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_LIB_FORMAT_WHITESPACEMANAGER_H
17#define LLVM_CLANG_LIB_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(const 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 unchangeable 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  /// Note: \p Spaces can be negative to retain information about initial
67  /// relative column offset between a line of a block comment and the start of
68  /// the comment. This negative offset may be compensated by trailing comment
69  /// alignment here. In all other cases negative \p Spaces will be truncated to
70  /// 0.
71  ///
72  /// When \p InPPDirective is true, escaped newlines are inserted. \p Spaces is
73  /// used to align backslashes correctly.
74  void replaceWhitespaceInToken(const FormatToken &Tok, unsigned Offset,
75                                unsigned ReplaceChars,
76                                StringRef PreviousPostfix,
77                                StringRef CurrentPrefix, bool InPPDirective,
78                                unsigned Newlines, unsigned IndentLevel,
79                                int Spaces);
80
81  /// \brief Returns all the \c Replacements created during formatting.
82  const tooling::Replacements &generateReplacements();
83
84  /// \brief Represents a change before a token, a break inside a token,
85  /// or the layout of an unchanged token (or whitespace within).
86  struct Change {
87    /// \brief Functor to sort changes in original source order.
88    class IsBeforeInFile {
89    public:
90      IsBeforeInFile(const SourceManager &SourceMgr) : SourceMgr(SourceMgr) {}
91      bool operator()(const Change &C1, const Change &C2) const;
92
93    private:
94      const SourceManager &SourceMgr;
95    };
96
97    Change() {}
98
99    /// \brief Creates a \c Change.
100    ///
101    /// The generated \c Change will replace the characters at
102    /// \p OriginalWhitespaceRange with a concatenation of
103    /// \p PreviousLinePostfix, \p NewlinesBefore line breaks, \p Spaces spaces
104    /// and \p CurrentLinePrefix.
105    ///
106    /// \p StartOfTokenColumn and \p InPPDirective will be used to lay out
107    /// trailing comments and escaped newlines.
108    Change(bool CreateReplacement, SourceRange OriginalWhitespaceRange,
109           unsigned IndentLevel, int Spaces, unsigned StartOfTokenColumn,
110           unsigned NewlinesBefore, StringRef PreviousLinePostfix,
111           StringRef CurrentLinePrefix, tok::TokenKind Kind,
112           bool ContinuesPPDirective, bool IsStartOfDeclName,
113           bool IsInsideToken);
114
115    bool CreateReplacement;
116    // Changes might be in the middle of a token, so we cannot just keep the
117    // FormatToken around to query its information.
118    SourceRange OriginalWhitespaceRange;
119    unsigned StartOfTokenColumn;
120    unsigned NewlinesBefore;
121    std::string PreviousLinePostfix;
122    std::string CurrentLinePrefix;
123    // The kind of the token whose whitespace this change replaces, or in which
124    // this change inserts whitespace.
125    // FIXME: Currently this is not set correctly for breaks inside comments, as
126    // the \c BreakableToken is still doing its own alignment.
127    tok::TokenKind Kind;
128    bool ContinuesPPDirective;
129    bool IsStartOfDeclName;
130
131    // The number of nested blocks the token is in. This is used to add tabs
132    // only for the indentation, and not for alignment, when
133    // UseTab = US_ForIndentation.
134    unsigned IndentLevel;
135
136    // The number of spaces in front of the token or broken part of the token.
137    // This will be adapted when aligning tokens.
138    // Can be negative to retain information about the initial relative offset
139    // of the lines in a block comment. This is used when aligning trailing
140    // comments. Uncompensated negative offset is truncated to 0.
141    int Spaces;
142
143    // If this change is inside of a token but not at the start of the token or
144    // directly after a newline.
145    bool IsInsideToken;
146
147    // \c IsTrailingComment, \c TokenLength, \c PreviousEndOfTokenColumn and
148    // \c EscapedNewlineColumn will be calculated in
149    // \c calculateLineBreakInformation.
150    bool IsTrailingComment;
151    unsigned TokenLength;
152    unsigned PreviousEndOfTokenColumn;
153    unsigned EscapedNewlineColumn;
154
155    // These fields are used to retain correct relative line indentation in a
156    // block comment when aligning trailing comments.
157    //
158    // If this Change represents a continuation of a block comment,
159    // \c StartOfBlockComment is pointer to the first Change in the block
160    // comment. \c IndentationOffset is a relative column offset to this
161    // change, so that the correct column can be reconstructed at the end of
162    // the alignment process.
163    const Change *StartOfBlockComment;
164    int IndentationOffset;
165  };
166
167private:
168  /// \brief Calculate \c IsTrailingComment, \c TokenLength for the last tokens
169  /// or token parts in a line and \c PreviousEndOfTokenColumn and
170  /// \c EscapedNewlineColumn for the first tokens or token parts in a line.
171  void calculateLineBreakInformation();
172
173  /// \brief Align consecutive assignments over all \c Changes.
174  void alignConsecutiveAssignments();
175
176  /// \brief Align consecutive declarations over all \c Changes.
177  void alignConsecutiveDeclarations();
178
179  /// \brief Align trailing comments over all \c Changes.
180  void alignTrailingComments();
181
182  /// \brief Align trailing comments from change \p Start to change \p End at
183  /// the specified \p Column.
184  void alignTrailingComments(unsigned Start, unsigned End, unsigned Column);
185
186  /// \brief Align escaped newlines over all \c Changes.
187  void alignEscapedNewlines();
188
189  /// \brief Align escaped newlines from change \p Start to change \p End at
190  /// the specified \p Column.
191  void alignEscapedNewlines(unsigned Start, unsigned End, unsigned Column);
192
193  /// \brief Fill \c Replaces with the replacements for all effective changes.
194  void generateChanges();
195
196  /// \brief Stores \p Text as the replacement for the whitespace in \p Range.
197  void storeReplacement(SourceRange Range, StringRef Text);
198  void appendNewlineText(std::string &Text, unsigned Newlines);
199  void appendNewlineText(std::string &Text, unsigned Newlines,
200                         unsigned PreviousEndOfTokenColumn,
201                         unsigned EscapedNewlineColumn);
202  void appendIndentText(std::string &Text, unsigned IndentLevel,
203                        unsigned Spaces, unsigned WhitespaceStartColumn);
204
205  SmallVector<Change, 16> Changes;
206  const SourceManager &SourceMgr;
207  tooling::Replacements Replaces;
208  const FormatStyle &Style;
209  bool UseCRLF;
210};
211
212} // namespace format
213} // namespace clang
214
215#endif
216