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 <list>
23 #include <stack>
24
25 namespace clang {
26 namespace format {
27
28 struct UnwrappedLineNode;
29
30 /// \brief An unwrapped line is a sequence of \c Token, that we would like to
31 /// put on a single line if there was no column limit.
32 ///
33 /// This is used as a main interface between the \c UnwrappedLineParser and the
34 /// \c UnwrappedLineFormatter. The key property is that changing the formatting
35 /// within an unwrapped line does not affect any other unwrapped lines.
36 struct UnwrappedLine {
37 UnwrappedLine();
38
39 // FIXME: Don't use std::list here.
40 /// \brief The \c Tokens comprising this \c UnwrappedLine.
41 std::list<UnwrappedLineNode> Tokens;
42
43 /// \brief The indent level of the \c UnwrappedLine.
44 unsigned Level;
45
46 /// \brief Whether this \c UnwrappedLine is part of a preprocessor directive.
47 bool InPPDirective;
48
49 bool MustBeDeclaration;
50 };
51
52 class UnwrappedLineConsumer {
53 public:
~UnwrappedLineConsumer()54 virtual ~UnwrappedLineConsumer() {}
55 virtual void consumeUnwrappedLine(const UnwrappedLine &Line) = 0;
56 virtual void finishRun() = 0;
57 };
58
59 class FormatTokenSource;
60
61 class UnwrappedLineParser {
62 public:
63 UnwrappedLineParser(const FormatStyle &Style,
64 const AdditionalKeywords &Keywords,
65 ArrayRef<FormatToken *> Tokens,
66 UnwrappedLineConsumer &Callback);
67
68 /// Returns true in case of a structural error.
69 bool parse();
70
71 private:
72 void reset();
73 void parseFile();
74 void parseLevel(bool HasOpeningBrace);
75 void parseBlock(bool MustBeDeclaration, bool AddLevel = true,
76 bool MunchSemi = true);
77 void parseChildBlock();
78 void parsePPDirective();
79 void parsePPDefine();
80 void parsePPIf(bool IfDef);
81 void parsePPElIf();
82 void parsePPElse();
83 void parsePPEndIf();
84 void parsePPUnknown();
85 void parseStructuralElement();
86 bool tryToParseBracedList();
87 bool parseBracedList(bool ContinueOnSemicolons = false);
88 void parseParens();
89 void parseSquare();
90 void parseIfThenElse();
91 void parseTryCatch();
92 void parseForOrWhileLoop();
93 void parseDoWhile();
94 void parseLabel();
95 void parseCaseLabel();
96 void parseSwitch();
97 void parseNamespace();
98 void parseAccessSpecifier();
99 void parseEnum();
100 void parseJavaEnumBody();
101 void parseRecord();
102 void parseObjCProtocolList();
103 void parseObjCUntilAtEnd();
104 void parseObjCInterfaceOrImplementation();
105 void parseObjCProtocol();
106 bool tryToParseLambda();
107 bool tryToParseLambdaIntroducer();
108 void tryToParseJSFunction();
109 void addUnwrappedLine();
110 bool eof() const;
111 void nextToken();
112 void readToken();
113 void flushComments(bool NewlineBeforeNext);
114 void pushToken(FormatToken *Tok);
115 void calculateBraceTypes();
116
117 // Marks a conditional compilation edge (for example, an '#if', '#ifdef',
118 // '#else' or merge conflict marker). If 'Unreachable' is true, assumes
119 // this branch either cannot be taken (for example '#if false'), or should
120 // not be taken in this round.
121 void conditionalCompilationCondition(bool Unreachable);
122 void conditionalCompilationStart(bool Unreachable);
123 void conditionalCompilationAlternative();
124 void conditionalCompilationEnd();
125
126 bool isOnNewLine(const FormatToken &FormatTok);
127
128 // FIXME: We are constantly running into bugs where Line.Level is incorrectly
129 // subtracted from beyond 0. Introduce a method to subtract from Line.Level
130 // and use that everywhere in the Parser.
