aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--include/clang/Format/Format.h75
-rw-r--r--lib/CMakeLists.txt1
-rw-r--r--lib/Format/CMakeLists.txt24
-rw-r--r--lib/Format/Format.cpp704
-rw-r--r--lib/Format/Makefile13
-rw-r--r--lib/Format/UnwrappedLineParser.cpp341
-rw-r--r--lib/Format/UnwrappedLineParser.h119
-rwxr-xr-xlib/Makefile2
-rw-r--r--unittests/CMakeLists.txt1
-rw-r--r--unittests/Format/CMakeLists.txt17
-rw-r--r--unittests/Format/FormatTest.cpp336
-rw-r--r--unittests/Format/Makefile19
12 files changed, 1651 insertions, 1 deletions
diff --git a/include/clang/Format/Format.h b/include/clang/Format/Format.h
new file mode 100644
index 0000000000..c2deccfda2
--- /dev/null
+++ b/include/clang/Format/Format.h
@@ -0,0 +1,75 @@
+//===--- Format.h - Format C++ code -----------------------------*- C++ -*-===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Various functions to configurably format source code.
+///
+/// This is EXPERIMENTAL code under heavy development. It is not in a state yet,
+/// where it can be used to format real code.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_FORMAT_FORMAT_H_
+#define LLVM_CLANG_FORMAT_FORMAT_H
+
+#include "clang/Frontend/FrontendAction.h"
+#include "clang/Tooling/Refactoring.h"
+
+namespace clang {
+
+class Lexer;
+class SourceManager;
+
+namespace format {
+
+/// \brief The \c FormatStyle is used to configure the formatting to follow
+/// specific guidelines.
+struct FormatStyle {
+ /// \brief The column limit.
+ unsigned ColumnLimit;
+
+ /// \brief The maximum number of consecutive empty lines to keep.
+ unsigned MaxEmptyLinesToKeep;
+
+ /// \brief Set whether & and * bind to the type as opposed to the variable.
+ bool PointerAndReferenceBindToType;
+
+ /// \brief The extra indent or outdent of access modifiers (e.g.: public:).
+ int AccessModifierOffset;
+
+ /// \brief Split two consecutive closing '>' by a space, i.e. use
+ /// A<A<int> > instead of A<A<int>>.
+ bool SplitTemplateClosingGreater;
+};
+
+/// \brief Returns a format style complying with the LLVM coding standards:
+/// http://llvm.org/docs/CodingStandards.html.
+FormatStyle getLLVMStyle();
+
+/// \brief Returns a format style complying with Google's C++ style guide:
+/// http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml.
+FormatStyle getGoogleStyle();
+
+/// \brief Reformats the given \p Ranges in the token stream coming out of
+/// \c Lex.
+///
+/// Each range is extended on either end to its next bigger logic unit, i.e.
+/// everything that might influence its formatting or might be influenced by its
+/// formatting.
+///
+/// Returns the \c Replacements necessary to make all \p Ranges comply with
+/// \p Style.
+tooling::Replacements reformat(const FormatStyle &Style, Lexer &Lex,
+ SourceManager &SourceMgr,
+ std::vector<CharSourceRange> Ranges);
+
+} // end namespace format
+} // end namespace clang
+
+#endif // LLVM_CLANG_FORMAT_FORMAT_H
diff --git a/lib/CMakeLists.txt b/lib/CMakeLists.txt
index 206c22818b..053320ced1 100644
--- a/lib/CMakeLists.txt
+++ b/lib/CMakeLists.txt
@@ -16,3 +16,4 @@ add_subdirectory(Frontend)
add_subdirectory(FrontendTool)
add_subdirectory(Tooling)
add_subdirectory(StaticAnalyzer)
+add_subdirectory(Format)
diff --git a/lib/Format/CMakeLists.txt b/lib/Format/CMakeLists.txt
new file mode 100644
index 0000000000..60ab14da24
--- /dev/null
+++ b/lib/Format/CMakeLists.txt
@@ -0,0 +1,24 @@
+set(LLVM_LINK_COMPONENTS support)
+
+add_clang_library(clangFormat
+ UnwrappedLineParser.cpp
+ Format.cpp
+ )
+
+add_dependencies(clangFormat
+ ClangAttrClasses
+ ClangAttrList
+ ClangDeclNodes
+ ClangDiagnosticCommon
+ ClangDiagnosticFrontend
+ ClangStmtNodes
+ )
+
+target_link_libraries(clangFormat
+ clangBasic
+ clangFrontend
+ clangAST
+ clangASTMatchers
+ clangRewriteCore
+ clangRewriteFrontend
+ )
diff --git a/lib/Format/Format.cpp b/lib/Format/Format.cpp
new file mode 100644
index 0000000000..a967768eec
--- /dev/null
+++ b/lib/Format/Format.cpp
@@ -0,0 +1,704 @@
+//===--- Format.cpp - Format C++ code -------------------------------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// \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.
+///
+//===----------------------------------------------------------------------===//
+
+#include "clang/Format/Format.h"
+
+#include "clang/Basic/SourceManager.h"
+#include "clang/Lex/Lexer.h"
+
+#include "UnwrappedLineParser.h"
+
+namespace clang {
+namespace format {
+
+// FIXME: Move somewhere sane.
+struct TokenAnnotation {
+ enum TokenType { TT_Unknown, TT_TemplateOpener, TT_BinaryOperator,
+ TT_UnaryOperator, TT_OverloadedOperator, TT_PointerOrReference,
+ TT_ConditionalExpr, TT_LineComment, TT_BlockComment };
+
+ TokenType Type;
+
+ /// \brief The current parenthesis level, i.e. the number of opening minus
+ /// the number of closing parenthesis left of the current position.
+ unsigned ParenLevel;
+
+ bool SpaceRequiredBefore;
+ bool CanBreakBefore;
+ bool MustBreakBefore;
+};
+
+using llvm::MutableArrayRef;
+
+FormatStyle getLLVMStyle() {
+ FormatStyle LLVMStyle;
+ LLVMStyle.ColumnLimit = 80;
+ LLVMStyle.MaxEmptyLinesToKeep = 1;
+ LLVMStyle.PointerAndReferenceBindToType = false;
+ LLVMStyle.AccessModifierOffset = -2;
+ LLVMStyle.SplitTemplateClosingGreater = true;
+ return LLVMStyle;
+}
+
+FormatStyle getGoogleStyle() {
+ FormatStyle GoogleStyle;
+ GoogleStyle.ColumnLimit = 80;
+ GoogleStyle.MaxEmptyLinesToKeep = 1;
+ GoogleStyle.PointerAndReferenceBindToType = true;
+ GoogleStyle.AccessModifierOffset = -1;
+ GoogleStyle.SplitTemplateClosingGreater = false;
+ return GoogleStyle;
+}
+
+struct OptimizationParameters {
+ unsigned PenaltyExtraLine;
+ unsigned PenaltyIndentLevel;
+};
+
+class UnwrappedLineFormatter {
+public:
+ UnwrappedLineFormatter(const FormatStyle &Style, SourceManager &SourceMgr,
+ const UnwrappedLine &Line,
+ const std::vector<TokenAnnotation> &Annotations,
+ tooling::Replacements &Replaces)
+ : Style(Style),
+ SourceMgr(SourceMgr),
+ Line(Line),
+ Annotations(Annotations),
+ Replaces(Replaces) {
+ Parameters.PenaltyExtraLine = 100;
+ Parameters.PenaltyIndentLevel = 5;
+ }
+
+ void format() {
+ formatFirstToken();
+ count = 0;
+ IndentState State;
+ State.Column = Line.Level * 2 + Line.Tokens[0].Tok.getLength();
+ State.CtorInitializerOnNewLine = false;
+ State.InCtorInitializer = false;
+ State.ConsumedTokens = 1;
+
+ //State.UsedIndent.push_back(Line.Level * 2);
+ State.Indent.push_back(Line.Level * 2 + 4);
+ State.LastSpace.push_back(Line.Level * 2);
+
+ // Start iterating at 1 as we have correctly formatted of Token #0 above.
