Hi djasper,
Allows formatting of C++11 braced init list constructs, like:
vector<int> v { 1, 2, 3 };
f({ 1, 2 });
http://llvm-reviews.chandlerc.com/D848
Files:
lib/Format/UnwrappedLineParser.cpp
lib/Format/UnwrappedLineParser.h
unittests/Format/FormatTest.cpp
Index: lib/Format/UnwrappedLineParser.cpp
===================================================================
--- lib/Format/UnwrappedLineParser.cpp
+++ lib/Format/UnwrappedLineParser.cpp
@@ -72,6 +72,15 @@
return Token;
}
+ virtual unsigned getPosition() {
+ return PreviousTokenSource->getPosition();
+ }
+
+ virtual FormatToken setPosition(unsigned Position) {
+ Token = PreviousTokenSource->setPosition(Position);
+ return Token;
+ }
+
private:
bool eof() { return Token.HasUnescapedNewline; }
@@ -124,15 +133,48 @@
UnwrappedLine *PreBlockLine;
};
+class AllTokenSource : public FormatTokenSource {
+public:
+ AllTokenSource(ArrayRef<FormatToken> Tokens) : Tokens(Tokens), Position(-1) {}
+
+ virtual FormatToken getNextToken() {
+ ++Position;
+ return Tokens[Position];
+ }
+
+ virtual unsigned getPosition() {
+ assert(Position >= 0);
+ return Position;
+ }
+
+ virtual FormatToken setPosition(unsigned P) {
+ Position = P;
+ return Tokens[Position];
+ }
+
+private:
+ ArrayRef<FormatToken> Tokens;
+ int Position;
+};
+
UnwrappedLineParser::UnwrappedLineParser(const FormatStyle &Style,
FormatTokenSource &Tokens,
UnwrappedLineConsumer &Callback)
: Line(new UnwrappedLine), MustBreakBeforeNextToken(false),
CurrentLines(&Lines), StructuralError(false), Style(Style),
- Tokens(&Tokens), Callback(Callback) {}
+ Tokens(NULL), Callback(Callback) {
+ FormatToken Tok;
+ do {
+ Tok = Tokens.getNextToken();
+ AllTokens.push_back(Tok);
+ } while (Tok.Tok.isNot(tok::eof));
+ LBraces.resize(AllTokens.size(), BS_Unknown);
+}
bool UnwrappedLineParser::parse() {
DEBUG(llvm::dbgs() << "----\n");
+ AllTokenSource TokenSource(AllTokens);
+ Tokens = &TokenSource;
readToken();
parseFile();
for (std::vector<UnwrappedLine>::iterator I = Lines.begin(), E = Lines.end();
@@ -183,6 +225,68 @@
} while (!eof());
}
+void UnwrappedLineParser::calculateBraceTypes() {
+ // We'll parse forward through the tokens until we hit
+ // a closing brace or eof - note that getNextToken() will
+ // parse macros, so this will magically work inside macro
+ // definitions, too.
+ unsigned StoredPosition = Tokens->getPosition();
+ unsigned Position = StoredPosition;
+ FormatToken Tok = FormatTok;
+ // Keep a stack of positions of lbrace tokens. We will
+ // update information about whether an lbrace starts a
+ // braced init list or a different block during the loop.
+ SmallVector<unsigned, 8> LBraceStack;
+ assert(Tok.Tok.is(tok::l_brace));
+ do {
+ FormatToken NextTok = Tokens->getNextToken();
+ switch (Tok.Tok.getKind()) {
+ case tok::l_brace:
+ LBraceStack.push_back(Position);
+ break;
+ case tok::r_brace:
+ if (!LBraceStack.empty()) {
+ if (LBraces[LBraceStack.back()] == BS_Unknown) {
+ // If there is a comma, semicolon or right paren after the closing
+ // brace, we assume this is a braced initializer list.
+
+ // FIXME: Note that this currently works only because we do not
+ // use the brace information while inside a braced init list.
