blob: 0c5a5284627c7d6d1db69420605f3b0b543de08f [file] [log] [blame]
Alexander Kornienko4b672072013-06-03 16:45:03 +00001//===--- 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
Benjamin Kramer2f5db8b2014-08-13 16:25:19 +000016#ifndef LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H
17#define LLVM_CLANG_LIB_FORMAT_FORMATTOKEN_H
Alexander Kornienko4b672072013-06-03 16:45:03 +000018
Daniel Jasperd0ec0d62014-11-04 12:41:02 +000019#include "clang/Basic/IdentifierTable.h"
Alexander Kornienko4b672072013-06-03 16:45:03 +000020#include "clang/Basic/OperatorPrecedence.h"
Daniel Jasper8de9ed02013-08-22 15:00:41 +000021#include "clang/Format/Format.h"
Alexander Kornienko4b672072013-06-03 16:45:03 +000022#include "clang/Lex/Lexer.h"
Ahmed Charlesdfca6f92014-03-09 11:36:40 +000023#include <memory>
Alexander Kornienko4b672072013-06-03 16:45:03 +000024
25namespace clang {
26namespace format {
27
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +000028#define LIST_TOKEN_TYPES \
29 TYPE(ArrayInitializerLSquare) \
30 TYPE(ArraySubscriptLSquare) \
31 TYPE(AttributeParen) \
32 TYPE(BinaryOperator) \
33 TYPE(BitFieldColon) \
34 TYPE(BlockComment) \
35 TYPE(CastRParen) \
36 TYPE(ConditionalExpr) \
37 TYPE(ConflictAlternative) \
38 TYPE(ConflictEnd) \
39 TYPE(ConflictStart) \
40 TYPE(CtorInitializerColon) \
41 TYPE(CtorInitializerComma) \
42 TYPE(DesignatedInitializerPeriod) \
43 TYPE(DictLiteral) \
44 TYPE(ForEachMacro) \
45 TYPE(FunctionAnnotationRParen) \
46 TYPE(FunctionDeclarationName) \
47 TYPE(FunctionLBrace) \
48 TYPE(FunctionTypeLParen) \
49 TYPE(ImplicitStringLiteral) \
50 TYPE(InheritanceColon) \
Andi-Bogdan Postelnicu0ef8ee12017-03-10 15:10:37 +000051 TYPE(InheritanceComma) \
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +000052 TYPE(InlineASMBrace) \
53 TYPE(InlineASMColon) \
54 TYPE(JavaAnnotation) \
55 TYPE(JsComputedPropertyName) \
Martin Probst4ef03702017-05-04 15:04:04 +000056 TYPE(JsExponentiation) \
57 TYPE(JsExponentiationEqual) \
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +000058 TYPE(JsFatArrow) \
Martin Probst22b8d2692017-03-13 09:14:23 +000059 TYPE(JsNonNullAssertion) \
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +000060 TYPE(JsTypeColon) \
Daniel Jasper91b1d1a2016-03-21 17:57:31 +000061 TYPE(JsTypeOperator) \
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +000062 TYPE(JsTypeOptionalQuestion) \
63 TYPE(LambdaArrow) \
64 TYPE(LambdaLSquare) \
65 TYPE(LeadingJavaAnnotation) \
66 TYPE(LineComment) \
67 TYPE(MacroBlockBegin) \
68 TYPE(MacroBlockEnd) \
69 TYPE(ObjCBlockLBrace) \
70 TYPE(ObjCBlockLParen) \
71 TYPE(ObjCDecl) \
72 TYPE(ObjCForIn) \
73 TYPE(ObjCMethodExpr) \
74 TYPE(ObjCMethodSpecifier) \
75 TYPE(ObjCProperty) \
76 TYPE(ObjCStringLiteral) \
77 TYPE(OverloadedOperator) \
78 TYPE(OverloadedOperatorLParen) \
79 TYPE(PointerOrReference) \
80 TYPE(PureVirtualSpecifier) \
81 TYPE(RangeBasedForLoopColon) \
82 TYPE(RegexLiteral) \
83 TYPE(SelectorName) \
84 TYPE(StartOfName) \
85 TYPE(TemplateCloser) \
86 TYPE(TemplateOpener) \
87 TYPE(TemplateString) \
88 TYPE(TrailingAnnotation) \
89 TYPE(TrailingReturnArrow) \
90 TYPE(TrailingUnaryOperator) \
91 TYPE(UnaryOperator) \
92 TYPE(Unknown)
93
Alexander Kornienko4b672072013-06-03 16:45:03 +000094enum TokenType {
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +000095#define TYPE(X) TT_##X,
96LIST_TOKEN_TYPES
97#undef TYPE
98 NUM_TOKEN_TYPES
Alexander Kornienko4b672072013-06-03 16:45:03 +000099};
100
Birunthan Mohanathas67d81c82015-07-13 16:19:34 +0000101/// \brief Determines the name of a token type.
102const char *getTokenTypeName(TokenType Type);
103
Daniel Jasperb1f74a82013-07-09 09:06:29 +0000104// Represents what type of block a set of braces open.