+ for (unsigned i = 1, n = Line.Tokens.size(); i != n; ++i) {
+ unsigned NoBreak = calcPenalty(State, false, UINT_MAX);
+ unsigned Break = calcPenalty(State, true, NoBreak);
+ addToken(Break < NoBreak, false, State);
+ }
+ }
+
+private:
+ /// \brief The current state when indenting a unwrapped line.
+ ///
+ /// As the indenting tries different combinations this is copied by value.
+ struct IndentState {
+ /// \brief The number of used columns in the current line.
+ unsigned Column;
+
+ /// \brief The number of tokens already consumed.
+ unsigned ConsumedTokens;
+
+ /// \brief The position to which a specific parenthesis level needs to be
+ /// indented.
+ std::vector<unsigned> Indent;
+
+ std::vector<unsigned> LastSpace;
+
+ bool CtorInitializerOnNewLine;
+ bool InCtorInitializer;
+
+ /// \brief Comparison operator to be able to used \c IndentState in \c map.
+ bool operator<(const IndentState &Other) const {
+ if (Other.ConsumedTokens != ConsumedTokens)
+ return Other.ConsumedTokens > ConsumedTokens;
+ if (Other.Column != Column)
+ return Other.Column > Column;
+ if (Other.Indent.size() != Indent.size())
+ return Other.Indent.size() > Indent.size();
+ for (int i = 0, e = Indent.size(); i != e; ++i) {
+ if (Other.Indent[i] != Indent[i])
+ return Other.Indent[i] > Indent[i];
+ }
+ if (Other.LastSpace.size() != LastSpace.size())
+ return Other.LastSpace.size() > LastSpace.size();
+ for (int i = 0, e = LastSpace.size(); i != e; ++i) {
+ if (Other.LastSpace[i] != LastSpace[i])
+ return Other.LastSpace[i] > LastSpace[i];
+ }
+ return false;
+ }
+ };
+
+ /// Append the next token to \p State.
+ void addToken(bool Newline, bool DryRun, IndentState &State) {
+ unsigned Index = State.ConsumedTokens;
+ const FormatToken &Current = Line.Tokens[Index];
+ const FormatToken &Previous = Line.Tokens[Index - 1];
+ unsigned ParenLevel = Annotations[Index].ParenLevel;
+
+ if (Current.Tok.is(tok::l_paren) || Current.Tok.is(tok::l_square) ||
+ Annotations[Index].Type == TokenAnnotation::TT_TemplateOpener) {
+ State.Indent.push_back(4 + State.LastSpace.back());
+ State.LastSpace.push_back(State.LastSpace.back());
+ }
+
+ if (Newline) {
+ if (Current.Tok.is(tok::string_literal) &&
+ Previous.Tok.is(tok::string_literal))
+ State.Column = State.Column - Previous.Tok.getLength();
+ else if (Previous.Tok.is(tok::equal) && ParenLevel != 0)
+ State.Column = State.Indent[ParenLevel] + 4;
+ else
+ State.Column = State.Indent[ParenLevel];
+ if (!DryRun)
+ replaceWhitespace(Current, 1, State.Column);
+
+ State.Column += Current.Tok.getLength();
+ State.LastSpace[ParenLevel] = State.Indent[ParenLevel];
+ if (Current.Tok.is(tok::colon) &&
+ Annotations[Index].Type != TokenAnnotation::TT_ConditionalExpr) {
+ State.Indent[ParenLevel] += 2;
+ State.CtorInitializerOnNewLine = true;
+ State.InCtorInitializer = true;
+ }
+ } else {
+ unsigned Spaces = Annotations[Index].SpaceRequiredBefore ? 1 : 0;
+ if (Annotations[Index].Type == TokenAnnotation::TT_LineComment)
+ Spaces = 2;
+ if (!DryRun)
+ replaceWhitespace(Current, 0, Spaces);
+ if (Previous.Tok.is(tok::l_paren))
+ State.Indent[ParenLevel] = State.Column;
+ if (Previous.Tok.is(tok::less) &&
+ Annotations[Index - 1].Type == TokenAnnotation::TT_TemplateOpener)
+ State.Indent[ParenLevel] = State.Column;
+ if (Current.Tok.is(tok::colon)) {
+ State.Indent[ParenLevel] = State.Column + 3;
+ State.InCtorInitializer = true;
+ }
+ // Top-level spaces are exempt as that mostly leads to better results.
+ if (Spaces > 0 && ParenLevel != 0)
+ State.LastSpace[ParenLevel] = State.Column + Spaces;
+ State.Column += Current.Tok.getLength() + Spaces;
+ }
+
+ if (Current.Tok.is(tok::r_paren) || Current.Tok.is(tok::r_square) ||
+ Annotations[Index].Type == TokenAnnotation::TT_TemplateOpener) {
+ State.Indent.pop_back();
+ State.LastSpace.pop_back();
+ }
+
+ ++State.ConsumedTokens;
+ }
+
+ typedef std::map<IndentState, unsigned> StateMap;
+ StateMap Memory;
+
+ unsigned splitPenalty(const FormatToken &Token) {
+ if (Token.Tok.is(tok::semi))
+ return 0;
+ if (Token.Tok.is(tok::comma))
+ return 1;
+ if (Token.Tok.is(tok::equal) || Token.Tok.is(tok::l_paren) ||
+ Token.Tok.is(tok::pipepipe) || Token.Tok.is(tok::ampamp))
+ return 2;
+ return 3;
+ }
+
+ /// \brief Calculate the number of lines needed to format the remaining part
+ /// of the unwrapped line.
+ ///
+ /// Assumes the formatting so far has led to
+ /// the \c IndentState \p State. If \p NewLine is set, a new line will be
+ /// added after the previous token.
+ ///
+ /// \param StopAt is used for optimization. If we can determine that we'll
+ /// definitely need at least \p StopAt additional lines, we already know of a
+ /// better solution.
+ unsigned calcPenalty(IndentState State, bool NewLine, unsigned StopAt) {
+ // We are at the end of the unwrapped line, so we don't need any more lines.
+ if (State.ConsumedTokens >= Line.Tokens.size())
+ return 0;
+
+ if (!NewLine && Annotations[State.ConsumedTokens].MustBreakBefore)
+ return UINT_MAX;
+ if (NewLine && !Annotations[State.ConsumedTokens].CanBreakBefore)
+ return UINT_MAX;
+
+ if (State.ConsumedTokens > 0 && !NewLine &&
+ State.CtorInitializerOnNewLine &&
+ Line.Tokens[State.ConsumedTokens - 1].Tok.is(tok::comma))
+ return UINT_MAX;
+
+ if (NewLine && State.InCtorInitializer && !State.CtorInitializerOnNewLine)
+ return UINT_MAX;
+
+ addToken(NewLine, true, State);
+
+ // Exceeding column limit is bad.
+ if (State.Column > Style.ColumnLimit)
+ return UINT_MAX;
+
+ unsigned CurrentPenalty = 0;
+ if (NewLine) {
+ CurrentPenalty += Parameters.PenaltyIndentLevel *
+ Annotations[State.ConsumedTokens - 1].ParenLevel +
+ Parameters.PenaltyExtraLine +
+ splitPenalty(Line.Tokens[State.ConsumedTokens - 2]);
+ }
+
+ if (StopAt <= CurrentPenalty)
+ return UINT_MAX;
+ StopAt -= CurrentPenalty;
+
+ // Has this state already been examined?
