1 //===--- UnwrappedLineParser.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 This file contains the declaration of the UnwrappedLineParser,
12 /// which turns a stream of tokens into UnwrappedLines.
13 ///
14 //===----------------------------------------------------------------------===//
15 
16 #ifndef LLVM_CLANG_LIB_FORMAT_UNWRAPPEDLINEPARSER_H
17 #define LLVM_CLANG_LIB_FORMAT_UNWRAPPEDLINEPARSER_H
18 
19 #include "FormatToken.h"
20 #include "clang/Basic/IdentifierTable.h"
21 #include "clang/Format/Format.h"
22 #include "llvm/Support/Regex.h"
23 #include <list>
24 #include <stack>
25 
26 namespace clang {
27 namespace format {
28 
29 struct UnwrappedLineNode;
30 
31 /// \brief An unwrapped line is a sequence of \c Token, that we would like to
32 /// put on a single line if there was no column limit.
33 ///
34 /// This is used as a main interface between the \c UnwrappedLineParser and the
35 /// \c UnwrappedLineFormatter. The key property is that changing the formatting
36 /// within an unwrapped line does not affect any other unwrapped lines.
37 struct UnwrappedLine {
38   UnwrappedLine();
39 
40   // FIXME: Don't use std::list here.
41   /// \brief The \c Tokens comprising this \c UnwrappedLine.
42   std::list<UnwrappedLineNode> Tokens;
43 
44   /// \brief The indent level of the \c UnwrappedLine.
45   unsigned Level;
46 
47   /// \brief Whether this \c UnwrappedLine is part of a preprocessor directive.
48   bool InPPDirective;
49 
50   bool MustBeDeclaration;
51 
52   /// \brief If this \c UnwrappedLine closes a block in a sequence of lines,
53   /// \c MatchingOpeningBlockLineIndex stores the index of the corresponding
54   /// opening line. Otherwise, \c MatchingOpeningBlockLineIndex must be
55   /// \c kInvalidIndex.
56   size_t MatchingOpeningBlockLineIndex;
57 
58   static const size_t kInvalidIndex = -1;
59 };
60 
61 class UnwrappedLineConsumer {
62 public:
63   virtual ~UnwrappedLineConsumer() {}
64   virtual void consumeUnwrappedLine(const UnwrappedLine &Line) = 0;
65   virtual void finishRun() = 0;
66 };
67 
68 class FormatTokenSource;
69 
70 class UnwrappedLineParser {
71 public:
72   UnwrappedLineParser(const FormatStyle &Style,
73                       const AdditionalKeywords &Keywords,
74                       ArrayRef<FormatToken *> Tokens,
75                       UnwrappedLineConsumer &Callback);
76 
77   void parse();
78 
79 private:
80   void reset();
81   void parseFile();
82   void parseLevel(bool HasOpeningBrace);
83   void parseBlock(bool MustBeDeclaration, bool AddLevel = true,
84                   bool MunchSemi = true);
85   void parseChildBlock();
86   void parsePPDirective();
87   void parsePPDefine();
88   void parsePPIf(bool IfDef);
89   void parsePPElIf();
90   void parsePPElse();
91   void parsePPEndIf();
92   void parsePPUnknown();
93   void readTokenWithJavaScriptASI();
94   void parseStructuralElement();
95   bool tryToParseBracedList();
96   bool parseBracedList(bool ContinueOnSemicolons = false);
97   void parseParens();
98   void parseSquare();
99   void parseIfThenElse();
100   void parseTryCatch();
101   void parseForOrWhileLoop();
102   void parseDoWhile();
103   void parseLabel();
104   void parseCaseLabel();
105   void parseSwitch();
106   void parseNamespace();
107   void parseNew();
108   void parseAccessSpecifier();
109   bool parseEnum();
110   void parseJavaEnumBody();
111   // Parses a record (aka class) as a top level element. If ParseAsExpr is true,
112   // parses the record as a child block, i.e. if the class declaration is an
113   // expression.
114   void parseRecord(bool ParseAsExpr = false);
115   void parseObjCProtocolList();
116   void parseObjCUntilAtEnd();
117   void parseObjCInterfaceOrImplementation();
118   void parseObjCProtocol();
119   void parseJavaScriptEs6ImportExport();
120   bool tryToParseLambda();
121   bool tryToParseLambdaIntroducer();
122   void tryToParseJSFunction();
123   void addUnwrappedLine();
124   bool eof() const;
125   void nextToken();
126   const FormatToken *getPreviousToken();
127   void readToken();
128 
129   // Decides which comment tokens should be added to the current line and which
130   // should be added as comments before the next token.
131   //
132   // Comments specifies the sequence of comment tokens to analyze. They get
133   // either pushed to the current line or added to the comments before the next
134   // token.
135   //
136   // NextTok specifies the next token. A null pointer NextTok is supported, and
137   // signifies either the absense of a next token, or that the next token
138   // shouldn't be taken into accunt for the analysis.
