aboutsummaryrefslogtreecommitdiff
path: root/lib/Format/Format.cpp
diff options
context:
space:
mode:
authorDaniel Jasper <djasper@google.com>2013-01-29 21:01:14 +0000
committerDaniel Jasper <djasper@google.com>2013-01-29 21:01:14 +0000
commit32d28ee6061930b2a9c170cbaaeea028c88a7b89 (patch)
tree897444a96325186bfc608451e86df3922b5c166e /lib/Format/Format.cpp
parentfcf896078e58aeb7adecb1a0ae5c8e0052b17f9f (diff)
Move the token annotator into separate files.
No functional changes. Also removed experimental-warning from all of clang-format's files, as it is no longer accurate. git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@173830 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Format/Format.cpp')
-rw-r--r--lib/Format/Format.cpp1031
1 files changed, 1 insertions, 1030 deletions
diff --git a/lib/Format/Format.cpp b/lib/Format/Format.cpp
index ecd6f5d385..9dfde34e8b 100644
--- a/lib/Format/Format.cpp
+++ b/lib/Format/Format.cpp
@@ -11,13 +11,11 @@
/// \brief This file implements functions declared in Format.h. This will be
/// split into separate files as we go.
///
-/// This is EXPERIMENTAL code under heavy development. It is not in a state yet,
-/// where it can be used to format real code.
-///
//===----------------------------------------------------------------------===//
#define DEBUG_TYPE "format-formatter"
+#include "TokenAnnotator.h"
#include "UnwrappedLineParser.h"
#include "clang/Basic/Diagnostic.h"
#include "clang/Basic/OperatorPrecedence.h"
@@ -34,140 +32,6 @@
namespace clang {
namespace format {
-enum TokenType {
- TT_BinaryOperator,
- TT_BlockComment,
- TT_CastRParen,
- TT_ConditionalExpr,
- TT_CtorInitializerColon,
- TT_ImplicitStringLiteral,
- TT_LineComment,
- TT_ObjCBlockLParen,
- TT_ObjCDecl,
- TT_ObjCMethodSpecifier,
- TT_ObjCMethodExpr,
- TT_ObjCProperty,
- TT_OverloadedOperator,
- TT_PointerOrReference,
- TT_PureVirtualSpecifier,
- TT_RangeBasedForLoopColon,
- TT_StartOfName,
- TT_TemplateCloser,
- TT_TemplateOpener,
- TT_TrailingUnaryOperator,
- TT_UnaryOperator,
- TT_Unknown
-};
-
-enum LineType {
- LT_Invalid,
- LT_Other,
- LT_BuilderTypeCall,
- LT_PreprocessorDirective,
- LT_VirtualFunctionDecl,
- LT_ObjCDecl, // An @interface, @implementation, or @protocol line.
- LT_ObjCMethodDecl,
- LT_ObjCProperty // An @property line.
-};
-
-class AnnotatedToken {
-public:
- explicit AnnotatedToken(const FormatToken &FormatTok)
- : FormatTok(FormatTok), Type(TT_Unknown), SpaceRequiredBefore(false),
- CanBreakBefore(false), MustBreakBefore(false),
- ClosesTemplateDeclaration(false), MatchingParen(NULL),
- ParameterCount(1), Parent(NULL) {
- }
-
- bool is(tok::TokenKind Kind) const { return FormatTok.Tok.is(Kind); }
- bool isNot(tok::TokenKind Kind) const { return FormatTok.Tok.isNot(Kind); }
-
- bool isObjCAtKeyword(tok::ObjCKeywordKind Kind) const {
- return FormatTok.Tok.isObjCAtKeyword(Kind);
- }
-
- FormatToken FormatTok;
-
- TokenType Type;
-
- bool SpaceRequiredBefore;
- bool CanBreakBefore;
- bool MustBreakBefore;
-
- bool ClosesTemplateDeclaration;
-
- AnnotatedToken *MatchingParen;
-
- /// \brief Number of parameters, if this is "(", "[" or "<".
- ///
- /// This is initialized to 1 as we don't need to distinguish functions with
- /// 0 parameters from functions with 1 parameter. Thus, we can simply count
- /// the number of commas.
- unsigned ParameterCount;
-
- /// \brief The total length of the line up to and including this token.
- unsigned TotalLength;
-
- /// \brief Penalty for inserting a line break before this token.
