WhitespaceManager.h revision 323112
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 Replaces the whitespace in front of \p Tok. Only call once for 45 /// each \c AnnotatedToken. 46 /// 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); 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, int Spaces); 79 80 /// \brief Returns all the \c Replacements created during formatting. 81 const tooling::Replacements &generateReplacements(); 82 83 /// \brief Represents a change before a token, a break inside a token, 84 /// or the layout of an unchanged token (or whitespace within). 85 struct Change { 86 /// \brief Functor to sort changes in original source order. 87 class IsBeforeInFile { 88 public: 89 IsBeforeInFile(const SourceManager &SourceMgr) : SourceMgr(SourceMgr) {} 90 bool operator()(const Change &C1, const Change &C2) const; 91 92 private: 93 const SourceManager &SourceMgr; 94 }; 95 96 /// \brief Creates a \c Change. 97 /// 98 /// The generated \c Change will replace the characters at 99 /// \p OriginalWhitespaceRange with a concatenation of 100 /// \p PreviousLinePostfix, \p NewlinesBefore line breaks, \p Spaces spaces 101 /// and \p CurrentLinePrefix. 102 /// 103 /// \p StartOfTokenColumn and \p InPPDirective will be used to lay out 104 /// trailing comments and escaped newlines. 105 Change(const FormatToken &Tok, bool CreateReplacement, 106 SourceRange OriginalWhitespaceRange, int Spaces, 107 unsigned StartOfTokenColumn, unsigned NewlinesBefore, 108 StringRef PreviousLinePostfix, StringRef CurrentLinePrefix, 109 bool ContinuesPPDirective, bool IsInsideToken); 110 111 // The kind of the token whose whitespace this change replaces, or in which 112 // this change inserts whitespace. 113 // FIXME: Currently this is not set correctly for breaks inside comments, as 114 // the \c BreakableToken is still doing its own alignment. 115 const FormatToken *Tok; 116 117 bool CreateReplacement; 118 // Changes might be in the middle of a token, so we cannot just keep the 119 // FormatToken around to query its information. 120 SourceRange OriginalWhitespaceRange; 121 unsigned StartOfTokenColumn; 122 unsigned NewlinesBefore; 123 std::string PreviousLinePostfix; 124 std::string CurrentLinePrefix; 125 bool ContinuesPPDirective; 126 127 // The number of spaces in front of the token or broken part of the token. 128 // This will be adapted when aligning tokens. 129 // Can be negative to retain information about the initial relative offset 130 // of the lines in a block comment. This is used when aligning trailing 131 // comments. Uncompensated negative offset is truncated to 0. 132 int Spaces; 133 134 // If this change is inside of a token but not at the start of the token or 135 // directly after a newline. 136 bool IsInsideToken; 137 138 // \c IsTrailingComment, \c TokenLength, \c PreviousEndOfTokenColumn and 139 // \c EscapedNewlineColumn will be calculated in 140 // \c calculateLineBreakInformation. 141 bool IsTrailingComment; 142 unsigned TokenLength; 143 unsigned PreviousEndOfTokenColumn; 144 unsigned EscapedNewlineColumn; 145 146 // These fields are used to retain correct relative line indentation in a 147 // block comment when aligning trailing comments. 148 // 149 // If this Change represents a continuation of a block comment, 150 // \c StartOfBlockComment is pointer to the first Change in the block 151 // comment. \c IndentationOffset is a relative column offset to this 152 // change, so that the correct column can be reconstructed at the end of 153 // the alignment process. 154 const Change *StartOfBlockComment; 155 int IndentationOffset; 156 157 // A combination of indent level and nesting level, which are used in 158 // tandem to compute lexical scope, for the purposes of deciding 159 // when to stop consecutive alignment runs. 160 std::pair<unsigned, unsigned> indentAndNestingLevel() const { 161 return std::make_pair(Tok->IndentLevel, Tok->NestingLevel); 162 } 163 }; 164 165private: 166 /// \brief Calculate \c IsTrailingComment, \c TokenLength for the last tokens 167 /// or token parts in a line and \c PreviousEndOfTokenColumn and 168 /// \c EscapedNewlineColumn for the first tokens or token parts in a line. 169 void calculateLineBreakInformation(); 170 171 /// \brief Align consecutive assignments over all \c Changes. 172 void alignConsecutiveAssignments(); 173 174 /// \brief Align consecutive declarations over all \c Changes. 175 void alignConsecutiveDeclarations(); 176 177 /// \brief Align trailing comments over all \c Changes. 178 void alignTrailingComments(); 179 180 /// \brief Align trailing comments from change \p Start to change \p End at 181 /// the specified \p Column. 182 void alignTrailingComments(unsigned Start, unsigned End, unsigned Column); 183 184 /// \brief Align escaped newlines over all \c Changes. 185 void alignEscapedNewlines(); 186 187 /// \brief Align escaped newlines from change \p Start to change \p End at 188 /// the specified \p Column. 189 void alignEscapedNewlines(unsigned Start, unsigned End, unsigned Column); 190 191 /// \brief Fill \c Replaces with the replacements for all effective changes. 192 void generateChanges(); 193 194 /// \brief Stores \p Text as the replacement for the whitespace in \p Range. 195 void storeReplacement(SourceRange Range, StringRef Text); 196 void appendNewlineText(std::string &Text, unsigned Newlines); 197 void appendNewlineText(std::string &Text, unsigned Newlines, 198 unsigned PreviousEndOfTokenColumn, 199 unsigned EscapedNewlineColumn); 200 void appendIndentText(std::string &Text, unsigned IndentLevel, 201 unsigned Spaces, unsigned WhitespaceStartColumn); 202 203 SmallVector<Change, 16> Changes; 204 const SourceManager &SourceMgr; 205 tooling::Replacements Replaces; 206 const FormatStyle &Style; 207 bool UseCRLF; 208}; 209 210} // namespace format 211} // namespace clang 212 213#endif 214