131 std::unique_ptr<UnwrappedLine> Line;
132
133 // Comments are sorted into unwrapped lines by whether they are in the same
134 // line as the previous token, or not. If not, they belong to the next token.
135 // Since the next token might already be in a new unwrapped line, we need to
136 // store the comments belonging to that token.
137 SmallVector<FormatToken *, 1> CommentsBeforeNextToken;
138 FormatToken *FormatTok;
139 bool MustBreakBeforeNextToken;
140
141 // The parsed lines. Only added to through \c CurrentLines.
142 SmallVector<UnwrappedLine, 8> Lines;
143
144 // Preprocessor directives are parsed out-of-order from other unwrapped lines.
145 // Thus, we need to keep a list of preprocessor directives to be reported
146 // after an unwarpped line that has been started was finished.
147 SmallVector<UnwrappedLine, 4> PreprocessorDirectives;
148
149 // New unwrapped lines are added via CurrentLines.
150 // Usually points to \c &Lines. While parsing a preprocessor directive when
151 // there is an unfinished previous unwrapped line, will point to
152 // \c &PreprocessorDirectives.
153 SmallVectorImpl<UnwrappedLine> *CurrentLines;
154
155 // We store for each line whether it must be a declaration depending on
156 // whether we are in a compound statement or not.
157 std::vector<bool> DeclarationScopeStack;
158
159 // Will be true if we encounter an error that leads to possibily incorrect
160 // indentation levels.
161 bool StructuralError;
162
163 const FormatStyle &Style;
164 const AdditionalKeywords &Keywords;
165
166 FormatTokenSource *Tokens;
167 UnwrappedLineConsumer &Callback;
168
169 // FIXME: This is a temporary measure until we have reworked the ownership
170 // of the format tokens. The goal is to have the actual tokens created and
171 // owned outside of and handed into the UnwrappedLineParser.
172 ArrayRef<FormatToken *> AllTokens;
173
174 // Represents preprocessor branch type, so we can find matching
175 // #if/#else/#endif directives.
176 enum PPBranchKind {
177 PP_Conditional, // Any #if, #ifdef, #ifndef, #elif, block outside #if 0
178 PP_Unreachable // #if 0 or a conditional preprocessor block inside #if 0
179 };
180
181 // Keeps a stack of currently active preprocessor branching directives.
182 SmallVector<PPBranchKind, 16> PPStack;
183
184 // The \c UnwrappedLineParser re-parses the code for each combination
185 // of preprocessor branches that can be taken.
186 // To that end, we take the same branch (#if, #else, or one of the #elif
187 // branches) for each nesting level of preprocessor branches.
188 // \c PPBranchLevel stores the current nesting level of preprocessor
189 // branches during one pass over the code.
190 int PPBranchLevel;
191
192 // Contains the current branch (#if, #else or one of the #elif branches)
193 // for each nesting level.
194 SmallVector<int, 8> PPLevelBranchIndex;
195
196 // Contains the maximum number of branches at each nesting level.
197 SmallVector<int, 8> PPLevelBranchCount;
198
199 // Contains the number of branches per nesting level we are currently
200 // in while parsing a preprocessor branch sequence.
201 // This is used to update PPLevelBranchCount at the end of a branch
202 // sequence.
203 std::stack<int> PPChainBranchIndex;
204
205 friend class ScopedLineState;
206 friend class CompoundStatementIndenter;
207 };
208
209 struct UnwrappedLineNode {
UnwrappedLineNodeUnwrappedLineNode210 UnwrappedLineNode() : Tok(nullptr) {}
UnwrappedLineNodeUnwrappedLineNode211 UnwrappedLineNode(FormatToken *Tok) : Tok(Tok) {}
212
213 FormatToken *Tok;
214 SmallVector<UnwrappedLine, 0> Children;
215 };
216
UnwrappedLine()217 inline UnwrappedLine::UnwrappedLine()
218 : Level(0), InPPDirective(false), MustBeDeclaration(false) {}
219
220 } // end namespace format
221 } // end namespace clang
222
223 #endif
224