Initial implementation of Solidity parser finished, not yet tested much.

AST.h

@@ -54,6 +54,8 @@ public:
 		: m_location(_location)
 	{}
 
+	virtual ~ASTNode() {}
+
 	Location getLocation() const { return m_location; }
 private:
 	Location m_location;
@@ -146,9 +148,7 @@ private:
 class TypeName : public ASTNode
 {
 public:
-	explicit TypeName(Location const& _location)
+	explicit TypeName(Location const& _location) : ASTNode(_location) {}
-		: ASTNode(_location)
-	{}
 };
 
 /// any pre-defined type that is not a mapping
@@ -192,15 +192,13 @@ private:
 class Statement : public ASTNode
 {
 public:
-	explicit Statement(Location const& _location)
+	explicit Statement(Location const& _location) : ASTNode(_location) {}
-		: ASTNode(_location)
-	{}
 };
 
 class Block : public Statement
 {
 public:
-	explicit Block(Location const& _location, vecptr<Statement> const& _statements)
+	Block(Location const& _location, vecptr<Statement> const& _statements)
 		: Statement(_location), m_statements(_statements)
 	{}
 private:
@@ -209,7 +207,12 @@ private:
 
 class IfStatement : public Statement
 {
-
+public:
+	IfStatement(Location const& _location, ptr<Expression> const& _condition,
+				ptr<Statement> const& _trueBody, ptr<Statement> const& _falseBody)
+		: Statement(_location), m_condition(_condition),
+		  m_trueBody(_trueBody), m_falseBody(_falseBody)
+	{}
 private:
 	ptr<Expression> m_condition;
 	ptr<Statement> m_trueBody;
@@ -218,11 +221,17 @@ private:
 
 class BreakableStatement : public Statement
 {
-
+public:
+	BreakableStatement(Location const& _location) : Statement(_location) {}
 };
 
 class WhileStatement : public BreakableStatement
 {
+public:
+	WhileStatement(Location const& _location, ptr<Expression> const& _condition,
+				   ptr<Statement> const& _body)
+		: BreakableStatement(_location), m_condition(_condition), m_body(_body)
+	{}
 private:
 	ptr<Expression> m_condition;
 	ptr<Statement> m_body;
@@ -230,31 +239,42 @@ private:
 
 class Continue : public Statement
 {
-
+public:
+	Continue(Location const& _location) : Statement(_location) {}
 };
 
 class Break : public Statement
 {
-
+public:
+	Break(Location const& _location) : Statement(_location) {}
 };
 
 class Return : public Statement
 {
+public:
+	Return(Location const& _location, ptr<Expression> _expression)
+		: Statement(_location), m_expression(_expression)
+	{}
 private:
 	ptr<Expression> m_expression;
 };
 
-class VariableAssignment : public Statement
+class VariableDefinition : public Statement
 {
+public:
+	VariableDefinition(Location const& _location, ptr<VariableDeclaration> _variable,
+					   ptr<Expression> _value)
+		: Statement(_location), m_variable(_variable), m_value(_value)
+	{}
 private:
 	ptr<VariableDeclaration> m_variable;
-	Token::Value m_assigmentOperator;
+	ptr<Expression> m_value; ///< can be missing
-	ptr<Expression> m_rightHandSide; ///< can be missing
 };
 
 class Expression : public Statement
 {
-private:
+public:
+	Expression(Location const& _location) : Statement(_location) {}
 };
 
 /// @}
@@ -264,6 +284,12 @@ private:
 
 class Assignment : public Expression
 {
+public:
+	Assignment(Location const& _location, ptr<Expression> const& _leftHandSide,
+			   Token::Value _assignmentOperator, ptr<Expression> const& _rightHandSide)
+		: Expression(_location), m_leftHandSide(_leftHandSide),
+		  m_assigmentOperator(_assignmentOperator), m_rightHandSide(_rightHandSide)
+	{}
 private:
 	ptr<Expression> m_leftHandSide;
 	Token::Value m_assigmentOperator;
@@ -272,31 +298,52 @@ private:
 