- unsigned SplitPenalty;
-
- std::vector<AnnotatedToken> Children;
- AnnotatedToken *Parent;
-
- const AnnotatedToken *getPreviousNoneComment() const {
- AnnotatedToken *Tok = Parent;
- while (Tok != NULL && Tok->is(tok::comment))
- Tok = Tok->Parent;
- return Tok;
- }
-};
-
-class AnnotatedLine {
-public:
- AnnotatedLine(const UnwrappedLine &Line)
- : First(Line.Tokens.front()), Level(Line.Level),
- InPPDirective(Line.InPPDirective),
- MustBeDeclaration(Line.MustBeDeclaration) {
- assert(!Line.Tokens.empty());
- AnnotatedToken *Current = &First;
- for (std::list<FormatToken>::const_iterator I = ++Line.Tokens.begin(),
- E = Line.Tokens.end();
- I != E; ++I) {
- Current->Children.push_back(AnnotatedToken(*I));
- Current->Children[0].Parent = Current;
- Current = &Current->Children[0];
- }
- Last = Current;
- }
- AnnotatedLine(const AnnotatedLine &Other)
- : First(Other.First), Type(Other.Type), Level(Other.Level),
- InPPDirective(Other.InPPDirective),
- MustBeDeclaration(Other.MustBeDeclaration) {
- Last = &First;
- while (!Last->Children.empty()) {
- Last->Children[0].Parent = Last;
- Last = &Last->Children[0];
- }
- }
-
- AnnotatedToken First;
- AnnotatedToken *Last;
-
- LineType Type;
- unsigned Level;
- bool InPPDirective;
- bool MustBeDeclaration;
-};
-
-static prec::Level getPrecedence(const AnnotatedToken &Tok) {
- return getBinOpPrecedence(Tok.FormatTok.Tok.getKind(), true, true);
-}
-
-bool isBinaryOperator(const AnnotatedToken &Tok) {
- // Comma is a binary operator, but does not behave as such wrt. formatting.
- return getPrecedence(Tok) > prec::Comma;
-}
-
FormatStyle getLLVMStyle() {
FormatStyle LLVMStyle;
LLVMStyle.ColumnLimit = 80;
@@ -338,15 +202,6 @@ private:
tooling::Replacements Replaces;
};
-/// \brief Returns if a token is an Objective-C selector name.
-///
-/// For example, "bar" is a selector name in [foo bar:(4 + 5)].
-static bool isObjCSelectorName(const AnnotatedToken &Tok) {
- return Tok.is(tok::identifier) && !Tok.Children.empty() &&
- Tok.Children[0].is(tok::colon) &&
- Tok.Children[0].Type == TT_ObjCMethodExpr;
-}
-
class UnwrappedLineFormatter {
public:
UnwrappedLineFormatter(const FormatStyle &Style, SourceManager &SourceMgr,
@@ -813,890 +668,6 @@ private:
OptimizationParameters Parameters;
};
-/// \brief Determines extra information about the tokens comprising an
-/// \c UnwrappedLine.
-class TokenAnnotator {
-public:
- TokenAnnotator(const FormatStyle &Style, SourceManager &SourceMgr, Lexer &Lex,
- AnnotatedLine &Line)
- : Style(Style), SourceMgr(SourceMgr), Lex(Lex), Line(Line) {
- }
-
- /// \brief A parser that gathers additional information about tokens.
- ///
- /// The \c TokenAnnotator tries to matches parenthesis and square brakets and
- /// store a parenthesis levels. It also tries to resolve matching "<" and ">"
- /// into template parameter lists.
- class AnnotatingParser {
- public:
- AnnotatingParser(AnnotatedToken &RootToken)
- : CurrentToken(&RootToken), KeywordVirtualFound(false),
- ColonIsObjCMethodExpr(false), ColonIsForRangeExpr(false) {
- }
-
- /// \brief A helper class to manage AnnotatingParser::ColonIsObjCMethodExpr.
- struct ObjCSelectorRAII {
- AnnotatingParser &P;
- bool ColonWasObjCMethodExpr;
-
- ObjCSelectorRAII(AnnotatingParser &P)
- : P(P), ColonWasObjCMethodExpr(P.ColonIsObjCMethodExpr) {
- }
-
- ~ObjCSelectorRAII() { P.ColonIsObjCMethodExpr = ColonWasObjCMethodExpr; }
-
- void markStart(AnnotatedToken &Left) {
- P.ColonIsObjCMethodExpr = true;
- Left.Type = TT_ObjCMethodExpr;
- }
-
- void markEnd(AnnotatedToken &Right) { Right.Type = TT_ObjCMethodExpr; }
- };
-
- bool parseAngle() {
- if (CurrentToken == NULL)
- return false;
- AnnotatedToken *Left = CurrentToken->Parent;
- while (CurrentToken != NULL) {
- if (CurrentToken->is(tok::greater)) {
- Left->MatchingParen = CurrentToken;
- CurrentToken->MatchingParen = Left;
- CurrentToken->Type = TT_TemplateCloser;
- next();
- return true;
- }
- if (CurrentToken->is(tok::r_paren) || CurrentToken->is(tok::r_square) ||
- CurrentToken->is(tok::r_brace))
- return false;
- if (CurrentToken->is(tok::pipepipe) || CurrentToken->is(tok::ampamp) ||
- CurrentToken->is(tok::question) || CurrentToken->is(tok::colon))
- return false;
- if (CurrentToken->is(tok::comma))
- ++Left->ParameterCount;
- if (!consumeToken())
- return false;
- }
- return false;
- }
-
- bool parseParens(bool LookForDecls = false) {
- if (CurrentToken == NULL)
- return false;
- bool StartsObjCMethodExpr = false;
- AnnotatedToken *Left = CurrentToken->Parent;
- if (CurrentToken->is(tok::caret)) {
- // ^( starts a block.