+ StateMap::iterator I = Memory.find(State);
+ if (I != Memory.end())
+ return I->second;
+ ++count;
+
+ unsigned NoBreak = calcPenalty(State, false, StopAt);
+ unsigned WithBreak = calcPenalty(State, true, std::min(StopAt, NoBreak));
+ unsigned Result = std::min(NoBreak, WithBreak);
+ if (Result != UINT_MAX)
+ Result += CurrentPenalty;
+ Memory[State] = Result;
+ assert(Memory.find(State) != Memory.end());
+ return Result;
+ }
+
+ /// \brief Replaces the whitespace in front of \p Tok. Only call once for
+ /// each \c FormatToken.
+ void replaceWhitespace(const FormatToken &Tok, unsigned NewLines,
+ unsigned Spaces) {
+ Replaces.insert(tooling::Replacement(
+ SourceMgr, Tok.WhiteSpaceStart, Tok.WhiteSpaceLength,
+ std::string(NewLines, '\n') + std::string(Spaces, ' ')));
+ }
+
+ /// \brief Add a new line and the required indent before the first Token
+ /// of the \c UnwrappedLine.
+ void formatFirstToken() {
+ const FormatToken &Token = Line.Tokens[0];
+ if (Token.WhiteSpaceStart.isValid()) {
+ unsigned Newlines =
+ std::min(Token.NewlinesBefore, Style.MaxEmptyLinesToKeep + 1);
+ unsigned Offset = SourceMgr.getFileOffset(Token.WhiteSpaceStart);
+ if (Newlines == 0 && Offset != 0)
+ Newlines = 1;
+ unsigned Indent = Line.Level * 2;
+ if (Token.Tok.is(tok::kw_public) || Token.Tok.is(tok::kw_protected) ||
+ Token.Tok.is(tok::kw_private))
+ Indent += Style.AccessModifierOffset;
+ replaceWhitespace(Token, Newlines, Indent);
+ }
+ }
+
+ FormatStyle Style;
+ SourceManager &SourceMgr;
+ const UnwrappedLine &Line;
+ const std::vector<TokenAnnotation> &Annotations;
+ tooling::Replacements &Replaces;
+ unsigned int count;
+
+ OptimizationParameters Parameters;
+};
+
+/// \brief Determines extra information about the tokens comprising an
+/// \c UnwrappedLine.
+class TokenAnnotator {
+public:
+ TokenAnnotator(const UnwrappedLine &Line, const FormatStyle &Style,
+ SourceManager &SourceMgr)
+ : Line(Line),
+ Style(Style),
+ SourceMgr(SourceMgr) {
+ }
+
+ /// \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(const SourceManager &SourceMgr,
+ const SmallVector<FormatToken, 16> &Tokens,
+ std::vector<TokenAnnotation> &Annotations)
+ : SourceMgr(SourceMgr),
+ Tokens(Tokens),
+ Annotations(Annotations),
+ Index(0) {
+ }
+
+ bool parseAngle(unsigned Level) {
+ while (Index < Tokens.size()) {
+ if (Tokens[Index].Tok.is(tok::greater)) {
+ Annotations[Index].Type = TokenAnnotation::TT_TemplateOpener;
+ Annotations[Index].ParenLevel = Level;
+ next();
+ return true;
+ }
+ if (Tokens[Index].Tok.is(tok::r_paren) ||
+ Tokens[Index].Tok.is(tok::r_square))
+ return false;
+ if (Tokens[Index].Tok.is(tok::pipepipe) ||
+ Tokens[Index].Tok.is(tok::ampamp) ||
+ Tokens[Index].Tok.is(tok::question) ||
+ Tokens[Index].Tok.is(tok::colon))
+ return false;
+ consumeToken(Level);
+ }
+ return false;
+ }
+
+ bool parseParens(unsigned Level) {
+ while (Index < Tokens.size()) {
+ if (Tokens[Index].Tok.is(tok::r_paren)) {
+ Annotations[Index].ParenLevel = Level;
+ next();
+ return true;
+ }
+ if (Tokens[Index].Tok.is(tok::r_square))
+ return false;
+ consumeToken(Level);
+ }
+ return false;
+ }
+
+ bool parseSquare(unsigned Level) {
+ while (Index < Tokens.size()) {
+ if (Tokens[Index].Tok.is(tok::r_square)) {
+ Annotations[Index].ParenLevel = Level;
+ next();
+ return true;
+ }
+ if (Tokens[Index].Tok.is(tok::r_paren))
+ return false;
+ consumeToken(Level);
+ }
+ return false;
+ }
+
+ bool parseConditional(unsigned Level) {
+ while (Index < Tokens.size()) {
+ if (Tokens[Index].Tok.is(tok::colon)) {
+ Annotations[Index].Type = TokenAnnotation::TT_ConditionalExpr;
+ next();
+ return true;
+ }
+ consumeToken(Level);
+ }
+ return false;
+ }
+
+ void consumeToken(unsigned Level) {
+ Annotations[Index].ParenLevel = Level;
+ unsigned CurrentIndex = Index;
+ next();
+ switch (Tokens[CurrentIndex].Tok.getKind()) {
+ case tok::l_paren:
+ parseParens(Level + 1);
+ break;
+ case tok::l_square:
+ parseSquare(Level + 1);
+ break;
+ case tok::less:
+ if (parseAngle(Level + 1))
+ Annotations[CurrentIndex].Type = TokenAnnotation::TT_TemplateOpener;
+ else {
+ Annotations[CurrentIndex].Type = TokenAnnotation::TT_BinaryOperator;
+ Index = CurrentIndex + 1;
+ }
+ break;
+ case tok::greater:
+ Annotations[CurrentIndex].Type = TokenAnnotation::TT_BinaryOperator;
+ break;
+ case tok::kw_operator:
+ if (!Tokens[Index].Tok.is(tok::l_paren))
+ Annotations[Index].Type = TokenAnnotation::TT_OverloadedOperator;
+ next();
+ break;
+ case tok::question:
+ parseConditional(Level);
+ break;
+ default:
+ break;
+ }
+ }
+
+ void parseLine() {
+ while (Index < Tokens.size()) {
+ consumeToken(0);
+ }
+ }
+
+ void next() {
+ ++Index;
+ }
+
+ private:
+ const SourceManager &SourceMgr;
+ const SmallVector<FormatToken, 16> &Tokens;
+ std::vector<TokenAnnotation> &Annotations;
+ unsigned Index;
+ };
+
+ void annotate() {
+ Annotations.clear();
+ for (int i = 0, e = Line.Tokens.size(); i != e; ++i) {
+ Annotations.push_back(TokenAnnotation());
+ }
+
+ AnnotatingParser Parser(SourceMgr, Line.Tokens, Annotations);
+ Parser.parseLine();
+
+ determineTokenTypes();
+
+ for (int i = 1, e = Line.Tokens.size(); i != e; ++i) {
+ TokenAnnotation &Annotation = Annotations[i];
+
+ Annotation.CanBreakBefore =
+ canBreakBetween(Line.Tokens[i - 1], Line.Tokens[i]);
+
+ if (Line.Tokens[i].Tok.is(tok::colon)) {
+ if (Line.Tokens[0].Tok.is(tok::kw_case) || i == e - 1) {
+ Annotation.SpaceRequiredBefore = false;
+ } else {
+ Annotation.SpaceRequiredBefore = TokenAnnotation::TT_ConditionalExpr;
+ }
+ } else if (Annotations[i - 1].Type == TokenAnnotation::TT_UnaryOperator) {
+ Annotation.SpaceRequiredBefore = false;
+ } else if (Annotation.Type == TokenAnnotation::TT_UnaryOperator) {
+ Annotation.SpaceRequiredBefore =
+ Line.Tokens[i - 1].Tok.isNot(tok::l_paren);
+ } else if (Line.Tokens[i - 1].Tok.is(tok::greater) &&
+ Line.Tokens[i].Tok.is(tok::greater)) {
+ if (Annotation.Type == TokenAnnotation::TT_TemplateOpener &&
+ Annotations[i - 1].Type == TokenAnnotation::TT_TemplateOpener)
+ Annotation.SpaceRequiredBefore = Style.SplitTemplateClosingGreater;
+ else
+ Annotation.SpaceRequiredBefore = false;
+ } else if (
+ Annotation.Type == TokenAnnotation::TT_BinaryOperator ||
+ Annotations[i - 1].Type == TokenAnnotation::TT_BinaryOperator) {
+ Annotation.SpaceRequiredBefore = true;
+ } else if (