Daniel Jaspere6fcf7d2015-06-17 13:08:06 +0000105enum BraceBlockKind { BK_Unknown, BK_Block, BK_BracedInit };
Daniel Jasperb1f74a82013-07-09 09:06:29 +0000106
Daniel Jasperb10cbc42013-07-10 14:02:49 +0000107// The packing kind of a function's parameters.
Daniel Jaspere6fcf7d2015-06-17 13:08:06 +0000108enum ParameterPackingKind { PPK_BinPacked, PPK_OnePerLine, PPK_Inconclusive };
Daniel Jasperb10cbc42013-07-10 14:02:49 +0000109
Daniel Jaspere6fcf7d2015-06-17 13:08:06 +0000110enum FormatDecision { FD_Unformatted, FD_Continue, FD_Break };
Manuel Klimek71814b42013-10-11 21:25:45 +0000111
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000112class TokenRole;
Daniel Jasper9fe0e8d2013-09-05 09:29:45 +0000113class AnnotatedLine;
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000114
Alexander Kornienko4b672072013-06-03 16:45:03 +0000115/// \brief A wrapper around a \c Token storing information about the
Alp Tokerf6a24ce2013-12-05 16:25:25 +0000116/// whitespace characters preceding it.
Alexander Kornienko4b672072013-06-03 16:45:03 +0000117struct FormatToken {
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000118 FormatToken() {}
Alexander Kornienko4b672072013-06-03 16:45:03 +0000119
120 /// \brief The \c Token.
121 Token Tok;
122
123 /// \brief The number of newlines immediately before the \c Token.
124 ///
125 /// This can be used to determine what the user wrote in the original code
126 /// and thereby e.g. leave an empty line between two function definitions.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000127 unsigned NewlinesBefore = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000128
129 /// \brief Whether there is at least one unescaped newline before the \c
130 /// Token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000131 bool HasUnescapedNewline = false;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000132
Alp Tokerf6a24ce2013-12-05 16:25:25 +0000133 /// \brief The range of the whitespace immediately preceding the \c Token.
Alexander Kornienko4b672072013-06-03 16:45:03 +0000134 SourceRange WhitespaceRange;
135
136 /// \brief The offset just past the last '\n' in this token's leading
137 /// whitespace (relative to \c WhiteSpaceStart). 0 if there is no '\n'.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000138 unsigned LastNewlineOffset = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000139
Alexander Kornienko39856b72013-09-10 09:38:25 +0000140 /// \brief The width of the non-whitespace parts of the token (or its first
141 /// line for multi-line tokens) in columns.
Alexander Kornienkoffcc0102013-06-05 14:09:10 +0000142 /// We need this to correctly measure number of columns a token spans.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000143 unsigned ColumnWidth = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000144
Alexander Kornienko39856b72013-09-10 09:38:25 +0000145 /// \brief Contains the width in columns of the last line of a multi-line
Alexander Kornienko632abb92013-09-02 13:58:14 +0000146 /// token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000147 unsigned LastLineColumnWidth = 0;
Alexander Kornienko632abb92013-09-02 13:58:14 +0000148
Alexander Kornienko39856b72013-09-10 09:38:25 +0000149 /// \brief Whether the token text contains newlines (escaped or not).
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000150 bool IsMultiline = false;
Alexander Kornienko632abb92013-09-02 13:58:14 +0000151
Martin Probst7ea9b6d2016-05-29 14:41:36 +0000152 /// \brief Indicates that this is the first token of the file.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000153 bool IsFirst = false;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000154
155 /// \brief Whether there must be a line break before this token.
156 ///
157 /// This happens for example when a preprocessor directive ended directly
158 /// before the token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000159 bool MustBreakBefore = false;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000160
161 /// \brief The raw text of the token.
162 ///
163 /// Contains the raw token text without leading whitespace and without leading
164 /// escaped newlines.
165 StringRef TokenText;
166
Daniel Jasper8369aa52013-07-16 20:28:33 +0000167 /// \brief Set to \c true if this token is an unterminated literal.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000168 bool IsUnterminatedLiteral = 0;
Daniel Jasper8369aa52013-07-16 20:28:33 +0000169
Daniel Jasperb1f74a82013-07-09 09:06:29 +0000170 /// \brief Contains the kind of block if this token is a brace.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000171 BraceBlockKind BlockKind = BK_Unknown;
Daniel Jasperb1f74a82013-07-09 09:06:29 +0000172
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000173 TokenType Type = TT_Unknown;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000174
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000175 /// \brief The number of spaces that should be inserted before this token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000176 unsigned SpacesRequiredBefore = 0;
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000177
178 /// \brief \c true if it is allowed to break before this token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000179 bool CanBreakBefore = false;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000180
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000181 /// \brief \c true if this is the ">" of "template<..>".
182 bool ClosesTemplateDeclaration = false;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000183
184 /// \brief Number of parameters, if this is "(", "[" or "<".
185 ///
186 /// This is initialized to 1 as we don't need to distinguish functions with
187 /// 0 parameters from functions with 1 parameter. Thus, we can simply count
188 /// the number of commas.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000189 unsigned ParameterCount = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000190
Daniel Jasper114a2bc2014-06-03 12:02:45 +0000191 /// \brief Number of parameters that are nested blocks,
192 /// if this is "(", "[" or "<".