139   void distributeComments(const SmallVectorImpl<FormatToken *> &Comments,
140                           const FormatToken *NextTok);
141 
142   // Adds the comment preceding the next token to unwrapped lines.
143   void flushComments(bool NewlineBeforeNext);
144   void pushToken(FormatToken *Tok);
145   void calculateBraceTypes(bool ExpectClassBody = false);
146 
147   // Marks a conditional compilation edge (for example, an '#if', '#ifdef',
148   // '#else' or merge conflict marker). If 'Unreachable' is true, assumes
149   // this branch either cannot be taken (for example '#if false'), or should
150   // not be taken in this round.
151   void conditionalCompilationCondition(bool Unreachable);
152   void conditionalCompilationStart(bool Unreachable);
153   void conditionalCompilationAlternative();
154   void conditionalCompilationEnd();
155 
156   bool isOnNewLine(const FormatToken &FormatTok);
157 
158   // FIXME: We are constantly running into bugs where Line.Level is incorrectly
159   // subtracted from beyond 0. Introduce a method to subtract from Line.Level
160   // and use that everywhere in the Parser.
161   std::unique_ptr<UnwrappedLine> Line;
162 
163   // Comments are sorted into unwrapped lines by whether they are in the same
164   // line as the previous token, or not. If not, they belong to the next token.
165   // Since the next token might already be in a new unwrapped line, we need to
166   // store the comments belonging to that token.
167   SmallVector<FormatToken *, 1> CommentsBeforeNextToken;
168   FormatToken *FormatTok;
169   bool MustBreakBeforeNextToken;
170 
171   // The parsed lines. Only added to through \c CurrentLines.
172   SmallVector<UnwrappedLine, 8> Lines;
173 
174   // Preprocessor directives are parsed out-of-order from other unwrapped lines.
175   // Thus, we need to keep a list of preprocessor directives to be reported
176   // after an unwarpped line that has been started was finished.
177   SmallVector<UnwrappedLine, 4> PreprocessorDirectives;
178 
179   // New unwrapped lines are added via CurrentLines.
180   // Usually points to \c &Lines. While parsing a preprocessor directive when
181   // there is an unfinished previous unwrapped line, will point to
182   // \c &PreprocessorDirectives.
183   SmallVectorImpl<UnwrappedLine> *CurrentLines;
184 
185   // We store for each line whether it must be a declaration depending on
186   // whether we are in a compound statement or not.
187   std::vector<bool> DeclarationScopeStack;
188 
189   const FormatStyle &Style;
190   const AdditionalKeywords &Keywords;
191 
192   llvm::Regex CommentPragmasRegex;
193 
194   FormatTokenSource *Tokens;
195   UnwrappedLineConsumer &Callback;
196 
197   // FIXME: This is a temporary measure until we have reworked the ownership
198   // of the format tokens. The goal is to have the actual tokens created and
199   // owned outside of and handed into the UnwrappedLineParser.
200   ArrayRef<FormatToken *> AllTokens;
201 
202   // Represents preprocessor branch type, so we can find matching
203   // #if/#else/#endif directives.
204   enum PPBranchKind {
205     PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0
206     PP_Unreachable  // #if 0 or a conditional preprocessor block inside #if 0
207   };
208 
209   // Keeps a stack of currently active preprocessor branching directives.
210   SmallVector<PPBranchKind, 16> PPStack;
211 
212   // The \c UnwrappedLineParser re-parses the code for each combination
213   // of preprocessor branches that can be taken.
214   // To that end, we take the same branch (#if, #else, or one of the #elif
215   // branches) for each nesting level of preprocessor branches.
216   // \c PPBranchLevel stores the current nesting level of preprocessor
217   // branches during one pass over the code.
218   int PPBranchLevel;
219 
220   // Contains the current branch (#if, #else or one of the #elif branches)
221   // for each nesting level.
222   SmallVector<int, 8> PPLevelBranchIndex;
223 
224   // Contains the maximum number of branches at each nesting level.
225   SmallVector<int, 8> PPLevelBranchCount;
226 
227   // Contains the number of branches per nesting level we are currently
228   // in while parsing a preprocessor branch sequence.
229   // This is used to update PPLevelBranchCount at the end of a branch
230   // sequence.
231   std::stack<int> PPChainBranchIndex;
232 
233   friend class ScopedLineState;
234   friend class CompoundStatementIndenter;
235 };
236 
237 struct UnwrappedLineNode {
238   UnwrappedLineNode() : Tok(nullptr) {}
239   UnwrappedLineNode(FormatToken *Tok) : Tok(Tok) {}
240 
241   FormatToken *Tok;
242   SmallVector<UnwrappedLine, 0> Children;
243 };
244 
245 inline UnwrappedLine::UnwrappedLine() : Level(0), InPPDirective(false),
246   MustBeDeclaration(false), MatchingOpeningBlockLineIndex(kInvalidIndex) {}
247 
248 } // end namespace format
249 } // end namespace clang
250 
251 #endif
252