- Left->Type = TT_ObjCBlockLParen;
- } else if (AnnotatedToken *MaybeSel = Left->Parent) {
- // @selector( starts a selector.
- if (MaybeSel->isObjCAtKeyword(tok::objc_selector) && MaybeSel->Parent &&
- MaybeSel->Parent->is(tok::at)) {
- StartsObjCMethodExpr = true;
- }
- }
-
- ObjCSelectorRAII objCSelector(*this);
- if (StartsObjCMethodExpr)
- objCSelector.markStart(*Left);
-
- while (CurrentToken != NULL) {
- // LookForDecls is set when "if (" has been seen. Check for
- // 'identifier' '*' 'identifier' followed by not '=' -- this
- // '*' has to be a binary operator but determineStarAmpUsage() will
- // categorize it as an unary operator, so set the right type here.
- if (LookForDecls && !CurrentToken->Children.empty()) {
- AnnotatedToken &Prev = *CurrentToken->Parent;
- AnnotatedToken &Next = CurrentToken->Children[0];
- if (Prev.Parent->is(tok::identifier) &&
- (Prev.is(tok::star) || Prev.is(tok::amp)) &&
- CurrentToken->is(tok::identifier) && Next.isNot(tok::equal)) {
- Prev.Type = TT_BinaryOperator;
- LookForDecls = false;
- }
- }
-
- if (CurrentToken->is(tok::r_paren)) {
- Left->MatchingParen = CurrentToken;
- CurrentToken->MatchingParen = Left;
-
- if (StartsObjCMethodExpr)
- objCSelector.markEnd(*CurrentToken);
-
- next();
- return true;
- }
- if (CurrentToken->is(tok::r_square) || CurrentToken->is(tok::r_brace))
- return false;
- if (CurrentToken->is(tok::comma))
- ++Left->ParameterCount;
- if (!consumeToken())
- return false;
- }
- return false;
- }
-
- bool parseSquare() {
- if (!CurrentToken)
- return false;
-
- // A '[' could be an index subscript (after an indentifier or after
- // ')' or ']'), or it could be the start of an Objective-C method
- // expression.
- AnnotatedToken *Left = CurrentToken->Parent;
- bool StartsObjCMethodExpr =
- !Left->Parent || Left->Parent->is(tok::colon) ||
- Left->Parent->is(tok::l_square) || Left->Parent->is(tok::l_paren) ||
- Left->Parent->is(tok::kw_return) || Left->Parent->is(tok::kw_throw) ||
- getBinOpPrecedence(Left->Parent->FormatTok.Tok.getKind(), true,
- true) > prec::Unknown;
-
- ObjCSelectorRAII objCSelector(*this);
- if (StartsObjCMethodExpr)
- objCSelector.markStart(*Left);
-
- while (CurrentToken != NULL) {
- if (CurrentToken->is(tok::r_square)) {
- if (!CurrentToken->Children.empty() &&
- CurrentToken->Children[0].is(tok::l_paren)) {
- // An ObjC method call can't be followed by an open parenthesis.
- // FIXME: Do we incorrectly label ":" with this?
- StartsObjCMethodExpr = false;
- Left->Type = TT_Unknown;
- }
- if (StartsObjCMethodExpr)
- objCSelector.markEnd(*CurrentToken);
- Left->MatchingParen = CurrentToken;
- CurrentToken->MatchingParen = Left;
- next();
- return true;
- }
- if (CurrentToken->is(tok::r_paren) || CurrentToken->is(tok::r_brace))
- return false;
- if (CurrentToken->is(tok::comma))
- ++Left->ParameterCount;
- if (!consumeToken())
- return false;
- }
- return false;
- }
-
- bool parseBrace() {
- // Lines are fine to end with '{'.
- if (CurrentToken == NULL)
- return true;
- AnnotatedToken *Left = CurrentToken->Parent;
- while (CurrentToken != NULL) {
- if (CurrentToken->is(tok::r_brace)) {
- Left->MatchingParen = CurrentToken;
- CurrentToken->MatchingParen = Left;
- next();
- return true;
- }
- if (CurrentToken->is(tok::r_paren) || CurrentToken->is(tok::r_square))
- return false;
- if (!consumeToken())
- return false;
- }
- return true;
- }
-
- bool parseConditional() {
- while (CurrentToken != NULL) {
- if (CurrentToken->is(tok::colon)) {
- CurrentToken->Type = TT_ConditionalExpr;
- next();
- return true;
- }
- if (!consumeToken())
- return false;
- }
- return false;
- }
-
- bool parseTemplateDeclaration() {
- if (CurrentToken != NULL && CurrentToken->is(tok::less)) {
- CurrentToken->Type = TT_TemplateOpener;
- next();
- if (!parseAngle())
- return false;
- CurrentToken->Parent->ClosesTemplateDeclaration = true;
- return true;
- }
- return false;
- }
-
- bool consumeToken() {
- AnnotatedToken *Tok = CurrentToken;
- next();
- switch (Tok->FormatTok.Tok.getKind()) {
- case tok::plus:
- case tok::minus:
- // At the start of the line, +/- specific ObjectiveC method
- // declarations.