+ Annotations[i - 1].Type == TokenAnnotation::TT_TemplateOpener &&
+ Line.Tokens[i].Tok.is(tok::l_paren)) {
+ Annotation.SpaceRequiredBefore = false;
+ } else {
+ Annotation.SpaceRequiredBefore =
+ spaceRequiredBetween(Line.Tokens[i - 1].Tok, Line.Tokens[i].Tok);
+ }
+
+ if (Annotations[i - 1].Type == TokenAnnotation::TT_LineComment ||
+ (Line.Tokens[i].Tok.is(tok::string_literal) &&
+ Line.Tokens[i - 1].Tok.is(tok::string_literal))) {
+ Annotation.MustBreakBefore = true;
+ }
+
+ if (Annotation.MustBreakBefore)
+ Annotation.CanBreakBefore = true;
+ }
+ }
+
+ const std::vector<TokenAnnotation> &getAnnotations() {
+ return Annotations;
+ }
+
+private:
+ void determineTokenTypes() {
+ for (int i = 0, e = Line.Tokens.size(); i != e; ++i) {
+ TokenAnnotation &Annotation = Annotations[i];
+ const FormatToken &Tok = Line.Tokens[i];
+
+ if (Tok.Tok.is(tok::star) || Tok.Tok.is(tok::amp))
+ Annotation.Type = determineStarAmpUsage(i);
+ else if (Tok.Tok.is(tok::minus) && Line.Tokens[i - 1].Tok.is(tok::equal))
+ Annotation.Type = TokenAnnotation::TT_UnaryOperator;
+ else if (isBinaryOperator(Line.Tokens[i]))
+ Annotation.Type = TokenAnnotation::TT_BinaryOperator;
+ else if (Tok.Tok.is(tok::comment)) {
+ StringRef Data(SourceMgr.getCharacterData(Tok.Tok.getLocation()),
+ Tok.Tok.getLength());
+ if (Data.startswith("//"))
+ Annotation.Type = TokenAnnotation::TT_LineComment;
+ else
+ Annotation.Type = TokenAnnotation::TT_BlockComment;
+ }
+ }
+ }
+
+ bool isBinaryOperator(const FormatToken &Tok) {
+ switch (Tok.Tok.getKind()) {
+ case tok::equal:
+ case tok::equalequal:
+ case tok::star:
+ //case tok::amp:
+ case tok::plus:
+ case tok::slash:
+ case tok::minus:
+ case tok::ampamp:
+ case tok::pipe:
+ case tok::pipepipe:
+ case tok::percent:
+ return true;
+ default:
+ return false;
+ }
+ }
+
+ TokenAnnotation::TokenType determineStarAmpUsage(unsigned Index) {
+ if (Index == Annotations.size())
+ return TokenAnnotation::TT_Unknown;
+
+ if (Index == 0 || Line.Tokens[Index - 1].Tok.is(tok::l_paren) ||
+ Line.Tokens[Index - 1].Tok.is(tok::comma) ||
+ Annotations[Index - 1].Type == TokenAnnotation::TT_BinaryOperator)
+ return TokenAnnotation::TT_UnaryOperator;
+
+ if (Line.Tokens[Index - 1].Tok.isLiteral() ||
+ Line.Tokens[Index + 1].Tok.isLiteral())
+ return TokenAnnotation::TT_BinaryOperator;
+
+ return TokenAnnotation::TT_PointerOrReference;
+ }
+
+ bool isIfForOrWhile(Token Tok) {
+ return Tok.is(tok::kw_if) || Tok.is(tok::kw_for) || Tok.is(tok::kw_while);
+ }
+
+ bool spaceRequiredBetween(Token Left, Token Right) {
+ if (Left.is(tok::kw_template) && Right.is(tok::less))
+ 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::less) || Right.is(tok::greater) || Right.is(tok::less))
+ return false;
+ if (Left.is(tok::amp) || Left.is(tok::star))
+ return Right.isLiteral() || Style.PointerAndReferenceBindToType;
+ if (Right.is(tok::star) && Left.is(tok::l_paren))
+ return false;
+ if (Right.is(tok::amp) || Right.is(tok::star))
+ return Left.isLiteral() || !Style.PointerAndReferenceBindToType;
+ if (Left.is(tok::l_square) || Right.is(tok::l_square) ||
+ Right.is(tok::r_square))
+ return false;
+ if (Left.is(tok::coloncolon) || Right.is(tok::coloncolon))
+ return false;
+ if (Left.is(tok::period) || Right.is(tok::period))
+ return false;
+ if (Left.is(tok::colon) || Right.is(tok::colon))
+ return true;
+ if ((Left.is(tok::plusplus) && Right.isAnyIdentifier()) ||
+ (Left.isAnyIdentifier() && Right.is(tok::plusplus)) ||
+ (Left.is(tok::minusminus) && Right.isAnyIdentifier()) ||
+ (Left.isAnyIdentifier() && Right.is(tok::minusminus)))
+ return false;
+ if (Left.is(tok::l_paren))
+ return false;
+ if (Left.is(tok::hash))
+ return false;
+ if (Right.is(tok::r_paren) || Right.is(tok::semi) || Right.is(tok::comma))
+ return false;
+ if (Right.is(tok::l_paren)) {
+ return !Left.isAnyIdentifier() || isIfForOrWhile(Left);
+ }
+ return true;
+ }
+
+ bool canBreakBetween(const FormatToken &Left, const FormatToken &Right) {
+ if (Right.Tok.is(tok::r_paren))
+ return false;
+ if (isBinaryOperator(Left))
+ return true;
+ return Right.Tok.is(tok::colon) || Left.Tok.is(tok::comma) || Left.Tok.is(
+ tok::semi) || Left.Tok.is(tok::equal) || Left.Tok.is(tok::ampamp) ||
+ (Left.Tok.is(tok::l_paren) && !Right.Tok.is(tok::r_paren));
+ }
+
+ const UnwrappedLine &Line;
+ FormatStyle Style;
+ SourceManager &SourceMgr;
+ std::vector<TokenAnnotation> Annotations;
+};
+
+class Formatter : public UnwrappedLineConsumer {
+public:
+ Formatter(const FormatStyle &Style, Lexer &Lex, SourceManager &SourceMgr,
+ const std::vector<CharSourceRange> &Ranges)
+ : Style(Style),
+ Lex(Lex),
+ SourceMgr(SourceMgr),
+ Ranges(Ranges) {
+ }
+
+ tooling::Replacements format() {
+ UnwrappedLineParser Parser(Lex, SourceMgr, *this);
+ Parser.parse();
+ return Replaces;
+ }
+
+private:
+ virtual void formatUnwrappedLine(const UnwrappedLine &TheLine) {
+ if (TheLine.Tokens.size() == 0)
+ return;
+
+ CharSourceRange LineRange =
+ CharSourceRange::getTokenRange(TheLine.Tokens.front().Tok.getLocation(),
+ TheLine.Tokens.back().Tok.getLocation());
+
+ for (unsigned i = 0, e = Ranges.size(); i != e; ++i) {
+ if (SourceMgr.isBeforeInTranslationUnit(LineRange.getEnd(),
+ Ranges[i].getBegin()) ||
+ SourceMgr.isBeforeInTranslationUnit(Ranges[i].getEnd(),
+ LineRange.getBegin()))
+ continue;
+
+ TokenAnnotator Annotator(TheLine, Style, SourceMgr);
+ Annotator.annotate();
+ UnwrappedLineFormatter Formatter(Style, SourceMgr, TheLine,
+ Annotator.getAnnotations(), Replaces);
+ Formatter.format();
+ return;
+ }
+ }
+
+ FormatStyle Style;
+ Lexer &Lex;
+ SourceManager &SourceMgr;
+ tooling::Replacements Replaces;
+ std::vector<CharSourceRange> Ranges;
+};
+
+tooling::Replacements reformat(const FormatStyle &Style, Lexer &Lex,
+ SourceManager &SourceMgr,
+ std::vector<CharSourceRange> Ranges) {
+ Formatter formatter(Style, Lex, SourceMgr, Ranges);
+ return formatter.format();
+}
+
+} // namespace format
+} // namespace clang
diff --git a/lib/Format/Makefile b/lib/Format/Makefile
new file mode 100644
index 0000000000..0d2e7a29bc
--- /dev/null
+++ b/lib/Format/Makefile
@@ -0,0 +1,13 @@
+##===- clang/lib/Tooling/Makefile ---------------------------*- Makefile -*-===##
+#
+# The LLVM Compiler Infrastructure
+#
+# This file is distributed under the University of Illinois Open Source
+# License. See LICENSE.TXT for details.
