1 //===--- WhitespaceManager.h - Format C++ code ------------------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
11 /// WhitespaceManager class manages whitespace around tokens and their
14 //===----------------------------------------------------------------------===//
16 #ifndef LLVM_CLANG_LIB_FORMAT_WHITESPACEMANAGER_H
17 #define LLVM_CLANG_LIB_FORMAT_WHITESPACEMANAGER_H
19 #include "TokenAnnotator.h"
20 #include "clang/Basic/SourceManager.h"
21 #include "clang/Format/Format.h"
27 /// Manages the whitespaces around tokens and their replacements.
29 /// This includes special handling for certain constructs, e.g. the alignment of
30 /// trailing line comments.
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.
37 /// There may be multiple calls to \c breakToken for a given token.
38 class WhitespaceManager {
40 WhitespaceManager(const SourceManager &SourceMgr, const FormatStyle &Style,
42 : SourceMgr(SourceMgr), Style(Style), UseCRLF(UseCRLF) {}
44 /// Replaces the whitespace in front of \p Tok. Only call once for
45 /// each \c AnnotatedToken.
47 /// \p StartOfTokenColumn is the column at which the token will start after
48 /// this replacement. It is needed for determining how \p Spaces is turned
49 /// into tabs and spaces for some format styles.
50 void replaceWhitespace(FormatToken &Tok, unsigned Newlines, unsigned Spaces,
51 unsigned StartOfTokenColumn,
52 bool InPPDirective = false);
54 /// Adds information about an unchangeable token's whitespace.
56 /// Needs to be called for every token for which \c replaceWhitespace
58 void addUntouchableToken(const FormatToken &Tok, bool InPPDirective);
60 llvm::Error addReplacement(const tooling::Replacement &Replacement);
62 /// Inserts or replaces whitespace in the middle of a token.
64 /// Inserts \p PreviousPostfix, \p Newlines, \p Spaces and \p CurrentPrefix
65 /// (in this order) at \p Offset inside \p Tok, replacing \p ReplaceChars
68 /// Note: \p Spaces can be negative to retain information about initial
69 /// relative column offset between a line of a block comment and the start of
70 /// the comment. This negative offset may be compensated by trailing comment
71 /// alignment here. In all other cases negative \p Spaces will be truncated to
74 /// When \p InPPDirective is true, escaped newlines are inserted. \p Spaces is
75 /// used to align backslashes correctly.
76 void replaceWhitespaceInToken(const FormatToken &Tok, unsigned Offset,
77 unsigned ReplaceChars,
78 StringRef PreviousPostfix,
79 StringRef CurrentPrefix, bool InPPDirective,
80 unsigned Newlines, int Spaces);
82 /// Returns all the \c Replacements created during formatting.
83 const tooling::Replacements &generateReplacements();
85 /// Represents a change before a token, a break inside a token,
86 /// or the layout of an unchanged token (or whitespace within).
88 /// Functor to sort changes in original source order.
89 class IsBeforeInFile {
91 IsBeforeInFile(const SourceManager &SourceMgr) : SourceMgr(SourceMgr) {}
92 bool operator()(const Change &C1, const Change &C2) const;
95 const SourceManager &SourceMgr;
98 /// Creates a \c Change.
100 /// The generated \c Change will replace the characters at
101 /// \p OriginalWhitespaceRange with a concatenation of
102 /// \p PreviousLinePostfix, \p NewlinesBefore line breaks, \p Spaces spaces
103 /// and \p CurrentLinePrefix.
105 /// \p StartOfTokenColumn and \p InPPDirective will be used to lay out
106 /// trailing comments and escaped newlines.
107 Change(const FormatToken &Tok, bool CreateReplacement,
108 SourceRange OriginalWhitespaceRange, int Spaces,
109 unsigned StartOfTokenColumn, unsigned NewlinesBefore,
110 StringRef PreviousLinePostfix, StringRef CurrentLinePrefix,
111 bool ContinuesPPDirective, bool IsInsideToken);
113 // The kind of the token whose whitespace this change replaces, or in which
114 // this change inserts whitespace.
115 // FIXME: Currently this is not set correctly for breaks inside comments, as
116 // the \c BreakableToken is still doing its own alignment.
117 const FormatToken *Tok;
119 bool CreateReplacement;
120 // Changes might be in the middle of a token, so we cannot just keep the
121 // FormatToken around to query its information.
122 SourceRange OriginalWhitespaceRange;
123 unsigned StartOfTokenColumn;
124 unsigned NewlinesBefore;
125 std::string PreviousLinePostfix;
126 std::string CurrentLinePrefix;
127 bool ContinuesPPDirective;
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 // Can be negative to retain information about the initial relative offset
132 // of the lines in a block comment. This is used when aligning trailing
133 // comments. Uncompensated negative offset is truncated to 0.
136 // If this change is inside of a token but not at the start of the token or
137 // directly after a newline.
140 // \c IsTrailingComment, \c TokenLength, \c PreviousEndOfTokenColumn and
141 // \c EscapedNewlineColumn will be calculated in
142 // \c calculateLineBreakInformation.
143 bool IsTrailingComment;
144 unsigned TokenLength;
145 unsigned PreviousEndOfTokenColumn;
146 unsigned EscapedNewlineColumn;
148 // These fields are used to retain correct relative line indentation in a
149 // block comment when aligning trailing comments.
151 // If this Change represents a continuation of a block comment,
152 // \c StartOfBlockComment is pointer to the first Change in the block
153 // comment. \c IndentationOffset is a relative column offset to this
154 // change, so that the correct column can be reconstructed at the end of
155 // the alignment process.
156 const Change *StartOfBlockComment;
157 int IndentationOffset;
159 // A combination of indent level and nesting level, which are used in
160 // tandem to compute lexical scope, for the purposes of deciding
161 // when to stop consecutive alignment runs.
162 std::pair<unsigned, unsigned> indentAndNestingLevel() const {
163 return std::make_pair(Tok->IndentLevel, Tok->NestingLevel);
168 /// 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();
173 /// Align consecutive assignments over all \c Changes.
174 void alignConsecutiveAssignments();
176 /// Align consecutive declarations over all \c Changes.
177 void alignConsecutiveDeclarations();
179 /// Align trailing comments over all \c Changes.
180 void alignTrailingComments();
182 /// 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);
186 /// Align escaped newlines over all \c Changes.
187 void alignEscapedNewlines();
189 /// 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);
193 /// Fill \c Replaces with the replacements for all effective changes.
194 void generateChanges();
196 /// 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 appendEscapedNewlineText(std::string &Text, unsigned Newlines,
200 unsigned PreviousEndOfTokenColumn,
201 unsigned EscapedNewlineColumn);
202 void appendIndentText(std::string &Text, unsigned IndentLevel,
203 unsigned Spaces, unsigned WhitespaceStartColumn);
205 SmallVector<Change, 16> Changes;
206 const SourceManager &SourceMgr;
207 tooling::Replacements Replaces;
208 const FormatStyle &Style;
212 } // namespace format