- if (Tok->Parent == NULL)
- Tok->Type = TT_ObjCMethodSpecifier;
- break;
- case tok::colon:
- // Colons from ?: are handled in parseConditional().
- if (Tok->Parent->is(tok::r_paren))
- Tok->Type = TT_CtorInitializerColon;
- else if (ColonIsObjCMethodExpr)
- Tok->Type = TT_ObjCMethodExpr;
- else if (ColonIsForRangeExpr)
- Tok->Type = TT_RangeBasedForLoopColon;
- break;
- case tok::kw_if:
- case tok::kw_while:
- if (CurrentToken != NULL && CurrentToken->is(tok::l_paren)) {
- next();
- if (!parseParens(/*LookForDecls=*/ true))
- return false;
- }
- break;
- case tok::kw_for:
- ColonIsForRangeExpr = true;
- next();
- if (!parseParens())
- return false;
- break;
- case tok::l_paren:
- if (!parseParens())
- return false;
- break;
- case tok::l_square:
- if (!parseSquare())
- return false;
- break;
- case tok::l_brace:
- if (!parseBrace())
- return false;
- break;
- case tok::less:
- if (parseAngle())
- Tok->Type = TT_TemplateOpener;
- else {
- Tok->Type = TT_BinaryOperator;
- CurrentToken = Tok;
- next();
- }
- break;
- case tok::r_paren:
- case tok::r_square:
- return false;
- case tok::r_brace:
- // Lines can start with '}'.
- if (Tok->Parent != NULL)
- return false;
- break;
- case tok::greater:
- Tok->Type = TT_BinaryOperator;
- break;
- case tok::kw_operator:
- if (CurrentToken != NULL && CurrentToken->is(tok::l_paren)) {
- CurrentToken->Type = TT_OverloadedOperator;
- next();
- if (CurrentToken != NULL && CurrentToken->is(tok::r_paren)) {
- CurrentToken->Type = TT_OverloadedOperator;
- next();
- }
- } else {
- while (CurrentToken != NULL && CurrentToken->isNot(tok::l_paren)) {
- CurrentToken->Type = TT_OverloadedOperator;
- next();
- }
- }
- break;
- case tok::question:
- parseConditional();
- break;
- case tok::kw_template:
- parseTemplateDeclaration();
- break;
- default:
- break;
- }
- return true;
- }
-
- void parseIncludeDirective() {
- next();
- if (CurrentToken != NULL && CurrentToken->is(tok::less)) {
- next();
- while (CurrentToken != NULL) {
- if (CurrentToken->isNot(tok::comment) ||
- !CurrentToken->Children.empty())
- CurrentToken->Type = TT_ImplicitStringLiteral;
- next();
- }
- } else {
- while (CurrentToken != NULL) {
- next();
- }
- }
- }
-
- void parseWarningOrError() {
- next();
- // We still want to format the whitespace left of the first token of the
- // warning or error.
- next();
- while (CurrentToken != NULL) {
- CurrentToken->Type = TT_ImplicitStringLiteral;
- next();
- }
- }
-
- void parsePreprocessorDirective() {
- next();
- if (CurrentToken == NULL)
- return;
- // Hashes in the middle of a line can lead to any strange token
- // sequence.
- if (CurrentToken->FormatTok.Tok.getIdentifierInfo() == NULL)
- return;
- switch (
- CurrentToken->FormatTok.Tok.getIdentifierInfo()->getPPKeywordID()) {
- case tok::pp_include:
- case tok::pp_import:
- parseIncludeDirective();
- break;
- case tok::pp_error:
- case tok::pp_warning:
- parseWarningOrError();
- break;
- default:
- break;
- }
- }
-
- LineType parseLine() {
- int PeriodsAndArrows = 0;
- bool CanBeBuilderTypeStmt = true;
- if (CurrentToken->is(tok::hash)) {
- parsePreprocessorDirective();
- return LT_PreprocessorDirective;
- }
- while (CurrentToken != NULL) {
- if (CurrentToken->is(tok::kw_virtual))
- KeywordVirtualFound = true;
- if (CurrentToken->is(tok::period) || CurrentToken->is(tok::arrow))
- ++PeriodsAndArrows;
- if (getPrecedence(*CurrentToken) > prec::Assignment &&
- CurrentToken->isNot(tok::less) && CurrentToken->isNot(tok::greater))
- CanBeBuilderTypeStmt = false;
- if (!consumeToken())
- return LT_Invalid;
- }
- if (KeywordVirtualFound)
- return LT_VirtualFunctionDecl;
-
- // Assume a builder-type call if there are 2 or more "." and "->".