 class UnaryOperation : public Expression
 {
+public:
+	UnaryOperation(Location const& _location, Token::Value _operator,
+				   ptr<Expression> const& _subExpression, bool _isPrefix)
+		: Expression(_location), m_operator(_operator),
+		  m_subExpression(_subExpression), m_isPrefix(_isPrefix)
+	{}
+
 private:
 	Token::Value m_operator;
 	ptr<Expression> m_subExpression;
-	bool isPrefix;
+	bool m_isPrefix;
 };
 
 class BinaryOperation : public Expression
 {
+public:
+	BinaryOperation(Location const& _location, ptr<Expression> const& _left,
+					Token::Value _operator, ptr<Expression> const& _right)
+		: Expression(_location), m_left(_left), m_operator(_operator), m_right(_right)
+	{}
 private:
 	ptr<Expression> m_left;
-	ptr<Expression> m_right;
 	Token::Value m_operator;
+	ptr<Expression> m_right;
 };
 
 /// Can be ordinary function call, type cast or struct construction.
 class FunctionCall : public Expression
 {
+public:
+	FunctionCall(Location const& _location, ptr<Expression> const& _expression,
+				 vecptr<Expression> const& _arguments)
+		: Expression(_location), m_expression(_expression), m_arguments(_arguments)
+	{}
 private:
-	// if m_functionName is the name of a type, store the token directly
+	ptr<Expression> m_expression;
-	std::string m_functionName; // "in place" calls of return values are not possible for now
 	vecptr<Expression> m_arguments;
 };
 
 class MemberAccess : public Expression
 {
+public:
+	MemberAccess(Location const& _location, ptr<Expression> _expression,
+				 std::string const& _memberName)
+		: Expression(_location), m_expression(_expression), m_memberName(_memberName)
+	{}
 private:
 	ptr<Expression> m_expression;
 	std::string m_memberName;
@@ -304,23 +351,48 @@ private:
 
 class IndexAccess : public Expression
 {
+public:
+	IndexAccess(Location const& _location, ptr<Expression> const& _base,
+				ptr<Expression> const& _index)
+		: Expression(_location), m_base(_base), m_index(_index)
+	{}
+private:
 	ptr<Expression> m_base;
 	ptr<Expression> m_index;
 };
 
 class PrimaryExpression : public Expression
 {
+public:
+	PrimaryExpression(Location const& _location) : Expression(_location) {}
 };
 
 class Identifier : public PrimaryExpression
 {
+public:
+	Identifier(Location const& _location, std::string const& _name)
+		: PrimaryExpression(_location), m_name(_name) {}
 private:
 	std::string m_name;
 };
 
+class ElementaryTypeNameExpression : public PrimaryExpression
+{
+public:
+	ElementaryTypeNameExpression(Location const& _location, Token::Value _type)
+		: PrimaryExpression(_location), m_type(_type) {}
+private:
+	Token::Value m_type;
+};
+
 class Literal : public PrimaryExpression
 {
+public:
+	Literal(Location const& _location, Token::Value _token, std::string const& _value)
+		: PrimaryExpression(_location), m_token(_token), m_value(_value)
+	{}
 private:
+	Token::Value m_token;
 	std::string m_value;
 };
 

Parser.cpp

@@ -231,12 +231,10 @@ ptr<ParameterList> Parser::parseParameterList()
 
 ptr<Block> Parser::parseBlock()
 {
-
 	ASTNodeFactory nodeFactory(*this);
 	expectToken(Token::LBRACE);
 	vecptr<Statement> statements;
 	while (m_scanner->getCurrentToken() != Token::RBRACE) {
-		m_scanner->next();
 		statements.push_back(parseStatement());
 	}
 	nodeFactory.markEndPosition();
@@ -246,6 +244,7 @@ ptr<Block> Parser::parseBlock()
 
 ptr<Statement> Parser::parseStatement()
 {
+	ptr<Statement> statement;
 