+#
+##===----------------------------------------------------------------------===##
+
+CLANG_LEVEL := ../..
+LIBRARYNAME := clangTooling
+
+include $(CLANG_LEVEL)/Makefile
diff --git a/lib/Format/UnwrappedLineParser.cpp b/lib/Format/UnwrappedLineParser.cpp
new file mode 100644
index 0000000000..8dc1278e04
--- /dev/null
+++ b/lib/Format/UnwrappedLineParser.cpp
@@ -0,0 +1,341 @@
+//===--- UnwrappedLineParser.cpp - Format C++ code ------------------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// \brief This file contains the implementation of the UnwrappedLineParser,
+/// which turns a stream of tokens into UnwrappedLines.
+///
+/// This is EXPERIMENTAL code under heavy development. It is not in a state yet,
+/// where it can be used to format real code.
+///
+//===----------------------------------------------------------------------===//
+
+#include "UnwrappedLineParser.h"
+
+#include "llvm/Support/raw_ostream.h"
+
+namespace clang {
+namespace format {
+
+UnwrappedLineParser::UnwrappedLineParser(Lexer &Lex, SourceManager &SourceMgr,
+ UnwrappedLineConsumer &Callback)
+ : GreaterStashed(false),
+ Lex(Lex),
+ SourceMgr(SourceMgr),
+ IdentTable(Lex.getLangOpts()),
+ Callback(Callback) {
+ Lex.SetKeepWhitespaceMode(true);
+}
+
+void UnwrappedLineParser::parse() {
+ parseToken();
+ parseLevel();
+}
+
+void UnwrappedLineParser::parseLevel() {
+ do {
+ switch (FormatTok.Tok.getKind()) {
+ case tok::hash:
+ parsePPDirective();
+ break;
+ case tok::comment:
+ parseComment();
+ break;
+ case tok::l_brace:
+ parseBlock();
+ addUnwrappedLine();
+ break;
+ case tok::r_brace:
+ return;
+ default:
+ parseStatement();
+ break;
+ }
+ } while (!eof());
+}
+
+void UnwrappedLineParser::parseBlock() {
+ nextToken();
+
+ // FIXME: Remove this hack to handle namespaces.
+ bool IsNamespace = Line.Tokens[0].Tok.is(tok::kw_namespace);
+
+ addUnwrappedLine();
+
+ if (!IsNamespace)
+ ++Line.Level;
+ parseLevel();
+ if (!IsNamespace)
+ --Line.Level;
+ assert(FormatTok.Tok.is(tok::r_brace) && "expected '}'");
+ nextToken();
+ if (FormatTok.Tok.is(tok::semi))
+ nextToken();
+}
+
+void UnwrappedLineParser::parsePPDirective() {
+ while (!eof()) {
+ nextToken();
+ if (FormatTok.NewlinesBefore > 0) {
+ addUnwrappedLine();
+ return;
+ }
+ }
+}
+
+void UnwrappedLineParser::parseComment() {
+ while (!eof()) {
+ nextToken();
+ if (FormatTok.NewlinesBefore > 0) {
+ addUnwrappedLine();
+ return;
+ }
+ }
+}
+
+void UnwrappedLineParser::parseStatement() {
+ if (FormatTok.Tok.is(tok::kw_public) || FormatTok.Tok.is(tok::kw_protected) ||
+ FormatTok.Tok.is(tok::kw_private)) {
+ parseAccessSpecifier();
+ return;
+ }
+ if (FormatTok.Tok.is(tok::kw_enum)) {
+ parseEnum();
+ return;
+ }
+ int TokenNumber = 0;
+ do {
+ ++TokenNumber;
+ switch (FormatTok.Tok.getKind()) {
+ case tok::semi:
+ nextToken();
+ addUnwrappedLine();
+ return;
+ case tok::l_paren:
+ parseParens();
+ break;
+ case tok::l_brace:
+ parseBlock();
+ addUnwrappedLine();
+ return;
+ case tok::kw_if:
+ parseIfThenElse();
+ return;
+ case tok::kw_do:
+ parseDoWhile();
+ return;
+ case tok::kw_switch:
+ parseSwitch();
+ return;
+ case tok::kw_default:
+ nextToken();
+ parseLabel();
+ return;
+ case tok::kw_case:
+ parseCaseLabel();
+ return;
+ case tok::raw_identifier:
+ nextToken();
+ break;
+ default:
+ nextToken();
+ if (TokenNumber == 1 && FormatTok.Tok.is(tok::colon)) {
+ parseLabel();
+ return;
+ }
+ break;
+ }
+ } while (!eof());
+}
+
+void UnwrappedLineParser::parseParens() {
+ assert(FormatTok.Tok.is(tok::l_paren) && "'(' expected.");
+ nextToken();
+ do {
+ switch (FormatTok.Tok.getKind()) {
+ case tok::l_paren:
+ parseParens();
+ break;
+ case tok::r_paren:
+ nextToken();
+ return;
+ default:
+ nextToken();
+ break;
+ }
+ } while (!eof());
+}
+
+void UnwrappedLineParser::parseIfThenElse() {
+ assert(FormatTok.Tok.is(tok::kw_if) && "'if' expected");
+ nextToken();
+ parseParens();
+ bool NeedsUnwrappedLine = false;
+ if (FormatTok.Tok.is(tok::l_brace)) {
+ parseBlock();
+ NeedsUnwrappedLine = true;
+ } else {
+ addUnwrappedLine();
+ ++Line.Level;
+ parseStatement();
+ --Line.Level;
+ }
+ if (FormatTok.Tok.is(tok::kw_else)) {
+ nextToken();
+ if (FormatTok.Tok.is(tok::l_brace)) {
+ parseBlock();
+ addUnwrappedLine();
+ } else if (FormatTok.Tok.is(tok::kw_if)) {
+ parseIfThenElse();
+ } else {
+ addUnwrappedLine();
+ ++Line.Level;
+ parseStatement();
+ --Line.Level;
+ }
+ } else if (NeedsUnwrappedLine) {
+ addUnwrappedLine();
+ }
+}
+
+void UnwrappedLineParser::parseDoWhile() {
+ assert(FormatTok.Tok.is(tok::kw_do) && "'do' expected");
+ nextToken();
+ if (FormatTok.Tok.is(tok::l_brace)) {
+ parseBlock();
+ } else {
+ addUnwrappedLine();
+ ++Line.Level;
+ parseStatement();
+ --Line.Level;
+ }
+
+ assert(FormatTok.Tok.is(tok::kw_while) && "'while' expected");
+ nextToken();
+ parseStatement();
+}
+
+void UnwrappedLineParser::parseLabel() {
+ // FIXME: remove all asserts.