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000193 unsigned BlockParameterCount = 0;
Daniel Jasper114a2bc2014-06-03 12:02:45 +0000194
Daniel Jasperde7ca752015-05-04 07:39:00 +0000195 /// \brief If this is a bracket ("<", "(", "[" or "{"), contains the kind of
196 /// the surrounding bracket.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000197 tok::TokenKind ParentBracket = tok::unknown;
Daniel Jasperde7ca752015-05-04 07:39:00 +0000198
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000199 /// \brief A token can have a special role that can carry extra information
200 /// about the token's formatting.
Ahmed Charlesb8984322014-03-07 20:03:18 +0000201 std::unique_ptr<TokenRole> Role;
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000202
Daniel Jasperb10cbc42013-07-10 14:02:49 +0000203 /// \brief If this is an opening parenthesis, how are the parameters packed?
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000204 ParameterPackingKind PackingKind = PPK_Inconclusive;
Daniel Jasperb10cbc42013-07-10 14:02:49 +0000205
Manuel Klimek31c85922013-08-29 15:21:40 +0000206 /// \brief The total length of the unwrapped line up to and including this
207 /// token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000208 unsigned TotalLength = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000209
Alexander Kornienko39856b72013-09-10 09:38:25 +0000210 /// \brief The original 0-based column of this token, including expanded tabs.
211 /// The configured TabWidth is used as tab width.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000212 unsigned OriginalColumn = 0;
Manuel Klimek31c85922013-08-29 15:21:40 +0000213
Alexander Kornienko4b672072013-06-03 16:45:03 +0000214 /// \brief The length of following tokens until the next natural split point,
215 /// or the next token that can be broken.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000216 unsigned UnbreakableTailLength = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000217
218 // FIXME: Come up with a 'cleaner' concept.
219 /// \brief The binding strength of a token. This is a combined value of
220 /// operator precedence, parenthesis nesting, etc.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000221 unsigned BindingStrength = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000222
Daniel Jasper63af7c42013-12-09 14:40:19 +0000223 /// \brief The nesting level of this token, i.e. the number of surrounding (),
224 /// [], {} or <>.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000225 unsigned NestingLevel = 0;
Daniel Jasper63af7c42013-12-09 14:40:19 +0000226
Daniel Jasper7d42f3f2017-01-31 11:25:01 +0000227 /// \brief The indent level of this token. Copied from the surrounding line.
228 unsigned IndentLevel = 0;
229
Alexander Kornienko4b672072013-06-03 16:45:03 +0000230 /// \brief Penalty for inserting a line break before this token.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000231 unsigned SplitPenalty = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000232
233 /// \brief If this is the first ObjC selector name in an ObjC method
234 /// definition or call, this contains the length of the longest name.
Daniel Jasperb88b25f2013-12-23 07:29:06 +0000235 ///
236 /// This being set to 0 means that the selectors should not be colon-aligned,
237 /// e.g. because several of them are block-type.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000238 unsigned LongestObjCSelectorName = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000239
240 /// \brief Stores the number of required fake parentheses and the
241 /// corresponding operator precedence.
242 ///
243 /// If multiple fake parentheses start at a token, this vector stores them in
244 /// reverse order, i.e. inner fake parenthesis first.
245 SmallVector<prec::Level, 4> FakeLParens;
246 /// \brief Insert this many fake ) after this token for correct indentation.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000247 unsigned FakeRParens = 0;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000248
Daniel Jasper562ecd42013-09-06 08:08:14 +0000249 /// \brief \c true if this token starts a binary expression, i.e. has at least
250 /// one fake l_paren with a precedence greater than prec::Unknown.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000251 bool StartsBinaryExpression = false;
Daniel Jasper562ecd42013-09-06 08:08:14 +0000252 /// \brief \c true if this token ends a binary expression.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000253 bool EndsBinaryExpression = false;
Daniel Jasper562ecd42013-09-06 08:08:14 +0000254
Daniel Jasper0e617842014-04-16 12:26:54 +0000255 /// \brief Is this is an operator (or "."/"->") in a sequence of operators
256 /// with the same precedence, contains the 0-based operator index.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000257 unsigned OperatorIndex = 0;
Daniel Jasper0e617842014-04-16 12:26:54 +0000258
Daniel Jasper00492f92016-01-05 13:03:50 +0000259 /// \brief If this is an operator (or "."/"->") in a sequence of operators
260 /// with the same precedence, points to the next operator.
261 FormatToken *NextOperator = nullptr;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000262
263 /// \brief Is this token part of a \c DeclStmt defining multiple variables?
264 ///
265 /// Only set if \c Type == \c TT_StartOfName.
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000266 bool PartOfMultiVariableDeclStmt = false;
Alexander Kornienko4b672072013-06-03 16:45:03 +0000267
Krasimir Georgievb6ccd382017-02-02 14:36:50 +0000268 /// \brief Does this line comment continue a line comment section?
269 ///
270 /// Only set to true if \c Type == \c TT_LineComment.
271 bool ContinuesLineCommentSection = false;
272
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000273 /// \brief If this is a bracket, this points to the matching one.
274 FormatToken *MatchingParen = nullptr;
275
276 /// \brief The previous token in the unwrapped line.
277 FormatToken *Previous = nullptr;
278
279 /// \brief The next token in the unwrapped line.
280 FormatToken *Next = nullptr;
281
282 /// \brief If this token starts a block, this contains all the unwrapped lines
283 /// in it.