 	switch (m_scanner->getCurrentToken()) {
 	case Token::IF:
@@ -254,10 +253,227 @@ ptr<Statement> Parser::parseStatement()
 		return parseWhileStatement();
 	case Token::LBRACE:
 		return parseBlock();
+
 	// starting from here, all statements must be terminated by a semicolon
-	case Token::CONTINUE: // all following
+	case Token::CONTINUE:
-		return
+		statement = ASTNodeFactory(*this).createNode<Continue>();
+		break;
+	case Token::BREAK:
+		statement = ASTNodeFactory(*this).createNode<Break>();
+		break;
+	case Token::RETURN:
+		{
+			ASTNodeFactory nodeFactory(*this);
+			ptr<Expression> expression;
+			if (m_scanner->next() != Token::SEMICOLON) {
+				expression = parseExpression();
+				nodeFactory.setEndPositionFromNode(expression);
+			}
+			statement = nodeFactory.createNode<Return>(expression);
+		}
+		break;
+	default:
+		// distinguish between variable definition (and potentially assignment) and expressions
+		// (which include assignments to other expressions and pre-declared variables)
+		// We have a variable definition if we ge a keyword that specifies a type name, or
+		// in the case of a user-defined type, we have two identifiers following each other.
+		if (m_scanner->getCurrentToken() == Token::MAPPING ||
+				m_scanner->getCurrentToken() == Token::VAR ||
+				Token::IsElementaryTypeName(m_scanner->getCurrentToken()) ||
+				(m_scanner->getCurrentToken() == Token::IDENTIFIER &&
+				 m_scanner->peek() == Token::IDENTIFIER)) {
+			statement = parseVariableDefinition();
+		} else {
+			// "ordinary" expression
+			statement = parseExpression();
+		}
 	}
+	expectToken(Token::SEMICOLON);
+	return statement;
+}
+
+ptr<IfStatement> Parser::parseIfStatement()
+{
+	ASTNodeFactory nodeFactory(*this);
+	expectToken(Token::IF);
+	expectToken(Token::LPAREN);
+	ptr<Expression> condition = parseExpression();
+	expectToken(Token::RPAREN);
+	ptr<Statement> trueBody = parseStatement();
+	ptr<Statement> falseBody;
+	if (m_scanner->getCurrentToken() == Token::ELSE) {
+		m_scanner->next();
+		falseBody = parseStatement();
+		nodeFactory.setEndPositionFromNode(falseBody);
+	} else {
+		nodeFactory.setEndPositionFromNode(trueBody);
+	}
+	return nodeFactory.createNode<IfStatement>(condition, trueBody, falseBody);
+}
+
+ptr<WhileStatement> Parser::parseWhileStatement()
+{
+	ASTNodeFactory nodeFactory(*this);
+	expectToken(Token::WHILE);
+	expectToken(Token::LPAREN);
+	ptr<Expression> condition = parseExpression();
+	expectToken(Token::RPAREN);
+	ptr<Statement> body = parseStatement();
+	nodeFactory.setEndPositionFromNode(body);
+	return nodeFactory.createNode<WhileStatement>(condition, body);
+}
+
+ptr<VariableDefinition> Parser::parseVariableDefinition()
+{
+	ASTNodeFactory nodeFactory(*this);
+	ptr<VariableDeclaration> variable = parseVariableDeclaration();
+	ptr<Expression> value;
+	if (m_scanner->getCurrentToken() == Token::ASSIGN) {
+		m_scanner->next();
+		value = parseExpression();
+		nodeFactory.setEndPositionFromNode(value);
+	} else {
+		nodeFactory.setEndPositionFromNode(variable);
+	}
+	return nodeFactory.createNode<VariableDefinition>(variable, value);
+}
+
+ptr<Expression> Parser::parseExpression()
+{
+	ASTNodeFactory nodeFactory(*this);
+	ptr<Expression> expression = parseBinaryExpression();
+	if (!Token::IsAssignmentOp(m_scanner->getCurrentToken()))
+		return expression;
+
+	Token::Value assignmentOperator = expectAssignmentOperator();
+	ptr<Expression> rightHandSide = parseExpression();
+	nodeFactory.setEndPositionFromNode(rightHandSide);
+	return nodeFactory.createNode<Assignment>(expression, assignmentOperator, rightHandSide);
+}
+
+ptr<Expression> Parser::parseBinaryExpression(int _minPrecedence)
+{
+	ASTNodeFactory nodeFactory(*this);
+	ptr<Expression> expression = parseUnaryExpression();
