/* This file is part of cpp-ethereum. cpp-ethereum is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. cpp-ethereum is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with cpp-ethereum. If not, see . */ /** * @author Christian * @date 2014 * Solidity parser. */ #include #include #include #include #include #include using namespace std; namespace dev { namespace solidity { /// AST node factory that also tracks the begin and end position of an AST node /// while it is being parsed class Parser::ASTNodeFactory { public: ASTNodeFactory(Parser const& _parser): m_parser(_parser), m_location(_parser.getPosition(), -1, _parser.getSourceName()) {} void markEndPosition() { m_location.end = m_parser.getEndPosition(); } void setLocationEmpty() { m_location.end = m_location.start; } /// Set the end position to the one of the given node. void setEndPositionFromNode(ASTPointer const& _node) { m_location.end = _node->getLocation().end; } template ASTPointer createNode(Args&& ... _args) { if (m_location.end < 0) markEndPosition(); return make_shared(m_location, forward(_args)...); } private: Parser const& m_parser; Location m_location; }; ASTPointer Parser::parse(shared_ptr const& _scanner) { m_scanner = _scanner; ASTNodeFactory nodeFactory(*this); vector> nodes; while (_scanner->getCurrentToken() != Token::EOS) { switch (m_scanner->getCurrentToken()) { case Token::IMPORT: nodes.push_back(parseImportDirective()); break; case Token::CONTRACT: nodes.push_back(parseContractDefinition()); break; default: BOOST_THROW_EXCEPTION(createParserError(std::string("Expected import directive or contract definition."))); } } return nodeFactory.createNode(nodes); } std::shared_ptr const& Parser::getSourceName() const { return m_scanner->getSourceName(); } int Parser::getPosition() const { return m_scanner->getCurrentLocation().start; } int Parser::getEndPosition() const { return m_scanner->getCurrentLocation().end; } ASTPointer Parser::parseImportDirective() { ASTNodeFactory nodeFactory(*this); expectToken(Token::IMPORT); if (m_scanner->getCurrentToken() != Token::STRING_LITERAL) BOOST_THROW_EXCEPTION(createParserError("Expected string literal (URL).")); ASTPointer url = getLiteralAndAdvance(); nodeFactory.markEndPosition(); expectToken(Token::SEMICOLON); return nodeFactory.createNode(url); } ASTPointer Parser::parseContractDefinition() { ASTNodeFactory nodeFactory(*this); ASTPointer docString; if (m_scanner->getCurrentCommentLiteral() != "") docString = make_shared(m_scanner->getCurrentCommentLiteral()); expectToken(Token::CONTRACT); ASTPointer name = expectIdentifierToken(); vector> baseContracts; vector> structs; vector> stateVariables; vector> functions; vector> modifiers; if (m_scanner->getCurrentToken() == Token::IS) do { m_scanner->next(); baseContracts.push_back(parseInheritanceSpecifier()); } while (m_scanner->getCurrentToken() == Token::COMMA); expectToken(Token::LBRACE); bool visibilityIsPublic = true; while (true) { Token::Value currentToken = m_scanner->getCurrentToken(); if (currentToken == Token::RBRACE) break; else if (currentToken == Token::PUBLIC || currentToken == Token::PRIVATE) { visibilityIsPublic = (m_scanner->getCurrentToken() == Token::PUBLIC); m_scanner->next(); expectToken(Token::COLON); } else if (currentToken == Token::FUNCTION) functions.push_back(parseFunctionDefinition(visibilityIsPublic, name.get())); else if (currentToken == Token::STRUCT) structs.push_back(parseStructDefinition()); else if (currentToken == Token::IDENTIFIER || currentToken == Token::MAPPING || Token::isElementaryTypeName(currentToken)) { bool const allowVar = false; stateVariables.push_back(parseVariableDeclaration(allowVar, visibilityIsPublic, true)); expectToken(Token::SEMICOLON); } else if (currentToken == Token::MODIFIER) modifiers.push_back(parseModifierDefinition()); else BOOST_THROW_EXCEPTION(createParserError("Function, variable, struct or modifier declaration expected.")); } nodeFactory.markEndPosition(); expectToken(Token::RBRACE); return nodeFactory.createNode(name, docString, baseContracts, structs, stateVariables, functions, modifiers); } ASTPointer Parser::parseInheritanceSpecifier() { ASTNodeFactory nodeFactory(*this); ASTPointer name(parseIdentifier()); vector> arguments; if (m_scanner->getCurrentToken() == Token::LPAREN) { m_scanner->next(); arguments = parseFunctionCallArguments(); nodeFactory.markEndPosition(); expectToken(Token::RPAREN); } else nodeFactory.setEndPositionFromNode(name); return nodeFactory.createNode(name, arguments); } ASTPointer Parser::parseFunctionDefinition(bool _isPublic, ASTString const* _contractName) { ASTNodeFactory nodeFactory(*this); ASTPointer docstring; if (m_scanner->getCurrentCommentLiteral() != "") docstring = make_shared(m_scanner->getCurrentCommentLiteral()); expectToken(Token::FUNCTION); ASTPointer name(expectIdentifierToken()); ASTPointer parameters(parseParameterList()); bool isDeclaredConst = false; vector> modifiers; while (true) { if (m_scanner->getCurrentToken() == Token::CONST) { isDeclaredConst = true; m_scanner->next(); } else if (m_scanner->getCurrentToken() == Token::IDENTIFIER) modifiers.push_back(parseModifierInvocation()); else