+ // Thus, if the parent is a braced init list, we consider all
+ // brace blocks inside it braced init list. That works good enough
+ // for now, but we will need to fix it to correctly handle lambdas.
+ if (NextTok.Tok.is(tok::comma) || NextTok.Tok.is(tok::semi) ||
+ NextTok.Tok.is(tok::r_paren))
+ LBraces[LBraceStack.back()] = BS_BracedInit;
+ else
+ LBraces[LBraceStack.back()] = BS_Block;
+ }
+ LBraceStack.pop_back();
+ }
+ break;
+ case tok::semi:
+ case tok::kw_if:
+ case tok::kw_while:
+ case tok::kw_for:
+ case tok::kw_switch:
+ case tok::kw_try:
+ if (!LBraceStack.empty())
+ LBraces[LBraceStack.back()] = BS_Block;
+ break;
+ default:
+ break;
+ }
+ Tok = NextTok;
+ ++Position;
+ } while (Tok.Tok.isNot(tok::eof));
+ // Assume other blocks for all unclosed opening braces.
+ for (unsigned i = 0, e = LBraceStack.size(); i != e; ++i) {
+ if (LBraces[LBraceStack[i]] == BS_Unknown)
+ LBraces[LBraceStack[i]] = BS_Block;
+ }
+ FormatTok = Tokens->setPosition(StoredPosition);
+}
+
void UnwrappedLineParser::parseBlock(bool MustBeDeclaration,
unsigned AddLevels) {
assert(FormatTok.Tok.is(tok::l_brace) && "'{' expected");
@@ -394,17 +498,21 @@
parseParens();
break;
case tok::l_brace:
- // A block outside of parentheses must be the last part of a
- // structural element.
- // FIXME: Figure out cases where this is not true, and add projections for
- // them (the one we know is missing are lambdas).
- if (Style.BreakBeforeBraces == FormatStyle::BS_Linux ||
- Style.BreakBeforeBraces == FormatStyle::BS_Stroustrup)
+ if (!tryToParseBracedList()) {
+ // A block outside of parentheses must be the last part of a
+ // structural element.
+ // FIXME: Figure out cases where this is not true, and add projections
+ // for them (the one we know is missing are lambdas).
+ if (Style.BreakBeforeBraces == FormatStyle::BS_Linux ||
+ Style.BreakBeforeBraces == FormatStyle::BS_Stroustrup)
+ addUnwrappedLine();
+ parseBlock(/*MustBeDeclaration=*/ false);
addUnwrappedLine();
-
- parseBlock(/*MustBeDeclaration=*/ false);
- addUnwrappedLine();
- return;
+ return;
+ }
+ // Otherwise this was a braced init list, and the structural
+ // element continues.
+ break;
case tok::identifier:
nextToken();
if (Line->Tokens.size() == 1) {
@@ -436,6 +544,17 @@
} while (!eof());
}
+bool UnwrappedLineParser::tryToParseBracedList() {
+ if (LBraces[Tokens->getPosition()] == BS_Unknown) {
+ calculateBraceTypes();
+ }
+ assert(LBraces[Tokens->getPosition()] != BS_Unknown);
+ if (LBraces[Tokens->getPosition()] == BS_Block)
+ return false;
+ parseBracedList();
+ return true;
+}
+
void UnwrappedLineParser::parseBracedList() {
nextToken();
@@ -517,13 +636,15 @@
nextToken();
return;
case tok::l_brace: {
- nextToken();
- ScopedLineState LineState(*this);
- ScopedDeclarationState DeclarationState(*Line, DeclarationScopeStack,
- /*MustBeDeclaration=*/ false);
- Line->Level += 1;
- parseLevel(/*HasOpeningBrace=*/ true);
- Line->Level -= 1;
+ if (!tryToParseBracedList()) {
+ nextToken();
+ ScopedLineState LineState(*this);
+ ScopedDeclarationState DeclarationState(*Line, DeclarationScopeStack,
+ /*MustBeDeclaration=*/ false);
+ Line->Level += 1;
+ parseLevel(/*HasOpeningBrace=*/ true);
+ Line->Level -= 1;
+ }
break;
}
case tok::at:
Index: lib/Format/UnwrappedLineParser.h
===================================================================
--- lib/Format/UnwrappedLineParser.h
+++ lib/Format/UnwrappedLineParser.h
@@ -122,6 +122,12 @@
virtual ~FormatTokenSource() {
}
virtual FormatToken getNextToken() = 0;
+
+ // FIXME: This interface will become an implementation detail of
+ // the UnwrappedLineParser once we switch to generate all tokens
+ // up-front.
