #if 0 or a conditional preprocessor block inside #if 0
224
  };
225
226
  struct PPBranch {
227
1.63k
    PPBranch(PPBranchKind Kind, size_t Line) : Kind(Kind), Line(Line) {}
228
    PPBranchKind Kind;
229
    size_t Line;
230
  };
231
232
  // Keeps a stack of currently active preprocessor branching directives.
233
  SmallVector<PPBranch, 16> PPStack;
234
235
  // The \c UnwrappedLineParser re-parses the code for each combination
236
  // of preprocessor branches that can be taken.
237
  // To that end, we take the same branch (#if, #else, or one of the #elif
238
  // branches) for each nesting level of preprocessor branches.
239
  // \c PPBranchLevel stores the current nesting level of preprocessor
240
  // branches during one pass over the code.
241
  int PPBranchLevel;
242
243
  // Contains the current branch (#if, #else or one of the #elif branches)
244
  // for each nesting level.
245
  SmallVector<int, 8> PPLevelBranchIndex;
246
247
  // Contains the maximum number of branches at each nesting level.
248
  SmallVector<int, 8> PPLevelBranchCount;
249
250
  // Contains the number of branches per nesting level we are currently
251
  // in while parsing a preprocessor branch sequence.
252
  // This is used to update PPLevelBranchCount at the end of a branch
253
  // sequence.
254
  std::stack<int> PPChainBranchIndex;
255
256
  // Include guard search state. Used to fixup preprocessor indent levels
257
  // so that include guards do not participate in indentation.
258
  enum IncludeGuardState {
259
    IG_Inited,   // Search started, looking for #ifndef.
260
    IG_IfNdefed, // #ifndef found, IncludeGuardToken points to condition.
261
    IG_Defined,  // Matching #define found, checking other requirements.
262
    IG_Found,    // All requirements met, need to fix indents.
263
    IG_Rejected, // Search failed or never started.
264
  };
265
266
  // Current state of include guard search.
267
  IncludeGuardState IncludeGuard;
268
269
  // Points to the #ifndef condition for a potential include guard. Null unless
270
  // IncludeGuardState == IG_IfNdefed.
271
  FormatToken *IncludeGuardToken;
272
273
  // Contains the first start column where the source begins. This is zero for
274
  // normal source code and may be nonzero when formatting a code fragment that
275
  // does not start at the beginning of the file.
276
  unsigned FirstStartColumn;
277
278
  friend class ScopedLineState;
279
  friend class CompoundStatementIndenter;
280
};
281
282
struct UnwrappedLineNode {
283
0
  UnwrappedLineNode() : Tok(nullptr) {}
284
567k
  UnwrappedLineNode(FormatToken *Tok) : Tok(Tok) {}
285
286
  FormatToken *Tok;
287
  SmallVector<UnwrappedLine, 0> Children;
288
};
289
290
inline UnwrappedLine::UnwrappedLine()
291
    : Level(0), InPPDirective(false), MustBeDeclaration(false),
292
76.5k
      MatchingOpeningBlockLineIndex(kInvalidIndex) {}
293
294
} // end namespace format
295
} // end namespace clang
296
297
#endif