Alexander Kornienko | 3b71155 | 2013-06-03 16:45:03 +0000 | [diff] [blame] | 1 | //===--- FormatToken.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 FormatToken, a wrapper |
| 12 | /// around Token with additional information related to formatting. |
| 13 | /// |
| 14 | //===----------------------------------------------------------------------===// |
| 15 | |
| 16 | #ifndef LLVM_CLANG_FORMAT_FORMAT_TOKEN_H |
| 17 | #define LLVM_CLANG_FORMAT_FORMAT_TOKEN_H |
| 18 | |
| 19 | #include "clang/Basic/OperatorPrecedence.h" |
| 20 | #include "clang/Lex/Lexer.h" |
| 21 | |
| 22 | namespace clang { |
| 23 | namespace format { |
| 24 | |
| 25 | enum TokenType { |
| 26 | TT_BinaryOperator, |
| 27 | TT_BlockComment, |
| 28 | TT_CastRParen, |
| 29 | TT_ConditionalExpr, |
| 30 | TT_CtorInitializerColon, |
| 31 | TT_DesignatedInitializerPeriod, |
| 32 | TT_ImplicitStringLiteral, |
| 33 | TT_InlineASMColon, |
| 34 | TT_InheritanceColon, |
| 35 | TT_FunctionTypeLParen, |
| 36 | TT_LineComment, |
| 37 | TT_ObjCArrayLiteral, |
| 38 | TT_ObjCBlockLParen, |
| 39 | TT_ObjCDecl, |
| 40 | TT_ObjCDictLiteral, |
| 41 | TT_ObjCForIn, |
| 42 | TT_ObjCMethodExpr, |
| 43 | TT_ObjCMethodSpecifier, |
| 44 | TT_ObjCProperty, |
| 45 | TT_ObjCSelectorName, |
| 46 | TT_OverloadedOperator, |
| 47 | TT_OverloadedOperatorLParen, |
| 48 | TT_PointerOrReference, |
| 49 | TT_PureVirtualSpecifier, |
| 50 | TT_RangeBasedForLoopColon, |
| 51 | TT_StartOfName, |
| 52 | TT_TemplateCloser, |
| 53 | TT_TemplateOpener, |
| 54 | TT_TrailingUnaryOperator, |
| 55 | TT_UnaryOperator, |
| 56 | TT_Unknown |
| 57 | }; |
| 58 | |
| 59 | /// \brief A wrapper around a \c Token storing information about the |
| 60 | /// whitespace characters preceeding it. |
| 61 | struct FormatToken { |
| 62 | FormatToken() |
| 63 | : NewlinesBefore(0), HasUnescapedNewline(false), LastNewlineOffset(0), |
Alexander Kornienko | 54e6c9d | 2013-06-07 17:45:07 +0000 | [diff] [blame] | 64 | CodePointCount(0), IsFirst(false), MustBreakBefore(false), |
| 65 | Type(TT_Unknown), SpacesRequiredBefore(0), CanBreakBefore(false), |
| 66 | ClosesTemplateDeclaration(false), ParameterCount(0), TotalLength(0), |
| 67 | UnbreakableTailLength(0), BindingStrength(0), SplitPenalty(0), |
| 68 | LongestObjCSelectorName(0), FakeRParens(0), LastInChainOfCalls(false), |
Alexander Kornienko | 3b71155 | 2013-06-03 16:45:03 +0000 | [diff] [blame] | 69 | PartOfMultiVariableDeclStmt(false), MatchingParen(NULL), Previous(NULL), |
| 70 | Next(NULL) {} |
| 71 | |
| 72 | /// \brief The \c Token. |
| 73 | Token Tok; |
| 74 | |
| 75 | /// \brief The number of newlines immediately before the \c Token. |
| 76 | /// |
| 77 | /// This can be used to determine what the user wrote in the original code |
| 78 | /// and thereby e.g. leave an empty line between two function definitions. |
| 79 | unsigned NewlinesBefore; |
| 80 | |
| 81 | /// \brief Whether there is at least one unescaped newline before the \c |
| 82 | /// Token. |
| 83 | bool HasUnescapedNewline; |
| 84 | |
| 85 | /// \brief The range of the whitespace immediately preceeding the \c Token. |
| 86 | SourceRange WhitespaceRange; |
| 87 | |
| 88 | /// \brief The offset just past the last '\n' in this token's leading |
| 89 | /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'. |
| 90 | unsigned LastNewlineOffset; |
| 91 | |
Alexander Kornienko | 0089510 | 2013-06-05 14:09:10 +0000 | [diff] [blame] | 92 | /// \brief The length of the non-whitespace parts of the token in CodePoints. |
| 93 | /// We need this to correctly measure number of columns a token spans. |
| 94 | unsigned CodePointCount; |
Alexander Kornienko | 3b71155 | 2013-06-03 16:45:03 +0000 | [diff] [blame] | 95 | |
| 96 | /// \brief Indicates that this is the first token. |