- if (PeriodsAndArrows >= 2 && CanBeBuilderTypeStmt)
- return LT_BuilderTypeCall;
-
- return LT_Other;
- }
-
- void next() {
- if (CurrentToken != NULL && !CurrentToken->Children.empty())
- CurrentToken = &CurrentToken->Children[0];
- else
- CurrentToken = NULL;
- }
-
- private:
- AnnotatedToken *CurrentToken;
- bool KeywordVirtualFound;
- bool ColonIsObjCMethodExpr;
- bool ColonIsForRangeExpr;
- };
-
- void calculateExtraInformation(AnnotatedToken &Current) {
- Current.SpaceRequiredBefore = spaceRequiredBefore(Current);
-
- if (Current.FormatTok.MustBreakBefore) {
- Current.MustBreakBefore = true;
- } else {
- if (Current.Type == TT_LineComment) {
- Current.MustBreakBefore = Current.FormatTok.NewlinesBefore > 0;
- } else if ((Current.Parent->is(tok::comment) &&
- Current.FormatTok.NewlinesBefore > 0) ||
- (Current.is(tok::string_literal) &&
- Current.Parent->is(tok::string_literal))) {
- Current.MustBreakBefore = true;
- } else {
- Current.MustBreakBefore = false;
- }
- }
- Current.CanBreakBefore = Current.MustBreakBefore || canBreakBefore(Current);
- if (Current.MustBreakBefore)
- Current.TotalLength = Current.Parent->TotalLength + Style.ColumnLimit;
- else
- Current.TotalLength =
- Current.Parent->TotalLength + Current.FormatTok.TokenLength +
- (Current.SpaceRequiredBefore ? 1 : 0);
- // FIXME: Only calculate this if CanBreakBefore is true once static
- // initializers etc. are sorted out.
- Current.SplitPenalty = splitPenalty(Current);
- if (!Current.Children.empty())
- calculateExtraInformation(Current.Children[0]);
- }
-
- void annotate() {
- AnnotatingParser Parser(Line.First);
- Line.Type = Parser.parseLine();
- if (Line.Type == LT_Invalid)
- return;
-
- bool LookForFunctionName = Line.MustBeDeclaration;
- determineTokenTypes(Line.First, /*IsExpression=*/ false,
- LookForFunctionName);
-
- if (Line.First.Type == TT_ObjCMethodSpecifier)
- Line.Type = LT_ObjCMethodDecl;
- else if (Line.First.Type == TT_ObjCDecl)
- Line.Type = LT_ObjCDecl;
- else if (Line.First.Type == TT_ObjCProperty)
- Line.Type = LT_ObjCProperty;
-
- Line.First.SpaceRequiredBefore = true;
- Line.First.MustBreakBefore = Line.First.FormatTok.MustBreakBefore;
- Line.First.CanBreakBefore = Line.First.MustBreakBefore;
-
- Line.First.TotalLength = Line.First.FormatTok.TokenLength;
- if (!Line.First.Children.empty())
- calculateExtraInformation(Line.First.Children[0]);
- }
-
-private:
- /// \brief Calculate the penalty for splitting before \c Tok.
- unsigned splitPenalty(const AnnotatedToken &Tok) {
- const AnnotatedToken &Left = *Tok.Parent;
- const AnnotatedToken &Right = Tok;
-
- if (Left.is(tok::l_brace) && Right.isNot(tok::l_brace))
- return 50;
- if (Left.is(tok::equal) && Right.is(tok::l_brace))
- return 150;
- if (Left.is(tok::coloncolon))
- return 500;
-
- if (Left.Type == TT_RangeBasedForLoopColon)
- return 5;
-
- if (Right.is(tok::arrow) || Right.is(tok::period)) {
- if (Left.is(tok::r_paren) && Line.Type == LT_BuilderTypeCall)
- return 5; // Should be smaller than breaking at a nested comma.
- return 150;
- }
-
- // In for-loops, prefer breaking at ',' and ';'.
- if (Line.First.is(tok::kw_for) &&
- (Left.isNot(tok::comma) && Left.isNot(tok::semi)))
- return 20;
-
- if (Left.is(tok::semi) || Left.is(tok::comma))
- return 0;
-
- // In Objective-C method expressions, prefer breaking before "param:" over
- // breaking after it.
- if (isObjCSelectorName(Right))
- return 0;
- if (Right.is(tok::colon) && Right.Type == TT_ObjCMethodExpr)
- return 20;
-
- if (Left.is(tok::l_paren))
- return 20;
- // FIXME: The penalty for a trailing "<" or "[" being higher than the
- // penalty for a trainling "(" is a temporary workaround until we can
- // properly avoid breaking in array subscripts or template parameters.