+ assert(FormatTok.Tok.is(tok::colon) && "':' expected");
+ nextToken();
+ unsigned OldLineLevel = Line.Level;
+ if (Line.Level > 0)
+ --Line.Level;
+ if (FormatTok.Tok.is(tok::l_brace)) {
+ parseBlock();
+ }
+ addUnwrappedLine();
+ Line.Level = OldLineLevel;
+}
+
+void UnwrappedLineParser::parseCaseLabel() {
+ assert(FormatTok.Tok.is(tok::kw_case) && "'case' expected");
+ // FIXME: fix handling of complex expressions here.
+ do {
+ nextToken();
+ } while (!eof() && !FormatTok.Tok.is(tok::colon));
+ parseLabel();
+}
+
+void UnwrappedLineParser::parseSwitch() {
+ assert(FormatTok.Tok.is(tok::kw_switch) && "'switch' expected");
+ nextToken();
+ parseParens();
+ if (FormatTok.Tok.is(tok::l_brace)) {
+ parseBlock();
+ addUnwrappedLine();
+ } else {
+ addUnwrappedLine();
+ ++Line.Level;
+ parseStatement();
+ --Line.Level;
+ }
+}
+
+void UnwrappedLineParser::parseAccessSpecifier() {
+ nextToken();
+ nextToken();
+ addUnwrappedLine();
+}
+
+void UnwrappedLineParser::parseEnum() {
+ do {
+ nextToken();
+ if (FormatTok.Tok.is(tok::semi)) {
+ nextToken();
+ addUnwrappedLine();
+ return;
+ }
+ } while (!eof());
+}
+
+void UnwrappedLineParser::addUnwrappedLine() {
+ // Consume trailing comments.
+ while (!eof() && FormatTok.NewlinesBefore == 0 &&
+ FormatTok.Tok.is(tok::comment)) {
+ nextToken();
+ }
+ Callback.formatUnwrappedLine(Line);
+ Line.Tokens.clear();
+}
+
+bool UnwrappedLineParser::eof() const {
+ return FormatTok.Tok.is(tok::eof);
+}
+
+void UnwrappedLineParser::nextToken() {
+ if (eof())
+ return;
+ Line.Tokens.push_back(FormatTok);
+ parseToken();
+}
+
+void UnwrappedLineParser::parseToken() {
+ if (GreaterStashed) {
+ FormatTok.NewlinesBefore = 0;
+ FormatTok.WhiteSpaceStart = FormatTok.Tok.getLocation().getLocWithOffset(1);
+ FormatTok.WhiteSpaceLength = 0;
+ GreaterStashed = false;
+ return;
+ }
+
+ FormatTok = FormatToken();
+ Lex.LexFromRawLexer(FormatTok.Tok);
+ FormatTok.WhiteSpaceStart = FormatTok.Tok.getLocation();
+
+ // Consume and record whitespace until we find a significant token.
+ while (FormatTok.Tok.is(tok::unknown)) {
+ FormatTok.NewlinesBefore += tokenText().count('\n');
+ FormatTok.WhiteSpaceLength += FormatTok.Tok.getLength();
+
+ if (eof())
+ return;
+ Lex.LexFromRawLexer(FormatTok.Tok);
+ }
+
+ if (FormatTok.Tok.is(tok::raw_identifier)) {
+ const IdentifierInfo &Info = IdentTable.get(tokenText());
+ FormatTok.Tok.setKind(Info.getTokenID());
+ }
+
+ if (FormatTok.Tok.is(tok::greatergreater)) {
+ FormatTok.Tok.setKind(tok::greater);
+ GreaterStashed = true;
+ }
+}
+
+StringRef UnwrappedLineParser::tokenText() {
+ StringRef Data(SourceMgr.getCharacterData(FormatTok.Tok.getLocation()),
+ FormatTok.Tok.getLength());
+ return Data;
+}
+
+} // end namespace format
+} // end namespace clang
diff --git a/lib/Format/UnwrappedLineParser.h b/lib/Format/UnwrappedLineParser.h
new file mode 100644
index 0000000000..1e6899e092
--- /dev/null
+++ b/lib/Format/UnwrappedLineParser.h
@@ -0,0 +1,119 @@
+//===--- UnwrappedLineParser.cpp - Format C++ code ------------------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// \brief This file contains the declaration of the UnwrappedLineParser,
+/// which turns a stream of tokens into UnwrappedLines.
+///
+/// This is EXPERIMENTAL code under heavy development. It is not in a state yet,
+/// where it can be used to format real code.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CLANG_FORMAT_UNWRAPPED_LINE_PARSER_H
+#define LLVM_CLANG_FORMAT_UNWRAPPED_LINE_PARSER_H
+
+#include "clang/Basic/SourceManager.h"
+#include "clang/Basic/IdentifierTable.h"
+#include "clang/Lex/Lexer.h"
+
+namespace clang {
+namespace format {
+
+/// \brief A wrapper around a \c Token storing information about the
+/// whitespace characters preceeding it.
+struct FormatToken {
+ FormatToken() : NewlinesBefore(0), WhiteSpaceLength(0) {
+ }
+
+ /// \brief The \c Token.
+ Token Tok;
+
+ /// \brief The number of newlines immediately before the \c Token.
+ ///
+ /// This can be used to determine what the user wrote in the original code
+ /// and thereby e.g. leave an empty line between two function definitions.
+ unsigned NewlinesBefore;
+
+ /// \brief The location of the start of the whitespace immediately preceeding
+ /// the \c Token.
+ ///
+ /// Used together with \c WhiteSpaceLength to create a \c Replacement.
+ SourceLocation WhiteSpaceStart;
+
+ /// \brief The length in characters of the whitespace immediately preceeding
+ /// the \c Token.
+ unsigned WhiteSpaceLength;
+};
+
+/// \brief An unwrapped line is a sequence of \c Token, that we would like to
+/// put on a single line if there was no column limit.
+///
+/// This is used as a main interface between the \c UnwrappedLineParser and the
+/// \c UnwrappedLineFormatter. The key property is that changing the formatting
+/// within an unwrapped line does not affect any other unwrapped lines.
+struct UnwrappedLine {
+ UnwrappedLine() : Level(0) {
+ }
+
+ /// \brief The \c Token comprising this \c UnwrappedLine.
+ SmallVector<FormatToken, 16> Tokens;
+
+ /// \brief The indent level of the \c UnwrappedLine.
+ unsigned Level;
+};
+
+class UnwrappedLineConsumer {
+public:
+ virtual void formatUnwrappedLine(const UnwrappedLine &Line) = 0;
+};
+
+class UnwrappedLineParser {
+public:
+ UnwrappedLineParser(Lexer &Lex, SourceManager &SourceMgr,
+ UnwrappedLineConsumer &Callback);
+
+ void parse();
+
+private:
+ void parseLevel();
+ void parseBlock();
+ void parsePPDirective();
+ void parseComment();
+ void parseStatement();
+ void parseParens();
+ void parseIfThenElse();
+ void parseDoWhile();
+ void parseLabel();
+ void parseCaseLabel();
+ void parseSwitch();
+ void parseAccessSpecifier();
+ void parseEnum();
+ void addUnwrappedLine();
+ bool eof() const;
+ void nextToken();
+ void parseToken();
+
+ /// Returns the text of \c FormatTok.