+ virtual unsigned getPosition() { return 0; }
+ virtual FormatToken setPosition(unsigned Position) { assert(false); }
};
class UnwrappedLineParser {
@@ -140,6 +146,7 @@
void parsePPDefine();
void parsePPUnknown();
void parseStructuralElement();
+ bool tryToParseBracedList();
void parseBracedList();
void parseReturn();
void parseParens();
@@ -163,6 +170,14 @@
void readToken();
void flushComments(bool NewlineBeforeNext);
void pushToken(const FormatToken &Tok);
+ void calculateBraceTypes();
+
+ // Represents what type of block a left brace opens.
+ enum LBraceState {
+ BS_Unknown,
+ BS_Block,
+ BS_BracedInit
+ };
// FIXME: We are constantly running into bugs where Line.Level is incorrectly
// subtracted from beyond 0. Introduce a method to subtract from Line.Level
@@ -203,6 +218,17 @@
FormatTokenSource *Tokens;
UnwrappedLineConsumer &Callback;
+ // FIXME: This is a temporary measure until le have reworked the ownership
+ // of the format tokens. The goal is to have the actual tokens created and
+ // owned outside of the UnwrappedLineParser, and handed into the
+ // UnwrappedLineParser.
+ SmallVector<FormatToken, 16> AllTokens;
+
+ // FIXME: Currently we cannot store attributes with tokens, as we treat
+ // them as read-only; thus, we now store the brace state indexed by the
+ // position of the token in the stream (see \c AllTokens).
+ SmallVector<LBraceState, 16> LBraces;
+
friend class ScopedLineState;
};
Index: unittests/Format/FormatTest.cpp
===================================================================
--- unittests/Format/FormatTest.cpp
+++ unittests/Format/FormatTest.cpp
@@ -2706,7 +2706,7 @@
verifyIndependentOfContext("Type **A = static_cast<Type **>(P);");
verifyGoogleFormat("Type** A = static_cast<Type**>(P);");
// FIXME: The newline is wrong.
- verifyFormat("auto a = [](int **&, int ***) {}\n;");
+ verifyFormat("auto a = [](int **&, int ***) {};");
verifyIndependentOfContext("InvalidRegions[*R] = 0;");
@@ -3100,6 +3100,14 @@
verifyFormat("return (a)(b) { 1, 2, 3 };");
}
+TEST_F(FormatTest, LayoutCxx11ConstructorBraceInitializers) {
+ verifyFormat("vector<int> x { 1, 2, 3, 4 };");
+ verifyFormat("vector<T> x { {}, {}, {}, {} };");
+ verifyFormat("f({ 1, 2 });");
+ verifyFormat("auto v = Foo { 1 };");
+ verifyFormat("f({ 1, 2 }, { { 2, 3 }, { 4, 5 } }, c, { d });");
+}
+
TEST_F(FormatTest, LayoutTokensFollowingBlockInParentheses) {
// FIXME: This is bad, find a better and more generic solution.
verifyFormat(
@@ -3859,8 +3867,7 @@
"{ @2 : @1 }\n"
"}");
verifyFormat("@{ @\"one\" : @\n"
- "{ @2 : @1 }\n"
- ",\n"
+ "{ @2 : @1 },\n"
"}");
verifyFormat("@{ 1 > 2 ? @\"one\" : @\"two\" : 1 > 2 ? @1 : @2 }");
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
