UnwrappedLineParser.h revision 526ed11ad9743c773df76bd1649d33fb92c2b8cb
1//===--- UnwrappedLineParser.h - Format C++ code ----------------*- C++ -*-===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9///
10/// \file
11/// \brief This file contains the declaration of the UnwrappedLineParser,
12/// which turns a stream of tokens into UnwrappedLines.
13///
14/// This is EXPERIMENTAL code under heavy development. It is not in a state yet,
15/// where it can be used to format real code.
16///
17//===----------------------------------------------------------------------===//
18
19#ifndef LLVM_CLANG_FORMAT_UNWRAPPED_LINE_PARSER_H
20#define LLVM_CLANG_FORMAT_UNWRAPPED_LINE_PARSER_H
21
22#include "clang/Basic/IdentifierTable.h"
23#include "clang/Basic/SourceManager.h"
24#include "clang/Format/Format.h"
25#include "clang/Lex/Lexer.h"
26
27#include <vector>
28
29namespace clang {
30namespace format {
31
32/// \brief A wrapper around a \c Token storing information about the
33/// whitespace characters preceeding it.
34struct FormatToken {
35  FormatToken()
36      : NewlinesBefore(0), HasUnescapedNewline(false), WhiteSpaceLength(0),
37        TokenLength(0), IsFirst(false), MustBreakBefore(false) {
38  }
39
40  /// \brief The \c Token.
41  Token Tok;
42
43  /// \brief The number of newlines immediately before the \c Token.
44  ///
45  /// This can be used to determine what the user wrote in the original code
46  /// and thereby e.g. leave an empty line between two function definitions.
47  unsigned NewlinesBefore;
48
49  /// \brief Whether there is at least one unescaped newline before the \c
50  /// Token.
51  bool HasUnescapedNewline;
52
53  /// \brief The location of the start of the whitespace immediately preceeding
54  /// the \c Token.
55  ///
56  /// Used together with \c WhiteSpaceLength to create a \c Replacement.
57  SourceLocation WhiteSpaceStart;
58
59  /// \brief The length in characters of the whitespace immediately preceeding
60  /// the \c Token.
61  unsigned WhiteSpaceLength;
62
63  /// \brief The length of the non-whitespace parts of the token. This is
64  /// necessary because we need to handle escaped newlines that are stored
65  /// with the token.
66  unsigned TokenLength;
67
68  /// \brief Indicates that this is the first token.
69  bool IsFirst;
70
71  /// \brief Whether there must be a line break before this token.
72  ///
73  /// This happens for example when a preprocessor directive ended directly
74  /// before the token.
75  bool MustBreakBefore;
76
77  // FIXME: We currently assume that there is exactly one token in this vector
78  // except for the very last token that does not have any children.
79  /// \brief All tokens that logically follow this token.
80  std::vector<FormatToken> Children;
81};
82
83/// \brief An unwrapped line is a sequence of \c Token, that we would like to
84/// put on a single line if there was no column limit.
85///
86/// This is used as a main interface between the \c UnwrappedLineParser and the
87/// \c UnwrappedLineFormatter. The key property is that changing the formatting
88/// within an unwrapped line does not affect any other unwrapped lines.
89struct UnwrappedLine {
90  UnwrappedLine() : Level(0), InPPDirective(false) {
91  }
92
93  /// \brief The \c Token comprising this \c UnwrappedLine.
94  FormatToken RootToken;
95
96  /// \brief The indent level of the \c UnwrappedLine.
97  unsigned Level;
98
99  /// \brief Whether this \c UnwrappedLine is part of a preprocessor directive.
100  bool InPPDirective;
101};
102
103class UnwrappedLineConsumer {
104public:
105  virtual ~UnwrappedLineConsumer() {
106  }
107  virtual void consumeUnwrappedLine(const UnwrappedLine &Line) = 0;
108};
109
110class FormatTokenSource {
111public:
112  virtual ~FormatTokenSource() {
113  }
114  virtual FormatToken getNextToken() = 0;
115};
116
117class UnwrappedLineParser {
118public:
119  UnwrappedLineParser(const FormatStyle &Style, FormatTokenSource &Tokens,
120                      UnwrappedLineConsumer &Callback);
121
122  /// Returns true in case of a structural error.
123  bool parse();
124
125private:
126  bool parseFile();
127  bool parseLevel(bool HasOpeningBrace);
128  bool parseBlock(unsigned AddLevels = 1);
129  void parsePPDirective();
130  void parsePPDefine();
131  void parsePPUnknown();
132  void parseComments();
133  void parseStructuralElement();
134  void parseParens();
135  void parseIfThenElse();
136  void parseForOrWhileLoop();
137  void parseDoWhile();
138  void parseLabel();
139  void parseCaseLabel();
140  void parseSwitch();
141  void parseNamespace();
142  void parseAccessSpecifier();
143  void parseEnum();
144  void parseStructOrClass();
145  void addUnwrappedLine();
146  bool eof() const;
147  void nextToken();
148  void readToken();
149
150  // FIXME: We are constantly running into bugs where Line.Level is incorrectly
151  // subtracted from beyond 0. Introduce a method to subtract from Line.Level
152  // and use that everywhere in the Parser.
153  llvm::OwningPtr<UnwrappedLine> Line;
154  bool RootTokenInitialized;
155  FormatToken *LastInCurrentLine;
156  FormatToken FormatTok;
157  bool MustBreakBeforeNextToken;
158
159  const FormatStyle &Style;
160  FormatTokenSource *Tokens;
161  UnwrappedLineConsumer &Callback;
162};
163
164} // end namespace format
165} // end namespace clang
166
167#endif // LLVM_CLANG_FORMAT_UNWRAPPED_LINE_PARSER_H
168