+ StringRef tokenText();
+
+ UnwrappedLine Line;
+ FormatToken FormatTok;
+ bool GreaterStashed;
+
+ Lexer &Lex;
+ SourceManager &SourceMgr;
+ IdentifierTable IdentTable;
+ UnwrappedLineConsumer &Callback;
+};
+
+} // end namespace format
+} // end namespace clang
+
+#endif // LLVM_CLANG_FORMAT_UNWRAPPED_LINE_PARSER_H
diff --git a/lib/Makefile b/lib/Makefile
index 1f14aa078e..17db50e8cd 100755
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -10,7 +10,7 @@ CLANG_LEVEL := ..
PARALLEL_DIRS = Headers Basic Lex Parse AST ASTMatchers Sema CodeGen Analysis \
StaticAnalyzer Edit Rewrite ARCMigrate Serialization Frontend \
- FrontendTool Tooling Driver
+ FrontendTool Tooling Driver Format
include $(CLANG_LEVEL)/Makefile
diff --git a/unittests/CMakeLists.txt b/unittests/CMakeLists.txt
index 989025a9eb..334ea41422 100644
--- a/unittests/CMakeLists.txt
+++ b/unittests/CMakeLists.txt
@@ -15,3 +15,4 @@ add_subdirectory(Basic)
add_subdirectory(Lex)
add_subdirectory(Frontend)
add_subdirectory(Tooling)
+add_subdirectory(Format)
diff --git a/unittests/Format/CMakeLists.txt b/unittests/Format/CMakeLists.txt
new file mode 100644
index 0000000000..66dd534d52
--- /dev/null
+++ b/unittests/Format/CMakeLists.txt
@@ -0,0 +1,17 @@
+set(LLVM_LINK_COMPONENTS
+ ${LLVM_TARGETS_TO_BUILD}
+ asmparser
+ support
+ mc
+ )
+
+add_clang_unittest(FormatTests
+ FormatTest.cpp
+ )
+
+target_link_libraries(FormatTests
+ clangAST
+ clangFormat
+ clangTooling
+ clangRewriteCore
+ )
diff --git a/unittests/Format/FormatTest.cpp b/unittests/Format/FormatTest.cpp
new file mode 100644
index 0000000000..9929eebd3b
--- /dev/null
+++ b/unittests/Format/FormatTest.cpp
@@ -0,0 +1,336 @@
+//===- unittest/Format/FormatTest.cpp - Formatting unit tests -------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#include "../Tooling/RewriterTestContext.h"
+#include "clang/Lex/Lexer.h"
+#include "clang/Format/Format.h"
+#include "gtest/gtest.h"
+
+namespace clang {
+namespace format {
+
+class FormatTest : public ::testing::Test {
+protected:
+ std::string format(llvm::StringRef Code, unsigned Offset, unsigned Length,
+ const FormatStyle &Style) {
+ RewriterTestContext Context;
+ FileID ID = Context.createInMemoryFile("input.cc", Code);
+ SourceLocation Start =
+ Context.Sources.getLocForStartOfFile(ID).getLocWithOffset(Offset);
+ std::vector<CharSourceRange> Ranges(
+ 1,
+ CharSourceRange::getCharRange(Start, Start.getLocWithOffset(Length)));
+ LangOptions LangOpts;
+ LangOpts.CPlusPlus = 1;
+ Lexer Lex(ID, Context.Sources.getBuffer(ID), Context.Sources, LangOpts);
+ tooling::Replacements Replace =
+ reformat(Style, Lex, Context.Sources, Ranges);
+ EXPECT_TRUE(applyAllReplacements(Replace, Context.Rewrite));
+ return Context.getRewrittenText(ID);
+ }
+
+ std::string format(llvm::StringRef Code,
+ const FormatStyle &Style = getLLVMStyle()) {
+ return format(Code, 0, Code.size(), Style);
+ }
+
+ void verifyFormat(llvm::StringRef Code) {
+ std::string WithoutFormat(Code.str());
+ for (unsigned i = 0, e = WithoutFormat.size(); i != e; ++i) {
+ if (WithoutFormat[i] == '\n')
+ WithoutFormat[i] = ' ';
+ }
+ EXPECT_EQ(Code.str(), format(WithoutFormat));
+ }
+
+ void verifyGoogleFormat(llvm::StringRef Code) {
+ std::string WithoutFormat(Code.str());
+ for (unsigned i = 0, e = WithoutFormat.size(); i != e; ++i) {
+ if (WithoutFormat[i] == '\n')
+ WithoutFormat[i] = ' ';
+ }
+ EXPECT_EQ(Code.str(), format(WithoutFormat, getGoogleStyle()));
+ }
+};
+
+TEST_F(FormatTest, DoesNotChangeCorrectlyFormatedCode) {
+ EXPECT_EQ(";", format(";"));
+}
+
+TEST_F(FormatTest, FormatsGlobalStatementsAt0) {
+ EXPECT_EQ("int i;", format(" int i;"));
+ EXPECT_EQ("\nint i;", format(" \n\t \r int i;"));
+ EXPECT_EQ("int i;\nint j;", format(" int i; int j;"));
+ EXPECT_EQ("int i;\nint j;", format(" int i;\n int j;"));
+}
+
+TEST_F(FormatTest, FormatsUnwrappedLinesAtFirstFormat) {
+ EXPECT_EQ("int i;", format("int\ni;"));
+}
+
+TEST_F(FormatTest, FormatsNestedBlockStatements) {
+ EXPECT_EQ("{\n {\n {\n }\n }\n}", format("{{{}}}"));
+}
+
+TEST_F(FormatTest, FormatsForLoop) {
+ verifyFormat(
+ "for (int VeryVeryLongLoopVariable = 0; VeryVeryLongLoopVariable < 10;\n"
+ " ++VeryVeryLongLoopVariable);");
+}
+
+TEST_F(FormatTest, FormatsWhileLoop) {
+ verifyFormat("while (true) {\n}");
+}
+
+TEST_F(FormatTest, FormatsNestedCall) {
+ verifyFormat("Method(f1, f2(f3));");
+ verifyFormat("Method(f1(f2, f3()));");
+}
+
+TEST_F(FormatTest, FormatsAwesomeMethodCall) {
+ verifyFormat(
+ "SomeLongMethodName(SomeReallyLongMethod(CallOtherReallyLongMethod(\n"
+ " parameter, parameter, parameter)), SecondLongCall(parameter));");
+}
+
+TEST_F(FormatTest, FormatsFunctionDefinition) {
+ verifyFormat("void f(int a, int b, int c, int d, int e, int f, int g,"
+ " int h, int j, int f,\n"
+ " int c, int ddddddddddddd) {\n"
+ "}");
+}
+
+TEST_F(FormatTest, FormatIfWithoutCompountStatement) {
+ verifyFormat("if (true)\n f();\ng();");
+ verifyFormat("if (a)\n if (b)\n if (c)\n g();\nh();");
+ verifyFormat("if (a)\n if (b) {\n f();\n }\ng();");
+}
+
+TEST_F(FormatTest, ParseIfThenElse) {
+ verifyFormat("if (true)\n"
+ " if (true)\n"
+ " if (true)\n"
+ " f();\n"
+ " else\n"
+ " g();\n"
+ " else\n"
+ " h();\n"
+ "else\n"
+ " i();");
+ verifyFormat("if (true)\n"
+ " if (true)\n"
+ " if (true) {\n"
+ " if (true)\n"
+ " f();\n"
+ " } else {\n"
+ " g();\n"
+ " }\n"
+ " else\n"
+ " h();\n"
+ "else {\n"
+ " i();\n"
+ "}");
+}
+
+TEST_F(FormatTest, UnderstandsSingleLineComments) {
+ EXPECT_EQ("// line 1\n// line 2\nvoid f() {\n}\n",
+ format("// line 1\n// line 2\nvoid f() {}\n"));
+
+ EXPECT_EQ("void f() {\n // Doesn't do anything\n}",
+ format("void f() {\n// Doesn't do anything\n}"));
+
+ EXPECT_EQ("int i // This is a fancy variable\n = 5;",