- if (Left.is(tok::l_square) || Left.Type == TT_TemplateOpener)
- return 50;
-
- if (Left.Type == TT_ConditionalExpr)
- return prec::Assignment;
- prec::Level Level = getPrecedence(Left);
-
- if (Level != prec::Unknown)
- return Level;
-
- return 3;
- }
-
- void determineTokenTypes(AnnotatedToken &Current, bool IsExpression,
- bool LookForFunctionName) {
- if (getPrecedence(Current) == prec::Assignment) {
- IsExpression = true;
- AnnotatedToken *Previous = Current.Parent;
- while (Previous != NULL) {
- if (Previous->Type == TT_BinaryOperator &&
- (Previous->is(tok::star) || Previous->is(tok::amp))) {
- Previous->Type = TT_PointerOrReference;
- }
- Previous = Previous->Parent;
- }
- }
- if (Current.is(tok::kw_return) || Current.is(tok::kw_throw) ||
- (Current.is(tok::l_paren) && !Line.MustBeDeclaration &&
- (Current.Parent == NULL || Current.Parent->isNot(tok::kw_for))))
- IsExpression = true;
-
- if (Current.Type == TT_Unknown) {
- if (LookForFunctionName && Current.is(tok::l_paren)) {
- findFunctionName(&Current);
- LookForFunctionName = false;
- } else if (Current.is(tok::star) || Current.is(tok::amp)) {
- Current.Type = determineStarAmpUsage(Current, IsExpression);
- } else if (Current.is(tok::minus) || Current.is(tok::plus) ||
- Current.is(tok::caret)) {
- Current.Type = determinePlusMinusCaretUsage(Current);
- } else if (Current.is(tok::minusminus) || Current.is(tok::plusplus)) {
- Current.Type = determineIncrementUsage(Current);
- } else if (Current.is(tok::exclaim)) {
- Current.Type = TT_UnaryOperator;
- } else if (isBinaryOperator(Current)) {
- Current.Type = TT_BinaryOperator;
- } else if (Current.is(tok::comment)) {
- std::string Data(Lexer::getSpelling(Current.FormatTok.Tok, SourceMgr,
- Lex.getLangOpts()));
- if (StringRef(Data).startswith("//"))
- Current.Type = TT_LineComment;
- else
- Current.Type = TT_BlockComment;
- } else if (Current.is(tok::r_paren) &&
- (Current.Parent->Type == TT_PointerOrReference ||
- Current.Parent->Type == TT_TemplateCloser) &&
- (Current.Children.empty() ||
- (Current.Children[0].isNot(tok::equal) &&
- Current.Children[0].isNot(tok::semi) &&
- Current.Children[0].isNot(tok::l_brace)))) {
- // FIXME: We need to get smarter and understand more cases of casts.
- Current.Type = TT_CastRParen;
- } else if (Current.is(tok::at) && Current.Children.size()) {
- switch (Current.Children[0].FormatTok.Tok.getObjCKeywordID()) {
- case tok::objc_interface:
- case tok::objc_implementation:
- case tok::objc_protocol:
- Current.Type = TT_ObjCDecl;
- break;
- case tok::objc_property:
- Current.Type = TT_ObjCProperty;
- break;
- default:
- break;
- }
- }
- }
-
- if (!Current.Children.empty())
- determineTokenTypes(Current.Children[0], IsExpression,
- LookForFunctionName);
- }
-
- /// \brief Starting from \p Current, this searches backwards for an
- /// identifier which could be the start of a function name and marks it.
- void findFunctionName(AnnotatedToken *Current) {
- AnnotatedToken *Parent = Current->Parent;
- while (Parent != NULL && Parent->Parent != NULL) {
- if (Parent->is(tok::identifier) &&
- (Parent->Parent->is(tok::identifier) ||
- Parent->Parent->Type == TT_PointerOrReference ||
- Parent->Parent->Type == TT_TemplateCloser)) {
- Parent->Type = TT_StartOfName;
- break;
- }
- Parent = Parent->Parent;
- }
- }
-
- /// \brief Returns the previous token ignoring comments.
- const AnnotatedToken *getPreviousToken(const AnnotatedToken &Tok) {
- const AnnotatedToken *PrevToken = Tok.Parent;
- while (PrevToken != NULL && PrevToken->is(tok::comment))
- PrevToken = PrevToken->Parent;
- return PrevToken;
- }
-
- /// \brief Returns the next token ignoring comments.
- const AnnotatedToken *getNextToken(const AnnotatedToken &Tok) {
- if (Tok.Children.empty())
- return NULL;
- const AnnotatedToken *NextToken = &Tok.Children[0];
- while (NextToken->is(tok::comment)) {
- if (NextToken->Children.empty())
- return NULL;
- NextToken = &NextToken->Children[0];
- }
- return NextToken;
- }
-
- /// \brief Return the type of the given token assuming it is * or &.