284 SmallVector<AnnotatedLine *, 1> Children;
285
286 /// \brief Stores the formatting decision for the token once it was made.
287 FormatDecision Decision = FD_Unformatted;
288
289 /// \brief If \c true, this token has been fully formatted (indented and
290 /// potentially re-formatted inside), and we do not allow further formatting
291 /// changes.
292 bool Finalized = false;
Daniel Jaspere1e43192014-04-01 12:55:11 +0000293
Alexander Kornienko4b672072013-06-03 16:45:03 +0000294 bool is(tok::TokenKind Kind) const { return Tok.is(Kind); }
Daniel Jasper8354ea82014-11-21 12:14:12 +0000295 bool is(TokenType TT) const { return Type == TT; }
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000296 bool is(const IdentifierInfo *II) const {
297 return II && II == Tok.getIdentifierInfo();
298 }
Daniel Jasperd7884572015-08-14 12:44:06 +0000299 bool is(tok::PPKeywordKind Kind) const {
300 return Tok.getIdentifierInfo() &&
301 Tok.getIdentifierInfo()->getPPKeywordID() == Kind;
302 }
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000303 template <typename A, typename B> bool isOneOf(A K1, B K2) const {
Alexander Kornienko4b672072013-06-03 16:45:03 +0000304 return is(K1) || is(K2);
305 }
Benjamin Kramerc5bc3cd2015-02-15 20:11:14 +0000306 template <typename A, typename B, typename... Ts>
307 bool isOneOf(A K1, B K2, Ts... Ks) const {
308 return is(K1) || isOneOf(K2, Ks...);
Aaron Ballman484ee9b2014-11-24 15:42:34 +0000309 }
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000310 template <typename T> bool isNot(T Kind) const { return !is(Kind); }
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000311
Martin Probst1244ecb2016-05-29 14:41:02 +0000312 /// \c true if this token starts a sequence with the given tokens in order,
313 /// following the ``Next`` pointers, ignoring comments.
314 template <typename A, typename... Ts>
315 bool startsSequence(A K1, Ts... Tokens) const {
316 return startsSequenceInternal(K1, Tokens...);
317 }
318
319 /// \c true if this token ends a sequence with the given tokens in order,
320 /// following the ``Previous`` pointers, ignoring comments.
321 template <typename A, typename... Ts>
322 bool endsSequence(A K1, Ts... Tokens) const {
323 return endsSequenceInternal(K1, Tokens...);
324 }
325
Daniel Jasper04b6a082013-12-20 06:22:01 +0000326 bool isStringLiteral() const { return tok::isStringLiteral(Tok.getKind()); }
Alexander Kornienko4b672072013-06-03 16:45:03 +0000327
328 bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const {
329 return Tok.isObjCAtKeyword(Kind);
330 }
331
332 bool isAccessSpecifier(bool ColonRequired = true) const {
333 return isOneOf(tok::kw_public, tok::kw_protected, tok::kw_private) &&
334 (!ColonRequired || (Next && Next->is(tok::colon)));
335 }
336
Daniel Jaspercb51cf42014-01-16 09:11:55 +0000337 /// \brief Determine whether the token is a simple-type-specifier.
338 bool isSimpleTypeSpecifier() const;
339
Alexander Kornienko4b672072013-06-03 16:45:03 +0000340 bool isObjCAccessSpecifier() const {
341 return is(tok::at) && Next && (Next->isObjCAtKeyword(tok::objc_public) ||
342 Next->isObjCAtKeyword(tok::objc_protected) ||
343 Next->isObjCAtKeyword(tok::objc_package) ||
344 Next->isObjCAtKeyword(tok::objc_private));
345 }
346
347 /// \brief Returns whether \p Tok is ([{ or a template opening <.
348 bool opensScope() const {
Daniel Jasper24de6fb2017-01-31 13:03:07 +0000349 if (is(TT_TemplateString) && TokenText.endswith("${"))
350 return true;
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000351 return isOneOf(tok::l_paren, tok::l_brace, tok::l_square,
352 TT_TemplateOpener);
Alexander Kornienko4b672072013-06-03 16:45:03 +0000353 }
Nico Weber0f987a62013-06-25 19:25:12 +0000354 /// \brief Returns whether \p Tok is )]} or a template closing >.
Alexander Kornienko4b672072013-06-03 16:45:03 +0000355 bool closesScope() const {
Daniel Jasper24de6fb2017-01-31 13:03:07 +0000356 if (is(TT_TemplateString) && TokenText.startswith("}"))
357 return true;
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000358 return isOneOf(tok::r_paren, tok::r_brace, tok::r_square,
359 TT_TemplateCloser);
Alexander Kornienko4b672072013-06-03 16:45:03 +0000360 }
361
Daniel Jasper4c6e0052013-08-27 14:24:43 +0000362 /// \brief Returns \c true if this is a "." or "->" accessing a member.