+ format("int i // This is a fancy variable\n= 5;"));
+
+ verifyFormat("f(/*test=*/ true);");
+}
+
+TEST_F(FormatTest, DoesNotBreakSemiAfterClassDecl) {
+ verifyFormat("class A {\n};");
+}
+
+TEST_F(FormatTest, BreaksAsHighAsPossible) {
+ verifyFormat(
+ "if ((aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa && aaaaaaaaaaaaaaaaaaaaaaaaaa) ||\n"
+ " (bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb && bbbbbbbbbbbbbbbbbbbbbbbbbb))\n"
+ " f();");
+}
+
+TEST_F(FormatTest, ElseIf) {
+ verifyFormat("if (a) {\n"
+ "} else if (b) {\n"
+ "}");
+ verifyFormat("if (a)\n"
+ " f();\n"
+ "else if (b)\n"
+ " g();\n"
+ "else\n"
+ " h();");
+}
+
+TEST_F(FormatTest, UnderstandsAccessSpecifiers) {
+ verifyFormat("class A {\n"
+ "public:\n"
+ "protected:\n"
+ "private:\n"
+ " void f() {\n"
+ " }\n"
+ "};");
+ verifyGoogleFormat("class A {\n"
+ " public:\n"
+ " protected:\n"
+ " private:\n"
+ " void f() {\n"
+ " }\n"
+ "};");
+}
+
+TEST_F(FormatTest, SwitchStatement) {
+ verifyFormat("switch (x) {\n"
+ "case 1:\n"
+ " f();\n"
+ " break;\n"
+ "case kFoo:\n"
+ "case ns::kBar:\n"
+ "case kBaz:\n"
+ " break;\n"
+ "default:\n"
+ " g();\n"
+ " break;\n"
+ "}");
+ verifyFormat("switch (x) {\n"
+ "case 1: {\n"
+ " f();\n"
+ " break;\n"
+ "}\n"
+ "}");
+ verifyFormat("switch (test)\n"
+ " ;");
+}
+
+TEST_F(FormatTest, Labels) {
+ verifyFormat("void f() {\n"
+ " some_code();\n"
+ "test_label:\n"
+ " some_other_code();\n"
+ " {\n"
+ " some_more_code();\n"
+ " another_label:\n"
+ " some_more_code();\n"
+ " }\n"
+ "}");
+ verifyFormat("some_code();\n"
+ "test_label:\n"
+ "some_other_code();");
+}
+
+TEST_F(FormatTest, DerivedClass) {
+ verifyFormat("class A : public B {\n"
+ "};");
+}
+
+TEST_F(FormatTest, DoWhile) {
+ verifyFormat("do {\n"
+ " do_something();\n"
+ "} while (something());");
+ verifyFormat("do\n"
+ " do_something();\n"
+ "while (something());");
+}
+
+TEST_F(FormatTest, BreaksDesireably) {
+ verifyFormat("if (aaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaa) ||\n"
+ " aaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaa) ||\n"
+ " aaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaa)) {\n};");
+
+ verifyFormat(
+ "aaaaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa,\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {\n}");
+
+ verifyFormat("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa));");
+}
+
+TEST_F(FormatTest, AlignsStringLiterals) {
+ verifyFormat("loooooooooooooooooooooooooongFunction(\"short literal \"\n"
+ " \"short literal\");");
+ verifyFormat(
+ "looooooooooooooooooooooooongFunction(\n"
+ " \"short literal\"\n"
+ " \"looooooooooooooooooooooooooooooooooooooooooooooooong literal\");");
+}
+
+TEST_F(FormatTest, UnderstandsEquals) {
+ verifyFormat(
+ "aaaaaaaaaaaaaaaaa =\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa;");
+ verifyFormat(
+ "if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa =\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {\n"
+ "}");
+ verifyFormat(
+ "if (a) {\n"
+ "} else if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa =\n"
+ " aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {\n"
+ "}");
+
+ verifyFormat("if (int aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa =\n"
+ " 100000000 + 100000000) {\n}");
+}
+
+TEST_F(FormatTest, UnderstandsTemplateParameters) {
+ verifyFormat("A<int> a;");
+ verifyFormat("A<A<A<int> > > a;");
+ verifyFormat("A<A<A<int, 2>, 3>, 4> a;");
+ verifyFormat("bool x = a < 1 || 2 > a;");
+ verifyFormat("bool x = 5 < f<int>();");
+ verifyFormat("bool x = f<int>() > 5;");
+ verifyFormat("bool x = 5 < a<int>::x;");
+ verifyFormat("bool x = a < 4 ? a > 2 : false;");
+ verifyFormat("bool x = f() ? a < 2 : a > 2;");
+
+ verifyGoogleFormat("A<A<int>> a;");
+ verifyGoogleFormat("A<A<A<int>>> a;");
+ verifyGoogleFormat("A<A<A<A<int>>>> a;");
+
+ verifyFormat("test >> a >> b;");
+ verifyFormat("test << a >> b;");
+
+ verifyFormat("f<int>();");
+ verifyFormat("template <typename T> void f() {\n}");
+}
+
+TEST_F(FormatTest, UndestandsUnaryOperators) {
+ verifyFormat("int a = -2;");
+}
+
+TEST_F(FormatTest, UndestandsOverloadedOperators) {
+ verifyFormat("bool operator<() {\n}");
+}
+
+TEST_F(FormatTest, UnderstandsUsesOfStar) {
+ verifyFormat("int *f(int *a) {\n}");
+ verifyFormat("f(a, *a);");
+ verifyFormat("f(*a);");
+ verifyFormat("int a = b * 10;");
+ verifyFormat("int a = 10 * b;");
+ // verifyFormat("int a = b * c;");
+ verifyFormat("int a = *b;");
+ // verifyFormat("int a = *b * c;");
+ // verifyFormat("int a = b * *c;");
+}
+
+//TEST_F(FormatTest, IncorrectDerivedClass) {
+// verifyFormat("public B {\n"
+// "};");
+//}
+
+} // end namespace tooling
+} // end namespace clang
diff --git a/unittests/Format/Makefile b/unittests/Format/Makefile
new file mode 100644
index 0000000000..4218622bd2
--- /dev/null
+++ b/unittests/Format/Makefile
@@ -0,0 +1,19 @@
+##===- unittests/Format/Makefile ---------------------------*- Makefile -*-===##
+#
+# The LLVM Compiler Infrastructure
+#
+# This file is distributed under the University of Illinois Open Source
+# License. See LICENSE.TXT for details.
+#
+##===----------------------------------------------------------------------===##
+
+CLANG_LEVEL = ../..
+TESTNAME = Format
+include $(CLANG_LEVEL)/../../Makefile.config
+LINK_COMPONENTS := $(TARGETS_TO_BUILD) asmparser support mc
+USEDLIBS = clangFormat.a clangTooling.a clangFrontend.a clangSerialization.a \
+ clangDriver.a clangParse.a clangRewriteCore.a
+ clangRewriteFrontend.a clangSema.a clangAnalysis.a clangEdit.a \
+ clangAST.a clangASTMatchers.a clangLex.a clangBasic.a
+
+include $(CLANG_LEVEL)/unittests/Makefile
generated by cgit v1.2.3-18-g5258 (git 2.32.0) at 2024-11-27 12:26:08 +0000