+	int precedence = Token::Precedence(m_scanner->getCurrentToken());
+	for (; precedence >= _minPrecedence; --precedence) {
+		while (Token::Precedence(m_scanner->getCurrentToken()) == precedence) {
+			Token::Value op = m_scanner->getCurrentToken();
+			m_scanner->next();
+			ptr<Expression> right = parseBinaryExpression(precedence + 1);
+			nodeFactory.setEndPositionFromNode(right);
+			expression = nodeFactory.createNode<BinaryOperation>(expression, op, right);
+		}
+	}
+	return expression;
+}
+
+ptr<Expression> Parser::parseUnaryExpression()
+{
+	ASTNodeFactory nodeFactory(*this);
+	Token::Value token = m_scanner->getCurrentToken();
+	if (Token::IsUnaryOp(token) || Token::IsCountOp(token)) {
+		// prefix expression
+		m_scanner->next();
+		ptr<Expression> subExpression = parseUnaryExpression();
+		nodeFactory.setEndPositionFromNode(subExpression);
+		return nodeFactory.createNode<UnaryOperation>(token, subExpression, true);
+	} else {
+		// potential postfix expression
+		ptr<Expression> subExpression = parseLeftHandSideExpression();
+		token = m_scanner->getCurrentToken();
+		if (!Token::IsCountOp(token))
+			return subExpression;
+		nodeFactory.markEndPosition();
+		m_scanner->next();
+		return nodeFactory.createNode<UnaryOperation>(token, subExpression, false);
+	}
+}
+
+ptr<Expression> Parser::parseLeftHandSideExpression()
+{
+	ASTNodeFactory nodeFactory(*this);
+	ptr<Expression> expression = parsePrimaryExpression();
+
+	while (true) {
+		switch (m_scanner->getCurrentToken()) {
+		case Token::LBRACK:
+			{
+				m_scanner->next();
+				ptr<Expression> index = parseExpression();
+				nodeFactory.markEndPosition();
+				expectToken(Token::RBRACK);
+				expression = nodeFactory.createNode<IndexAccess>(expression, index);
+			}
+			break;
+		case Token::PERIOD:
+			{
+				m_scanner->next();
+				nodeFactory.markEndPosition();
+				std::string memberName = expectIdentifier();
+				expression = nodeFactory.createNode<MemberAccess>(expression, memberName);
+			}
+			break;
+		case Token::LPAREN:
+			{
+				m_scanner->next();
+				vecptr<Expression> arguments = parseFunctionCallArguments();
+				nodeFactory.markEndPosition();
+				expectToken(Token::RPAREN);
+				expression = nodeFactory.createNode<FunctionCall>(expression, arguments);
+			}
+			break;
+		default:
+			return expression;
+		}
+	}
+}
+
+ptr<Expression> Parser::parsePrimaryExpression()
+{
+	Token::Value token = m_scanner->getCurrentToken();
+	switch (token) {
+	case Token::TRUE_LITERAL:
+	case Token::FALSE_LITERAL:
+		m_scanner->next();
+		return ASTNodeFactory(*this).createNode<Literal>(token, std::string());
+	case Token::NUMBER:
+	case Token::STRING_LITERAL:
+		m_scanner->next();
+		return ASTNodeFactory(*this).createNode<Literal>(token, m_scanner->getCurrentLiteral());
+	case Token::IDENTIFIER:
+		m_scanner->next();
+		return ASTNodeFactory(*this).createNode<Identifier>(m_scanner->getCurrentLiteral());
+	case Token::LPAREN:
+		{
+			m_scanner->next();
+			ptr<Expression> expression = parseExpression();
+			expectToken(Token::RPAREN);
+			return expression;
+		}
+	default:
+		if (Token::IsElementaryTypeName(token)) {
+			// used for casts
+			m_scanner->next();
+			return ASTNodeFactory(*this).createNode<ElementaryTypeNameExpression>(token);
+		} else {
+			throwExpectationError("Expected primary expression.");
+			return ptr<Expression>(); // this is not reached
+		}
+	}
+}
+
+vecptr<Expression> Parser::parseFunctionCallArguments()
+{
+	vecptr<Expression> arguments;
+	if (m_scanner->getCurrentToken() != Token::RPAREN) {
+		arguments.push_back(parseExpression());
+		while (m_scanner->getCurrentToken() != Token::RPAREN) {
+			expectToken(Token::COMMA);
+			arguments.push_back(parseExpression());
+		}
+	}
+	return arguments;
 }
 