363 bool isMemberAccess() const {
Daniel Jasper85bcadc2014-07-09 13:07:57 +0000364 return isOneOf(tok::arrow, tok::period, tok::arrowstar) &&
Daniel Jasper05cd9292015-03-26 18:46:28 +0000365 !isOneOf(TT_DesignatedInitializerPeriod, TT_TrailingReturnArrow,
366 TT_LambdaArrow);
Daniel Jasper4c6e0052013-08-27 14:24:43 +0000367 }
368
Alexander Kornienko4b672072013-06-03 16:45:03 +0000369 bool isUnaryOperator() const {
370 switch (Tok.getKind()) {
371 case tok::plus:
372 case tok::plusplus:
373 case tok::minus:
374 case tok::minusminus:
375 case tok::exclaim:
376 case tok::tilde:
377 case tok::kw_sizeof:
378 case tok::kw_alignof:
379 return true;
380 default:
381 return false;
382 }
383 }
Daniel Jasper4c6e0052013-08-27 14:24:43 +0000384
Alexander Kornienko4b672072013-06-03 16:45:03 +0000385 bool isBinaryOperator() const {
386 // Comma is a binary operator, but does not behave as such wrt. formatting.
387 return getPrecedence() > prec::Comma;
388 }
Daniel Jasper4c6e0052013-08-27 14:24:43 +0000389
Alexander Kornienko4b672072013-06-03 16:45:03 +0000390 bool isTrailingComment() const {
Daniel Jasper610381f2014-08-26 09:37:52 +0000391 return is(tok::comment) &&
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000392 (is(TT_LineComment) || !Next || Next->NewlinesBefore > 0);
Alexander Kornienko4b672072013-06-03 16:45:03 +0000393 }
394
Daniel Jasper78b194992014-08-06 14:15:41 +0000395 /// \brief Returns \c true if this is a keyword that can be used
396 /// like a function call (e.g. sizeof, typeid, ...).
397 bool isFunctionLikeKeyword() const {
398 switch (Tok.getKind()) {
399 case tok::kw_throw:
400 case tok::kw_typeid:
401 case tok::kw_return:
402 case tok::kw_sizeof:
403 case tok::kw_alignof:
404 case tok::kw_alignas:
405 case tok::kw_decltype:
406 case tok::kw_noexcept:
407 case tok::kw_static_assert:
408 case tok::kw___attribute:
409 return true;
410 default:
411 return false;
412 }
413 }
414
Daniel Jasper7209bb92016-12-13 11:16:42 +0000415 /// \brief Returns \c true if this is a string literal that's like a label,
416 /// e.g. ends with "=" or ":".
417 bool isLabelString() const {
418 if (!is(tok::string_literal))
419 return false;
420 StringRef Content = TokenText;
421 if (Content.startswith("\"") || Content.startswith("'"))
422 Content = Content.drop_front(1);
423 if (Content.endswith("\"") || Content.endswith("'"))
424 Content = Content.drop_back(1);
425 Content = Content.trim();
426 return Content.size() > 1 &&
427 (Content.back() == ':' || Content.back() == '=');
428 }
429
Daniel Jasper3cea45d2015-05-04 08:51:40 +0000430 /// \brief Returns actual token start location without leading escaped
431 /// newlines and whitespace.
432 ///
433 /// This can be different to Tok.getLocation(), which includes leading escaped
434 /// newlines.
435 SourceLocation getStartOfNonWhitespace() const {
436 return WhitespaceRange.getEnd();
437 }
438
Alexander Kornienko4b672072013-06-03 16:45:03 +0000439 prec::Level getPrecedence() const {
440 return getBinOpPrecedence(Tok.getKind(), true, true);
441 }
442
443 /// \brief Returns the previous token ignoring comments.
Alexander Kornienko1efe0a02013-07-04 14:47:51 +0000444 FormatToken *getPreviousNonComment() const {
Alexander Kornienko4b672072013-06-03 16:45:03 +0000445 FormatToken *Tok = Previous;
Craig Topper2145bc02014-05-09 08:15:10 +0000446 while (Tok && Tok->is(tok::comment))
Alexander Kornienko4b672072013-06-03 16:45:03 +0000447 Tok = Tok->Previous;
448 return Tok;
449 }
450
451 /// \brief Returns the next token ignoring comments.
Alexander Kornienko1efe0a02013-07-04 14:47:51 +0000452 const FormatToken *getNextNonComment() const {
Alexander Kornienko4b672072013-06-03 16:45:03 +0000453 const FormatToken *Tok = Next;
Craig Topper2145bc02014-05-09 08:15:10 +0000454 while (Tok && Tok->is(tok::comment))
Alexander Kornienko4b672072013-06-03 16:45:03 +0000455 Tok = Tok->Next;
456 return Tok;
457 }
458
Daniel Jasperb8f61682013-10-22 15:45:58 +0000459 /// \brief Returns \c true if this tokens starts a block-type list, i.e. a
460 /// list that should be indented with a block indent.
Daniel Jasperbd73bcf2015-10-27 13:42:08 +0000461 bool opensBlockOrBlockTypeList(const FormatStyle &Style) const {
Daniel Jasper98e0b122017-02-20 14:51:16 +0000462 if (is(TT_TemplateString) && opensScope())
463 return true;
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000464 return is(TT_ArrayInitializerLSquare) ||
Daniel Jasper11a0ac62014-12-12 09:40:58 +0000465 (is(tok::l_brace) &&
466 (BlockKind == BK_Block || is(TT_DictLiteral) ||
467 (!Style.Cpp11BracedListStyle && NestingLevel == 0)));
Daniel Jasperb8f61682013-10-22 15:45:58 +0000468 }
469
Daniel Jasperbd73bcf2015-10-27 13:42:08 +0000470 /// \brief Same as opensBlockOrBlockTypeList, but for the closing token.
