WhitespaceManager.h revision 263508 
1//===--- WhitespaceManager.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 WhitespaceManager class manages whitespace around tokens and their
12/// replacements.
13///
14//===----------------------------------------------------------------------===//
15
16#ifndef LLVM_CLANG_FORMAT_WHITESPACEMANAGER_H
17#define LLVM_CLANG_FORMAT_WHITESPACEMANAGER_H
18
19#include "TokenAnnotator.h"
20#include "clang/Basic/SourceManager.h"
21#include "clang/Format/Format.h"
22#include <string>
23
24namespace clang {
25namespace format {
26
27/// \brief Manages the whitespaces around tokens and their replacements.
28///
29/// This includes special handling for certain constructs, e.g. the alignment of
30/// trailing line comments.
31///
32/// To guarantee correctness of alignment operations, the \c WhitespaceManager
33/// must be informed about every token in the source file; for each token, there
34/// must be exactly one call to either \c replaceWhitespace or
35/// \c addUntouchableToken.
36///
37/// There may be multiple calls to \c breakToken for a given token.
38class WhitespaceManager {
39public:
40  WhitespaceManager(SourceManager &SourceMgr, const FormatStyle &Style,
41                    bool UseCRLF)
42      : SourceMgr(SourceMgr), Style(Style), UseCRLF(UseCRLF) {}
43
44  /// \brief Prepares the \c WhitespaceManager for another run.
45  void reset();
46
47  /// \brief Replaces the whitespace in front of \p Tok. Only call once for
48  /// each \c AnnotatedToken.
49  void replaceWhitespace(FormatToken &Tok, unsigned Newlines,
50                         unsigned IndentLevel, unsigned Spaces,
51                         unsigned StartOfTokenColumn,
52                         bool InPPDirective = false);
53
54  /// \brief Adds information about an unchangable token's whitespace.
55  ///
56  /// Needs to be called for every token for which \c replaceWhitespace
57  /// was not called.
58  void addUntouchableToken(const FormatToken &Tok, bool InPPDirective);
59
60  /// \brief Inserts or replaces whitespace in the middle of a token.
61  ///
62  /// Inserts \p PreviousPostfix, \p Newlines, \p Spaces and \p CurrentPrefix
63  /// (in this order) at \p Offset inside \p Tok, replacing \p ReplaceChars
64  /// characters.
65  ///
66  /// When \p InPPDirective is true, escaped newlines are inserted. \p Spaces is
67  /// used to align backslashes correctly.
68  void replaceWhitespaceInToken(const FormatToken &Tok, unsigned Offset,
69                                unsigned ReplaceChars,
70                                StringRef PreviousPostfix,
71                                StringRef CurrentPrefix, bool InPPDirective,
72                                unsigned Newlines, unsigned IndentLevel,
73                                unsigned Spaces);
74
75  /// \brief Returns all the \c Replacements created during formatting.
76  const tooling::Replacements &generateReplacements();
77
78private:
79  /// \brief Represents a change before a token, a break inside a token,
80  /// or the layout of an unchanged token (or whitespace within).
81  struct Change {
82    /// \brief Functor to sort changes in original source order.
83    class IsBeforeInFile {
84    public:
85      IsBeforeInFile(const SourceManager &SourceMgr) : SourceMgr(SourceMgr) {}
86      bool operator()(const Change &C1, const Change &C2) const;
87
88    private:
89      const SourceManager &SourceMgr;
90    };
91
92    Change() {}
93
94    /// \brief Creates a \c Change.
95    ///
96    /// The generated \c Change will replace the characters at
97    /// \p OriginalWhitespaceRange with a concatenation of
98    /// \p PreviousLinePostfix, \p NewlinesBefore line breaks, \p Spaces spaces
99    /// and \p CurrentLinePrefix.
100    ///
101    /// \p StartOfTokenColumn and \p InPPDirective will be used to lay out
102    /// trailing comments and escaped newlines.
103    Change(bool CreateReplacement, const SourceRange &OriginalWhitespaceRange,
104           unsigned IndentLevel, unsigned Spaces, unsigned StartOfTokenColumn,
105           unsigned NewlinesBefore, StringRef PreviousLinePostfix,
106           StringRef CurrentLinePrefix, tok::TokenKind Kind,
107           bool ContinuesPPDirective);
108
109    bool CreateReplacement;
110    // Changes might be in the middle of a token, so we cannot just keep the
111    // FormatToken around to query its information.
112    SourceRange OriginalWhitespaceRange;
113    unsigned StartOfTokenColumn;
114    unsigned NewlinesBefore;
115    std::string PreviousLinePostfix;
116    std::string CurrentLinePrefix;
117    // The kind of the token whose whitespace this change replaces, or in which
118    // this change inserts whitespace.
119    // FIXME: Currently this is not set correctly for breaks inside comments, as
120    // the \c BreakableToken is still doing its own alignment.
121    tok::TokenKind Kind;
122    bool ContinuesPPDirective;
123
124    // The number of nested blocks the token is in. This is used to add tabs
125    // only for the indentation, and not for alignment, when
126    // UseTab = US_ForIndentation.
127    unsigned IndentLevel;
128
129    // The number of spaces in front of the token or broken part of the token.
130    // This will be adapted when aligning tokens.
131    unsigned Spaces;
132
133    // \c IsTrailingComment, \c TokenLength, \c PreviousEndOfTokenColumn and
134    // \c EscapedNewlineColumn will be calculated in
135    // \c calculateLineBreakInformation.
136    bool IsTrailingComment;
137    unsigned TokenLength;
138    unsigned PreviousEndOfTokenColumn;
139    unsigned EscapedNewlineColumn;
140  };
141
142  /// \brief Calculate \c IsTrailingComment, \c TokenLength for the last tokens
143  /// or token parts in a line and \c PreviousEndOfTokenColumn and
144  /// \c EscapedNewlineColumn for the first tokens or token parts in a line.
145  void calculateLineBreakInformation();
146
147  /// \brief Align trailing comments over all \c Changes.
148  void alignTrailingComments();
149
150  /// \brief Align trailing comments from change \p Start to change \p End at
151  /// the specified \p Column.
152  void alignTrailingComments(unsigned Start, unsigned End, unsigned Column);
153
154  /// \brief Align escaped newlines over all \c Changes.
155  void alignEscapedNewlines();
156
157  /// \brief Align escaped newlines from change \p Start to change \p End at
158  /// the specified \p Column.
159  void alignEscapedNewlines(unsigned Start, unsigned End, unsigned Column);
160
161  /// \brief Fill \c Replaces with the replacements for all effective changes.
162  void generateChanges();
163
164  /// \brief Stores \p Text as the replacement for the whitespace in \p Range.
165  void storeReplacement(const SourceRange &Range, StringRef Text);
166  void appendNewlineText(std::string &Text, unsigned Newlines);
167  void appendNewlineText(std::string &Text, unsigned Newlines,
168                         unsigned PreviousEndOfTokenColumn,
169                         unsigned EscapedNewlineColumn);
170  void appendIndentText(std::string &Text, unsigned IndentLevel,
171                        unsigned Spaces, unsigned WhitespaceStartColumn);
172
173  SmallVector<Change, 16> Changes;
174  SourceManager &SourceMgr;
175  tooling::Replacements Replaces;
176  const FormatStyle &Style;
177  bool UseCRLF;
178};
179
180} // namespace format
181} // namespace clang
182
183#endif // LLVM_CLANG_FORMAT_WHITESPACEMANAGER_H
184