 void Parser::expectToken(Token::Value _value)
@@ -267,6 +483,15 @@ void Parser::expectToken(Token::Value _value)
 	m_scanner->next();
 }
 
+Token::Value Parser::expectAssignmentOperator()
+{
+	Token::Value op = m_scanner->getCurrentToken();
+	if (!Token::IsAssignmentOp(op))
+		throwExpectationError(std::string("Expected assignment operator"));
+	m_scanner->next();
+	return op;
+}
+
 std::string Parser::expectIdentifier()
 {
 	if (m_scanner->getCurrentToken() != Token::IDENTIFIER)

Parser.h

@@ -53,12 +53,22 @@ private:
 	ptr<ParameterList> parseParameterList();
 	ptr<Block> parseBlock();
 	ptr<Statement> parseStatement();
+	ptr<IfStatement> parseIfStatement();
+	ptr<WhileStatement> parseWhileStatement();
+	ptr<VariableDefinition> parseVariableDefinition();
+	ptr<Expression> parseExpression();
+	ptr<Expression> parseBinaryExpression(int _minPrecedence = 4);
+	ptr<Expression> parseUnaryExpression();
+	ptr<Expression> parseLeftHandSideExpression();
+	ptr<Expression> parsePrimaryExpression();
+	vecptr<Expression> parseFunctionCallArguments();
 	/// @}
 
 	/// Helper functions
 	/// @{
 	/// If current token value is not _value, throw exception otherwise advance token.
 	void expectToken(Token::Value _value);
+	Token::Value expectAssignmentOperator();
 	std::string expectIdentifier();
 	void throwExpectationError(const std::string& _description);
 	/// @}

grammar.txt

@@ -15,18 +15,21 @@ TypeName = ElementaryTypeName | Identifier | Mapping
 Mapping = 'mapping' '(' ElementaryTypeName '=>' TypeName ')'
 
 Block = '{' Statement* '}'
-Statement = IfStatement | WhileStatement | Continue | Break | Return | VariableAssignment | Expression ';' | Block
+Statement = IfStatement | WhileStatement | Block |
+			( Continue | Break | Return | VariableDefinition | Expression ) ';'
 
 IfStatement = 'if' '(' Expression ')' Statement ( 'else' Statement )?
 WhileStatement = 'while' '(' Expression ')' Statement
 Continue = 'continue' ';'
 Break = 'break' ';'
 Return = 'return' Expression? ';'
-VariableAssignment = VariableDeclaration ( AssignmentOp Expression )? ';'
+VariableDefinition = VariableDeclaration ( = Expression )? ';'
 
-Expression = Assignment | UnaryOperation | BinaryOperation | FunctionCall | IndexAccess | MemberAccess | PrimaryExpression
+Expression = Assignment | UnaryOperation | BinaryOperation | FunctionCall | IndexAccess |
+			 MemberAccess | PrimaryExpression
+// The expression syntax is actually much more complicated
 Assignment = Expression (AssignmentOp Expression)
-FunctionCall = Identifier '(' ( Expression ( ',' Expression )* ) ')'
+FunctionCall = Expression '(' ( Expression ( ',' Expression )* ) ')'
 MemberAccess = Expression '.' Identifier
 IndexAccess = Expression '[' Expresison ']'
-PrimaryExpression = Identifier | NumberLiteral | StringLiteral | '(' Expression ')'
+PrimaryExpression = Identifier | NumberLiteral | StringLiteral | ElementaryTypeName | '(' Expression ')'