471 bool closesBlockOrBlockTypeList(const FormatStyle &Style) const {
Daniel Jasper98e0b122017-02-20 14:51:16 +0000472 if (is(TT_TemplateString) && closesScope())
473 return true;
Daniel Jasperbd73bcf2015-10-27 13:42:08 +0000474 return MatchingParen && MatchingParen->opensBlockOrBlockTypeList(Style);
Daniel Jasper1db6c382013-10-22 15:30:28 +0000475 }
476
Alexander Kornienko4b672072013-06-03 16:45:03 +0000477private:
478 // Disallow copying.
Aaron Ballmanabc18922015-02-15 22:54:08 +0000479 FormatToken(const FormatToken &) = delete;
480 void operator=(const FormatToken &) = delete;
Martin Probst1244ecb2016-05-29 14:41:02 +0000481
482 template <typename A, typename... Ts>
483 bool startsSequenceInternal(A K1, Ts... Tokens) const {
484 if (is(tok::comment) && Next)
485 return Next->startsSequenceInternal(K1, Tokens...);
486 return is(K1) && Next && Next->startsSequenceInternal(Tokens...);
487 }
488
489 template <typename A>
490 bool startsSequenceInternal(A K1) const {
491 if (is(tok::comment) && Next)
492 return Next->startsSequenceInternal(K1);
493 return is(K1);
494 }
495
496 template <typename A, typename... Ts>
497 bool endsSequenceInternal(A K1) const {
498 if (is(tok::comment) && Previous)
499 return Previous->endsSequenceInternal(K1);
500 return is(K1);
501 }
502
503 template <typename A, typename... Ts>
504 bool endsSequenceInternal(A K1, Ts... Tokens) const {
505 if (is(tok::comment) && Previous)
506 return Previous->endsSequenceInternal(K1, Tokens...);
507 return is(K1) && Previous && Previous->endsSequenceInternal(Tokens...);
508 }
Alexander Kornienko4b672072013-06-03 16:45:03 +0000509};
510
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000511class ContinuationIndenter;
512struct LineState;
513
514class TokenRole {
515public:
516 TokenRole(const FormatStyle &Style) : Style(Style) {}
517 virtual ~TokenRole();
518
519 /// \brief After the \c TokenAnnotator has finished annotating all the tokens,
520 /// this function precomputes required information for formatting.
521 virtual void precomputeFormattingInfos(const FormatToken *Token);
522
523 /// \brief Apply the special formatting that the given role demands.
524 ///
Daniel Jasper01603472014-01-09 13:42:56 +0000525 /// Assumes that the token having this role is already formatted.
526 ///
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000527 /// Continues formatting from \p State leaving indentation to \p Indenter and
528 /// returns the total penalty that this formatting incurs.
Daniel Jasper01603472014-01-09 13:42:56 +0000529 virtual unsigned formatFromToken(LineState &State,
530 ContinuationIndenter *Indenter,
531 bool DryRun) {
532 return 0;
533 }
534
535 /// \brief Same as \c formatFromToken, but assumes that the first token has
536 /// already been set thereby deciding on the first line break.
537 virtual unsigned formatAfterToken(LineState &State,
538 ContinuationIndenter *Indenter,
539 bool DryRun) {
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000540 return 0;
541 }
542
543 /// \brief Notifies the \c Role that a comma was found.
544 virtual void CommaFound(const FormatToken *Token) {}
545
546protected:
547 const FormatStyle &Style;
548};
549
550class CommaSeparatedList : public TokenRole {
551public:
Daniel Jasper01603472014-01-09 13:42:56 +0000552 CommaSeparatedList(const FormatStyle &Style)
553 : TokenRole(Style), HasNestedBracedList(false) {}
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000554
Craig Topperfb6b25b2014-03-15 04:29:04 +0000555 void precomputeFormattingInfos(const FormatToken *Token) override;
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000556
Craig Topperfb6b25b2014-03-15 04:29:04 +0000557 unsigned formatAfterToken(LineState &State, ContinuationIndenter *Indenter,
558 bool DryRun) override;
Daniel Jasper01603472014-01-09 13:42:56 +0000559
Craig Topperfb6b25b2014-03-15 04:29:04 +0000560 unsigned formatFromToken(LineState &State, ContinuationIndenter *Indenter,
561 bool DryRun) override;
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000562
563 /// \brief Adds \p Token as the next comma to the \c CommaSeparated list.
Daniel Jasperb05a81d2014-05-09 13:11:16 +0000564 void CommaFound(const FormatToken *Token) override {
565 Commas.push_back(Token);
566 }
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000567
568private:
569 /// \brief A struct that holds information on how to format a given list with
570 /// a specific number of columns.
571 struct ColumnFormat {
572 /// \brief The number of columns to use.
573 unsigned Columns;
574
575 /// \brief The total width in characters.
576 unsigned TotalWidth;
577
578 /// \brief The number of lines required for this format.
579 unsigned LineCount;
580
581 /// \brief The size of each column in characters.