- TokenType
- determineStarAmpUsage(const AnnotatedToken &Tok, bool IsExpression) {
- const AnnotatedToken *PrevToken = getPreviousToken(Tok);
- if (PrevToken == NULL)
- return TT_UnaryOperator;
-
- const AnnotatedToken *NextToken = getNextToken(Tok);
- if (NextToken == NULL)
- return TT_Unknown;
-
- if (NextToken->is(tok::l_square) && NextToken->Type != TT_ObjCMethodExpr)
- return TT_PointerOrReference;
-
- if (PrevToken->is(tok::l_paren) || PrevToken->is(tok::l_square) ||
- PrevToken->is(tok::l_brace) || PrevToken->is(tok::comma) ||
- PrevToken->is(tok::kw_return) || PrevToken->is(tok::colon) ||
- PrevToken->Type == TT_BinaryOperator ||
- PrevToken->Type == TT_UnaryOperator || PrevToken->Type == TT_CastRParen)
- return TT_UnaryOperator;
-
- if (PrevToken->FormatTok.Tok.isLiteral() || PrevToken->is(tok::r_paren) ||
- PrevToken->is(tok::r_square) || NextToken->FormatTok.Tok.isLiteral() ||
- NextToken->is(tok::plus) || NextToken->is(tok::minus) ||
- NextToken->is(tok::plusplus) || NextToken->is(tok::minusminus) ||
- NextToken->is(tok::tilde) || NextToken->is(tok::exclaim) ||
- NextToken->is(tok::l_paren) || NextToken->is(tok::l_square) ||
- NextToken->is(tok::kw_alignof) || NextToken->is(tok::kw_sizeof))
- return TT_BinaryOperator;
-
- if (NextToken->is(tok::comma) || NextToken->is(tok::r_paren) ||
- NextToken->is(tok::greater))
- return TT_PointerOrReference;
-
- // It is very unlikely that we are going to find a pointer or reference type
- // definition on the RHS of an assignment.
- if (IsExpression)
- return TT_BinaryOperator;
-
- return TT_PointerOrReference;
- }
-
- TokenType determinePlusMinusCaretUsage(const AnnotatedToken &Tok) {
- const AnnotatedToken *PrevToken = getPreviousToken(Tok);
- if (PrevToken == NULL)
- return TT_UnaryOperator;
-
- // Use heuristics to recognize unary operators.
- if (PrevToken->is(tok::equal) || PrevToken->is(tok::l_paren) ||
- PrevToken->is(tok::comma) || PrevToken->is(tok::l_square) ||
- PrevToken->is(tok::question) || PrevToken->is(tok::colon) ||
- PrevToken->is(tok::kw_return) || PrevToken->is(tok::kw_case) ||
- PrevToken->is(tok::at) || PrevToken->is(tok::l_brace))
- return TT_UnaryOperator;
-
- // There can't be to consecutive binary operators.
- if (PrevToken->Type == TT_BinaryOperator)
- return TT_UnaryOperator;
-
- // Fall back to marking the token as binary operator.
- return TT_BinaryOperator;
- }
-
- /// \brief Determine whether ++/-- are pre- or post-increments/-decrements.
- TokenType determineIncrementUsage(const AnnotatedToken &Tok) {
- const AnnotatedToken *PrevToken = getPreviousToken(Tok);
- if (PrevToken == NULL)
- return TT_UnaryOperator;
- if (PrevToken->is(tok::r_paren) || PrevToken->is(tok::r_square) ||
- PrevToken->is(tok::identifier))
- return TT_TrailingUnaryOperator;
-
- return TT_UnaryOperator;
- }
-
- bool spaceRequiredBetween(const AnnotatedToken &Left,
- const AnnotatedToken &Right) {
- if (Right.is(tok::hashhash))
- return Left.is(tok::hash);
- if (Left.is(tok::hashhash) || Left.is(tok::hash))
- return Right.is(tok::hash);
- if (Right.is(tok::r_paren) || Right.is(tok::semi) || Right.is(tok::comma))
- return false;
- if (Right.is(tok::less) &&
- (Left.is(tok::kw_template) ||
- (Line.Type == LT_ObjCDecl && Style.ObjCSpaceBeforeProtocolList)))
- return true;
- if (Left.is(tok::arrow) || Right.is(tok::arrow))
- return false;
- if (Left.is(tok::exclaim) || Left.is(tok::tilde))
- return false;
- if (Left.is(tok::at) &&
- (Right.is(tok::identifier) || Right.is(tok::string_literal) ||
- Right.is(tok::char_constant) || Right.is(tok::numeric_constant) ||
- Right.is(tok::l_paren) || Right.is(tok::l_brace) ||
- Right.is(tok::kw_true) || Right.is(tok::kw_false)))
- return false;
- if (Left.is(tok::coloncolon))
- return false;