| 97 | bool IsFirst; |
| 98 | |
| 99 | /// \brief Whether there must be a line break before this token. |
| 100 | /// |
| 101 | /// This happens for example when a preprocessor directive ended directly |
| 102 | /// before the token. |
| 103 | bool MustBreakBefore; |
| 104 | |
| 105 | /// \brief Returns actual token start location without leading escaped |
| 106 | /// newlines and whitespace. |
| 107 | /// |
| 108 | /// This can be different to Tok.getLocation(), which includes leading escaped |
| 109 | /// newlines. |
| 110 | SourceLocation getStartOfNonWhitespace() const { |
| 111 | return WhitespaceRange.getEnd(); |
| 112 | } |
| 113 | |
| 114 | /// \brief The raw text of the token. |
| 115 | /// |
| 116 | /// Contains the raw token text without leading whitespace and without leading |
| 117 | /// escaped newlines. |
| 118 | StringRef TokenText; |
| 119 | |
| 120 | TokenType Type; |
| 121 | |
| 122 | unsigned SpacesRequiredBefore; |
| 123 | bool CanBreakBefore; |
| 124 | |
| 125 | bool ClosesTemplateDeclaration; |
| 126 | |
| 127 | /// \brief Number of parameters, if this is "(", "[" or "<". |
| 128 | /// |
| 129 | /// This is initialized to 1 as we don't need to distinguish functions with |
| 130 | /// 0 parameters from functions with 1 parameter. Thus, we can simply count |
| 131 | /// the number of commas. |
| 132 | unsigned ParameterCount; |
| 133 | |
| 134 | /// \brief The total length of the line up to and including this token. |
| 135 | unsigned TotalLength; |
| 136 | |
| 137 | /// \brief The length of following tokens until the next natural split point, |
| 138 | /// or the next token that can be broken. |
| 139 | unsigned UnbreakableTailLength; |
| 140 | |
| 141 | // FIXME: Come up with a 'cleaner' concept. |
| 142 | /// \brief The binding strength of a token. This is a combined value of |
| 143 | /// operator precedence, parenthesis nesting, etc. |
| 144 | unsigned BindingStrength; |
| 145 | |
| 146 | /// \brief Penalty for inserting a line break before this token. |
| 147 | unsigned SplitPenalty; |
| 148 | |
| 149 | /// \brief If this is the first ObjC selector name in an ObjC method |
| 150 | /// definition or call, this contains the length of the longest name. |
| 151 | unsigned LongestObjCSelectorName; |
| 152 | |
| 153 | /// \brief Stores the number of required fake parentheses and the |
| 154 | /// corresponding operator precedence. |
| 155 | /// |
| 156 | /// If multiple fake parentheses start at a token, this vector stores them in |
| 157 | /// reverse order, i.e. inner fake parenthesis first. |
| 158 | SmallVector<prec::Level, 4> FakeLParens; |
| 159 | /// \brief Insert this many fake ) after this token for correct indentation. |
| 160 | unsigned FakeRParens; |
| 161 | |
| 162 | /// \brief Is this the last "." or "->" in a builder-type call? |
| 163 | bool LastInChainOfCalls; |
| 164 | |
| 165 | /// \brief Is this token part of a \c DeclStmt defining multiple variables? |
| 166 | /// |
| 167 | /// Only set if \c Type == \c TT_StartOfName. |
| 168 | bool PartOfMultiVariableDeclStmt; |
| 169 | |
| 170 | bool is(tok::TokenKind Kind) const { return Tok.is(Kind); } |
| 171 | |
| 172 | bool isOneOf(tok::TokenKind K1, tok::TokenKind K2) const { |
| 173 | return is(K1) || is(K2); |
| 174 | } |
| 175 | |
| 176 | bool isOneOf(tok::TokenKind K1, tok::TokenKind K2, tok::TokenKind K3) const { |
| 177 | return is(K1) || is(K2) || is(K3); |
| 178 | } |
| 179 | |
| 180 | bool isOneOf(tok::TokenKind K1, tok::TokenKind K2, tok::TokenKind K3, |
| 181 | tok::TokenKind K4, tok::TokenKind K5 = tok::NUM_TOKENS, |
| 182 | tok::TokenKind K6 = tok::NUM_TOKENS, |
| 183 | tok::TokenKind K7 = tok::NUM_TOKENS, |
| 184 | tok::TokenKind K8 = tok::NUM_TOKENS, |
| 185 | tok::TokenKind K9 = tok::NUM_TOKENS, |