582 SmallVector<unsigned, 8> ColumnSizes;
583 };
584
585 /// \brief Calculate which \c ColumnFormat fits best into
586 /// \p RemainingCharacters.
587 const ColumnFormat *getColumnFormat(unsigned RemainingCharacters) const;
588
589 /// \brief The ordered \c FormatTokens making up the commas of this list.
590 SmallVector<const FormatToken *, 8> Commas;
591
592 /// \brief The length of each of the list's items in characters including the
593 /// trailing comma.
594 SmallVector<unsigned, 8> ItemLengths;
595
596 /// \brief Precomputed formats that can be used for this list.
597 SmallVector<ColumnFormat, 4> Formats;
Daniel Jasper01603472014-01-09 13:42:56 +0000598
599 bool HasNestedBracedList;
Daniel Jasper8de9ed02013-08-22 15:00:41 +0000600};
601
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000602/// \brief Encapsulates keywords that are context sensitive or for languages not
603/// properly supported by Clang's lexer.
604struct AdditionalKeywords {
605 AdditionalKeywords(IdentifierTable &IdentTable) {
Daniel Jasper5d7f06f2015-08-24 14:28:08 +0000606 kw_final = &IdentTable.get("final");
607 kw_override = &IdentTable.get("override");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000608 kw_in = &IdentTable.get("in");
Daniel Jasperb7fda112016-02-11 13:24:15 +0000609 kw_of = &IdentTable.get("of");
Daniel Jasper31f6c542014-12-05 10:42:21 +0000610 kw_CF_ENUM = &IdentTable.get("CF_ENUM");
611 kw_CF_OPTIONS = &IdentTable.get("CF_OPTIONS");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000612 kw_NS_ENUM = &IdentTable.get("NS_ENUM");
Daniel Jasper31f6c542014-12-05 10:42:21 +0000613 kw_NS_OPTIONS = &IdentTable.get("NS_OPTIONS");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000614
Martin Probstc4a0dd42016-05-20 11:24:24 +0000615 kw_as = &IdentTable.get("as");
Martin Probst5f8445b2016-04-24 22:05:09 +0000616 kw_async = &IdentTable.get("async");
617 kw_await = &IdentTable.get("await");
Martin Probst72fd75a2016-11-10 16:20:58 +0000618 kw_declare = &IdentTable.get("declare");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000619 kw_finally = &IdentTable.get("finally");
Daniel Jasper8fc7a1e2016-03-22 14:32:20 +0000620 kw_from = &IdentTable.get("from");
Martin Probst5f8445b2016-04-24 22:05:09 +0000621 kw_function = &IdentTable.get("function");
Martin Probst19c7de02017-04-26 12:34:15 +0000622 kw_get = &IdentTable.get("get");
Daniel Jasper354aa512015-02-19 16:07:32 +0000623 kw_import = &IdentTable.get("import");
Daniel Jasper779c66f2015-12-30 08:00:58 +0000624 kw_is = &IdentTable.get("is");
Daniel Jasper9f642f72015-09-28 14:28:08 +0000625 kw_let = &IdentTable.get("let");
Martin Probst72fd75a2016-11-10 16:20:58 +0000626 kw_module = &IdentTable.get("module");
Martin Probst19c7de02017-04-26 12:34:15 +0000627 kw_set = &IdentTable.get("set");
Martin Probst1b7f9802016-06-23 19:52:32 +0000628 kw_type = &IdentTable.get("type");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000629 kw_var = &IdentTable.get("var");
Martin Probst5f8445b2016-04-24 22:05:09 +0000630 kw_yield = &IdentTable.get("yield");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000631
Daniel Jasper82c92752014-11-21 12:19:07 +0000632 kw_abstract = &IdentTable.get("abstract");
Nico Weber4f113492015-09-15 23:48:17 +0000633 kw_assert = &IdentTable.get("assert");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000634 kw_extends = &IdentTable.get("extends");
635 kw_implements = &IdentTable.get("implements");
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000636 kw_instanceof = &IdentTable.get("instanceof");
Nico Webera644d7f2014-11-10 16:30:02 +0000637 kw_interface = &IdentTable.get("interface");
Nico Webered501662015-01-13 22:32:50 +0000638 kw_native = &IdentTable.get("native");
Daniel Jasper9b9e0762014-11-26 18:03:42 +0000639 kw_package = &IdentTable.get("package");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000640 kw_synchronized = &IdentTable.get("synchronized");
641 kw_throws = &IdentTable.get("throws");
Nico Weberfac23712015-02-04 15:26:27 +0000642 kw___except = &IdentTable.get("__except");
Nico Weber21088802017-02-10 19:36:52 +0000643 kw___has_include = &IdentTable.get("__has_include");
644 kw___has_include_next = &IdentTable.get("__has_include_next");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000645
Daniel Jasperee4a8a12015-04-22 09:45:42 +0000646 kw_mark = &IdentTable.get("mark");
647
Daniel Jasperd8d98392015-11-20 14:32:54 +0000648 kw_extend = &IdentTable.get("extend");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000649 kw_option = &IdentTable.get("option");
650 kw_optional = &IdentTable.get("optional");
651 kw_repeated = &IdentTable.get("repeated");
652 kw_required = &IdentTable.get("required");
653 kw_returns = &IdentTable.get("returns");
Daniel Jasper53395402015-04-07 15:04:40 +0000654
655 kw_signals = &IdentTable.get("signals");
Daniel Jaspera00de632015-12-01 12:05:04 +0000656 kw_qsignals = &IdentTable.get("Q_SIGNALS");
Daniel Jasper53395402015-04-07 15:04:40 +0000657 kw_slots = &IdentTable.get("slots");
658 kw_qslots = &IdentTable.get("Q_SLOTS");
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000659 }
660
Nico Weberfac23712015-02-04 15:26:27 +0000661 // Context sensitive keywords.
