Home | History | Annotate | Download | only in Format 
      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 
     24 namespace clang {
     25 namespace 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.
     38 class WhitespaceManager {
     39 public:
     40   WhitespaceManager(SourceManager &SourceMgr, const FormatStyle &Style)
     41       : SourceMgr(SourceMgr), Style(Style) {}
     42 
     43   /// \brief Replaces the whitespace in front of \p Tok. Only call once for
     44   /// each \c AnnotatedToken.
     45   void replaceWhitespace(const FormatToken &Tok, unsigned Newlines,
     46                          unsigned Spaces, unsigned StartOfTokenColumn,
     47                          bool InPPDirective = false);
     48 
     49   /// \brief Adds information about an unchangable token's whitespace.
     50   ///
     51   /// Needs to be called for every token for which \c replaceWhitespace
     52   /// was not called.
     53   void addUntouchableToken(const FormatToken &Tok, bool InPPDirective);
     54 
     55   /// \brief Inserts or replaces whitespace in the middle of a token.
     56   ///
     57   /// Inserts \p PreviousPostfix, \p Newlines, \p Spaces and \p CurrentPrefix
     58   /// (in this order) at \p Offset inside \p Tok, replacing \p ReplaceChars
     59   /// characters.
     60   ///
     61   /// When \p InPPDirective is true, escaped newlines are inserted. \p Spaces is
     62   /// used to align backslashes correctly.
     63   void replaceWhitespaceInToken(const FormatToken &Tok, unsigned Offset,
     64                                 unsigned ReplaceChars,
     65                                 StringRef PreviousPostfix,
     66                                 StringRef CurrentPrefix, bool InPPDirective,
     67                                 unsigned Newlines, unsigned Spaces);
     68 
     69   /// \brief Returns all the \c Replacements created during formatting.
     70   const tooling::Replacements &generateReplacements();
     71 
     72 private:
     73   /// \brief Represents a change before a token, a break inside a token,
     74   /// or the layout of an unchanged token (or whitespace within).
     75   struct Change {
     76     /// \brief Functor to sort changes in original source order.
     77     class IsBeforeInFile {
     78     public:
     79       IsBeforeInFile(const SourceManager &SourceMgr) : SourceMgr(SourceMgr) {}
     80       bool operator()(const Change &C1, const Change &C2) const;
     81 
     82     private:
     83       const SourceManager &SourceMgr;
     84     };
     85 
     86     Change() {}
     87 
     88     /// \brief Creates a \c Change.
     89     ///
     90     /// The generated \c Change will replace the characters at
     91     /// \p OriginalWhitespaceRange with a concatenation of
     92     /// \p PreviousLinePostfix, \p NewlinesBefore line breaks, \p Spaces spaces
     93     /// and \p CurrentLinePrefix.
     94     ///
     95     /// \p StartOfTokenColumn and \p InPPDirective will be used to lay out
     96     /// trailing comments and escaped newlines.
     97     Change(bool CreateReplacement, const SourceRange &OriginalWhitespaceRange,
     98            unsigned Spaces, unsigned StartOfTokenColumn,
     99            unsigned NewlinesBefore, StringRef PreviousLinePostfix,
    100            StringRef CurrentLinePrefix, tok::TokenKind Kind,
    101            bool ContinuesPPDirective);
    102 
    103     bool CreateReplacement;
    104     // Changes might be in the middle of a token, so we cannot just keep the
    105     // FormatToken around to query its information.
    106     SourceRange OriginalWhitespaceRange;
    107     unsigned StartOfTokenColumn;
    108     unsigned NewlinesBefore;
    109     std::string PreviousLinePostfix;
    110     std::string CurrentLinePrefix;
    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     tok::TokenKind Kind;
    116     bool ContinuesPPDirective;
    117 
    118     // The number of spaces in front of the token or broken part of the token.
    119     // This will be adapted when aligning tokens.
    120     unsigned Spaces;
    121 
    122     // \c IsTrailingComment, \c TokenLength, \c PreviousEndOfTokenColumn and
    123     // \c EscapedNewlineColumn will be calculated in
    124     // \c calculateLineBreakInformation.
    125     bool IsTrailingComment;
    126     unsigned TokenLength;
    127     unsigned PreviousEndOfTokenColumn;
    128     unsigned EscapedNewlineColumn;
    129   };
    130 
    131   /// \brief Calculate \c IsTrailingComment, \c TokenLength for the last tokens
    132   /// or token parts in a line and \c PreviousEndOfTokenColumn and
    133   /// \c EscapedNewlineColumn for the first tokens or token parts in a line.
    134   void calculateLineBreakInformation();
    135 
    136   /// \brief Align trailing comments over all \c Changes.
    137   void alignTrailingComments();
    138 
    139   /// \brief Align trailing comments from change \p Start to change \p End at
    140   /// the specified \p Column.
    141   void alignTrailingComments(unsigned Start, unsigned End, unsigned Column);
    142 
    143   /// \brief Align escaped newlines over all \c Changes.
    144   void alignEscapedNewlines();
    145 
    146   /// \brief Align escaped newlines from change \p Start to change \p End at
    147   /// the specified \p Column.
    148   void alignEscapedNewlines(unsigned Start, unsigned End, unsigned Column);
    149 
    150   /// \brief Fill \c Replaces with the replacements for all effective changes.
    151   void generateChanges();
    152 
    153   /// \brief Stores \p Text as the replacement for the whitespace in \p Range.
    154   void storeReplacement(const SourceRange &Range, StringRef Text);
    155   std::string getNewlineText(unsigned Newlines, unsigned Spaces);
    156   std::string getNewlineText(unsigned Newlines, unsigned Spaces,
    157                              unsigned PreviousEndOfTokenColumn,
    158                              unsigned EscapedNewlineColumn);
    159   std::string getIndentText(unsigned Spaces);
    160 
    161   SmallVector<Change, 16> Changes;
    162   SourceManager &SourceMgr;
    163   tooling::Replacements Replaces;
    164   const FormatStyle &Style;
    165 };
    166 
    167 } // namespace format
    168 } // namespace clang
    169 
    170 #endif // LLVM_CLANG_FORMAT_WHITESPACEMANAGER_H
    171