| 186 | tok::TokenKind K10 = tok::NUM_TOKENS, |
| 187 | tok::TokenKind K11 = tok::NUM_TOKENS, |
| 188 | tok::TokenKind K12 = tok::NUM_TOKENS) const { |
| 189 | return is(K1) || is(K2) || is(K3) || is(K4) || is(K5) || is(K6) || is(K7) || |
| 190 | is(K8) || is(K9) || is(K10) || is(K11) || is(K12); |
| 191 | } |
| 192 | |
| 193 | bool isNot(tok::TokenKind Kind) const { return Tok.isNot(Kind); } |
| 194 | |
| 195 | bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const { |
| 196 | return Tok.isObjCAtKeyword(Kind); |
| 197 | } |
| 198 | |
| 199 | bool isAccessSpecifier(bool ColonRequired = true) const { |
| 200 | return isOneOf(tok::kw_public, tok::kw_protected, tok::kw_private) && |
| 201 | (!ColonRequired || (Next && Next->is(tok::colon))); |
| 202 | } |
| 203 | |
| 204 | bool isObjCAccessSpecifier() const { |
| 205 | return is(tok::at) && Next && (Next->isObjCAtKeyword(tok::objc_public) || |
| 206 | Next->isObjCAtKeyword(tok::objc_protected) || |
| 207 | Next->isObjCAtKeyword(tok::objc_package) || |
| 208 | Next->isObjCAtKeyword(tok::objc_private)); |
| 209 | } |
| 210 | |
| 211 | /// \brief Returns whether \p Tok is ([{ or a template opening <. |
| 212 | bool opensScope() const { |
| 213 | return isOneOf(tok::l_paren, tok::l_brace, tok::l_square) || |
| 214 | Type == TT_TemplateOpener; |
| 215 | |
| 216 | } |
Nico Weber | 0e5a888 | 2013-06-25 19:25:12 +0000 | [diff] [blame] | 217 | /// \brief Returns whether \p Tok is )]} or a template closing >. |
Alexander Kornienko | 3b71155 | 2013-06-03 16:45:03 +0000 | [diff] [blame] | 218 | bool closesScope() const { |
| 219 | return isOneOf(tok::r_paren, tok::r_brace, tok::r_square) || |
| 220 | Type == TT_TemplateCloser; |
| 221 | } |
| 222 | |
| 223 | bool isUnaryOperator() const { |
| 224 | switch (Tok.getKind()) { |
| 225 | case tok::plus: |
| 226 | case tok::plusplus: |
| 227 | case tok::minus: |
| 228 | case tok::minusminus: |
| 229 | case tok::exclaim: |
| 230 | case tok::tilde: |
| 231 | case tok::kw_sizeof: |
| 232 | case tok::kw_alignof: |
| 233 | return true; |
| 234 | default: |
| 235 | return false; |
| 236 | } |
| 237 | } |
| 238 | bool isBinaryOperator() const { |
| 239 | // Comma is a binary operator, but does not behave as such wrt. formatting. |
| 240 | return getPrecedence() > prec::Comma; |
| 241 | } |
| 242 | bool isTrailingComment() const { |
| 243 | return is(tok::comment) && (!Next || Next->NewlinesBefore > 0); |
| 244 | } |
| 245 | |
| 246 | prec::Level getPrecedence() const { |
| 247 | return getBinOpPrecedence(Tok.getKind(), true, true); |
| 248 | } |
| 249 | |
| 250 | /// \brief Returns the previous token ignoring comments. |
| 251 | FormatToken *getPreviousNoneComment() const { |
| 252 | FormatToken *Tok = Previous; |
| 253 | while (Tok != NULL && Tok->is(tok::comment)) |
| 254 | Tok = Tok->Previous; |
| 255 | return Tok; |
| 256 | } |
| 257 | |
| 258 | /// \brief Returns the next token ignoring comments. |
| 259 | const FormatToken *getNextNoneComment() const { |
| 260 | const FormatToken *Tok = Next; |
| 261 | while (Tok != NULL && Tok->is(tok::comment)) |
| 262 | Tok = Tok->Next; |
| 263 | return Tok; |
| 264 | } |
| 265 | |
| 266 | FormatToken *MatchingParen; |
| 267 | |
| 268 | FormatToken *Previous; |
| 269 | FormatToken *Next; |
| 270 | |
| 271 | private: |
| 272 | // Disallow copying. |
Craig Topper | 53d4f31 | 2013-07-01 04:07:34 +0000 | [diff] [blame^] | 273 | FormatToken(const FormatToken &) LLVM_DELETED_FUNCTION; |
| 274 | void operator=(const FormatToken &) LLVM_DELETED_FUNCTION; |
Alexander Kornienko | 3b71155 | 2013-06-03 16:45:03 +0000 | [diff] [blame] | 275 | }; |
| 276 | |
| 277 | } // namespace format |
| 278 | } // namespace clang |
| 279 | |
| 280 | #endif // LLVM_CLANG_FORMAT_FORMAT_TOKEN_H |