Daniel Jasper5d7f06f2015-08-24 14:28:08 +0000662 IdentifierInfo *kw_final;
663 IdentifierInfo *kw_override;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000664 IdentifierInfo *kw_in;
Daniel Jasperb7fda112016-02-11 13:24:15 +0000665 IdentifierInfo *kw_of;
Daniel Jasper31f6c542014-12-05 10:42:21 +0000666 IdentifierInfo *kw_CF_ENUM;
667 IdentifierInfo *kw_CF_OPTIONS;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000668 IdentifierInfo *kw_NS_ENUM;
Daniel Jasper31f6c542014-12-05 10:42:21 +0000669 IdentifierInfo *kw_NS_OPTIONS;
Nico Weberfac23712015-02-04 15:26:27 +0000670 IdentifierInfo *kw___except;
Nico Weber21088802017-02-10 19:36:52 +0000671 IdentifierInfo *kw___has_include;
672 IdentifierInfo *kw___has_include_next;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000673
674 // JavaScript keywords.
Martin Probstc4a0dd42016-05-20 11:24:24 +0000675 IdentifierInfo *kw_as;
Martin Probst5f8445b2016-04-24 22:05:09 +0000676 IdentifierInfo *kw_async;
677 IdentifierInfo *kw_await;
Martin Probst72fd75a2016-11-10 16:20:58 +0000678 IdentifierInfo *kw_declare;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000679 IdentifierInfo *kw_finally;
Daniel Jasper8fc7a1e2016-03-22 14:32:20 +0000680 IdentifierInfo *kw_from;
Martin Probst5f8445b2016-04-24 22:05:09 +0000681 IdentifierInfo *kw_function;
Martin Probst19c7de02017-04-26 12:34:15 +0000682 IdentifierInfo *kw_get;
Daniel Jasper354aa512015-02-19 16:07:32 +0000683 IdentifierInfo *kw_import;
Daniel Jasper779c66f2015-12-30 08:00:58 +0000684 IdentifierInfo *kw_is;
Daniel Jasper9f642f72015-09-28 14:28:08 +0000685 IdentifierInfo *kw_let;
Martin Probst72fd75a2016-11-10 16:20:58 +0000686 IdentifierInfo *kw_module;
Martin Probst19c7de02017-04-26 12:34:15 +0000687 IdentifierInfo *kw_set;
Martin Probst1b7f9802016-06-23 19:52:32 +0000688 IdentifierInfo *kw_type;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000689 IdentifierInfo *kw_var;
Martin Probst5f8445b2016-04-24 22:05:09 +0000690 IdentifierInfo *kw_yield;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000691
692 // Java keywords.
Daniel Jasper82c92752014-11-21 12:19:07 +0000693 IdentifierInfo *kw_abstract;
Nico Weber4f113492015-09-15 23:48:17 +0000694 IdentifierInfo *kw_assert;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000695 IdentifierInfo *kw_extends;
696 IdentifierInfo *kw_implements;
Daniel Jaspera98b7b02014-11-25 10:05:17 +0000697 IdentifierInfo *kw_instanceof;
Nico Webera644d7f2014-11-10 16:30:02 +0000698 IdentifierInfo *kw_interface;
Nico Webered501662015-01-13 22:32:50 +0000699 IdentifierInfo *kw_native;
Daniel Jasper9b9e0762014-11-26 18:03:42 +0000700 IdentifierInfo *kw_package;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000701 IdentifierInfo *kw_synchronized;
702 IdentifierInfo *kw_throws;
703
Daniel Jasperee4a8a12015-04-22 09:45:42 +0000704 // Pragma keywords.
705 IdentifierInfo *kw_mark;
706
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000707 // Proto keywords.
Daniel Jasperd8d98392015-11-20 14:32:54 +0000708 IdentifierInfo *kw_extend;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000709 IdentifierInfo *kw_option;
710 IdentifierInfo *kw_optional;
711 IdentifierInfo *kw_repeated;
712 IdentifierInfo *kw_required;
713 IdentifierInfo *kw_returns;
Daniel Jasper53395402015-04-07 15:04:40 +0000714
715 // QT keywords.
716 IdentifierInfo *kw_signals;
Daniel Jaspera00de632015-12-01 12:05:04 +0000717 IdentifierInfo *kw_qsignals;
Daniel Jasper53395402015-04-07 15:04:40 +0000718 IdentifierInfo *kw_slots;
719 IdentifierInfo *kw_qslots;
Daniel Jasperd0ec0d62014-11-04 12:41:02 +0000720};
721
Alexander Kornienko4b672072013-06-03 16:45:03 +0000722} // namespace format
723} // namespace clang
724
Benjamin Kramer2f5db8b2014-08-13 16:25:19 +0000725#endif