- if (Right.is(tok::coloncolon))
- return Left.isNot(tok::identifier) && Left.isNot(tok::greater);
- if (Left.is(tok::less) || Right.is(tok::greater) || Right.is(tok::less))
- return false;
- if (Right.is(tok::amp) || Right.is(tok::star))
- return Left.FormatTok.Tok.isLiteral() ||
- (Left.isNot(tok::star) && Left.isNot(tok::amp) &&
- !Style.PointerAndReferenceBindToType);
- if (Left.is(tok::amp) || Left.is(tok::star))
- return Right.FormatTok.Tok.isLiteral() ||
- Style.PointerAndReferenceBindToType;
- if (Right.is(tok::star) && Left.is(tok::l_paren))
- return false;
- if (Left.is(tok::l_square) || Right.is(tok::r_square))
- return false;
- if (Right.is(tok::l_square) && Right.Type != TT_ObjCMethodExpr)
- return false;
- if (Left.is(tok::period) || Right.is(tok::period))
- return false;
- if (Left.is(tok::colon))
- return Left.Type != TT_ObjCMethodExpr;
- if (Right.is(tok::colon))
- return Right.Type != TT_ObjCMethodExpr;
- if (Left.is(tok::l_paren))
- return false;
- if (Right.is(tok::l_paren)) {
- return Line.Type == LT_ObjCDecl || Left.is(tok::kw_if) ||
- Left.is(tok::kw_for) || Left.is(tok::kw_while) ||
- Left.is(tok::kw_switch) || Left.is(tok::kw_return) ||
- Left.is(tok::kw_catch) || Left.is(tok::kw_new) ||
- Left.is(tok::kw_delete);
- }
- if (Left.is(tok::at) &&
- Right.FormatTok.Tok.getObjCKeywordID() != tok::objc_not_keyword)
- return false;
- if (Left.is(tok::l_brace) && Right.is(tok::r_brace))
- return false;
- return true;
- }
-
- bool spaceRequiredBefore(const AnnotatedToken &Tok) {
- if (Line.Type == LT_ObjCMethodDecl) {
- if (Tok.is(tok::identifier) && !Tok.Children.empty() &&
- Tok.Children[0].is(tok::colon) && Tok.Parent->is(tok::identifier))
- return true;
- if (Tok.is(tok::colon))
- return false;
- if (Tok.Parent->Type == TT_ObjCMethodSpecifier)
- return true;
- if (Tok.Parent->is(tok::r_paren) && Tok.is(tok::identifier))
- // Don't space between ')' and <id>
- return false;
- if (Tok.Parent->is(tok::colon) && Tok.is(tok::l_paren))
- // Don't space between ':' and '('
- return false;
- }
- if (Line.Type == LT_ObjCProperty &&
- (Tok.is(tok::equal) || Tok.Parent->is(tok::equal)))
- return false;
-
- if (Tok.Parent->is(tok::comma))
- return true;
- if (Tok.Type == TT_CtorInitializerColon || Tok.Type == TT_ObjCBlockLParen)
- return true;
- if (Tok.Type == TT_OverloadedOperator)
- return Tok.is(tok::identifier) || Tok.is(tok::kw_new) ||
- Tok.is(tok::kw_delete) || Tok.is(tok::kw_bool);
- if (Tok.Parent->Type == TT_OverloadedOperator)
- return false;
- if (Tok.is(tok::colon))
- return Line.First.isNot(tok::kw_case) && !Tok.Children.empty() &&
- Tok.Type != TT_ObjCMethodExpr;
- if (Tok.Parent->Type == TT_UnaryOperator ||
- Tok.Parent->Type == TT_CastRParen)
- return false;
- if (Tok.Type == TT_UnaryOperator)
- return Tok.Parent->isNot(tok::l_paren) &&
- Tok.Parent->isNot(tok::l_square) && Tok.Parent->isNot(tok::at) &&
- (Tok.Parent->isNot(tok::colon) ||
- Tok.Parent->Type != TT_ObjCMethodExpr);
- if (Tok.Parent->is(tok::greater) && Tok.is(tok::greater)) {
- return Tok.Type == TT_TemplateCloser && Tok.Parent->Type ==
- TT_TemplateCloser && Style.SplitTemplateClosingGreater;
- }
- if (Tok.Type == TT_BinaryOperator || Tok.Parent->Type == TT_BinaryOperator)
- return true;
- if (Tok.Parent->Type == TT_TemplateCloser && Tok.is(tok::l_paren))
- return false;
- if (Tok.is(tok::less) && Line.First.is(tok::hash))
- return true;
- if (Tok.Type == TT_TrailingUnaryOperator)
- return false;
- return spaceRequiredBetween(*Tok.Parent, Tok);
- }
-
- bool canBreakBefore(const AnnotatedToken &Right) {
- const AnnotatedToken &Left = *Right.Parent;
- if (Line.Type == LT_ObjCMethodDecl) {
- if (Right.is(tok::identifier) && !Right.Children.empty() &&
- Right.Children[0].is(tok::colon) && Left.is(tok::identifier))
- return t