break; } ASTPointer returnParameters; if (m_scanner->getCurrentToken() == Token::RETURNS) { bool const permitEmptyParameterList = false; m_scanner->next(); returnParameters = parseParameterList(permitEmptyParameterList); } else { // create an empty parameter list at a zero-length location ASTNodeFactory nodeFactory(*this); nodeFactory.setLocationEmpty(); returnParameters = nodeFactory.createNode(vector>()); } ASTPointer block = parseBlock(); nodeFactory.setEndPositionFromNode(block); bool const c_isConstructor = (_contractName && *name == *_contractName); return nodeFactory.createNode(name, _isPublic, c_isConstructor, docstring, parameters, isDeclaredConst, modifiers, returnParameters, block); } ASTPointer Parser::parseStructDefinition() { ASTNodeFactory nodeFactory(*this); expectToken(Token::STRUCT); ASTPointer name = expectIdentifierToken(); vector> members; expectToken(Token::LBRACE); while (m_scanner->getCurrentToken() != Token::RBRACE) { bool const allowVar = false; members.push_back(parseVariableDeclaration(allowVar)); expectToken(Token::SEMICOLON); } nodeFactory.markEndPosition(); expectToken(Token::RBRACE); return nodeFactory.createNode(name, members); } ASTPointer Parser::parseVariableDeclaration(bool _allowVar, bool _isPublic, bool _isStateVariable) { ASTNodeFactory nodeFactory(*this); ASTPointer type = parseTypeName(_allowVar); nodeFactory.markEndPosition(); return nodeFactory.createNode(type, expectIdentifierToken(), _isPublic, _isStateVariable); } ASTPointer Parser::parseModifierDefinition() { ScopeGuard resetModifierFlag([this]() { m_insideModifier = false; }); m_insideModifier = true; ASTNodeFactory nodeFactory(*this); ASTPointer docstring; if (m_scanner->getCurrentCommentLiteral() != "") docstring = make_shared(m_scanner->getCurrentCommentLiteral()); expectToken(Token::MODIFIER); ASTPointer name(expectIdentifierToken()); ASTPointer parameters; if (m_scanner->getCurrentToken() == Token::LPAREN) parameters = parseParameterList(); else { // create an empty parameter list at a zero-length location ASTNodeFactory nodeFactory(*this); nodeFactory.setLocationEmpty(); parameters = nodeFactory.createNode(vector>()); } ASTPointer block = parseBlock(); nodeFactory.setEndPositionFromNode(block); return nodeFactory.createNode(name, docstring, parameters, block); } ASTPointer Parser::parseModifierInvocation() { ASTNodeFactory nodeFactory(*this); ASTPointer name(parseIdentifier()); vector> arguments; if (m_scanner->getCurrentToken() == Token::LPAREN) { m_scanner->next(); arguments = parseFunctionCallArguments(); nodeFactory.markEndPosition(); expectToken(Token::RPAREN); } else nodeFactory.setEndPositionFromNode(name); return nodeFactory.createNode(name, arguments); } ASTPointer Parser::parseIdentifier() { ASTNodeFactory nodeFactory(*this); nodeFactory.markEndPosition(); return nodeFactory.createNode(expectIdentifierToken()); } ASTPointer Parser::parseTypeName(bool _allowVar) { ASTPointer type; Token::Value token = m_scanner->getCurrentToken(); if (Token::isElementaryTypeName(token)) { type = ASTNodeFactory(*this).createNode(token); m_scanner->next(); } else if (token == Token::VAR) { if (!_allowVar) BOOST_THROW_EXCEPTION(createParserError("Expected explicit type name.")); m_scanner->next(); } else if (token == Token::MAPPING) { type = parseMapping(); } else if (token == Token::IDENTIFIER) { ASTNodeFactory nodeFactory(*this); nodeFactory.markEndPosition(); type = nodeFactory.createNode(expectIdentifierToken()); } else BOOST_THROW_EXCEPTION(createParserError("Expected type name")); return type; } ASTPointer Parser::parseMapping() { ASTNodeFactory nodeFactory(*this); expectToken(Token::MAPPING); expectToken(Token::LPAREN); if (!Token::isElementaryTypeName(m_scanner->getCurrentToken())) BOOST_THROW_EXCEPTION(createParserError("Expected elementary type name for mapping key type")); ASTPointer keyType; keyType = ASTNodeFactory(*this).createNode(m_scanner->getCurrentToken()); m_scanner->next(); expectToken(Token::ARROW); bool const allowVar = false; ASTPointer valueType = parseTypeName(allowVar); nodeFactory.markEndPosition(); expectToken(Token::RPAREN); return nodeFactory.createNode(keyType, valueType); } ASTPointer Parser::parseParameterList(bool _allowEmpty) { ASTNodeFactory nodeFactory(*this); vector> parameters; expectToken(Token::LPAREN); if (!_allowEmpty || m_scanner->getCurrentToken() != Token::RPAREN) { bool const allowVar = false; parameters.push_back(parseVariableDeclaration(allowVar)); while (m_scanner->getCurrentToken() != Token::RPAREN) { expectToken(Token::COMMA); parameters.push_back(parseVariableDeclaration(allowVar)); } } nodeFactory.markEndPosition(); m_scanner->next(); return nodeFactory.createNode(parameters); } ASTPointer Parser::parseBlock() { ASTNodeFactory nodeFactory(*this); expectToken(Token::LBRACE); vector> statements; while (m_scanner->getCurrentToken() != Token::RBRACE) statements.push_back(parseStatement()); nodeFactory.markEndPosition(); expectToken(Token::RBRACE); return nodeFactory.createNode(statements); } ASTPointer Parser::parseStatement() { ASTPointer statement; switch (m_scanner->getCurrentToken()) { case Token::IF: return parseIfStatement(); case Token::WHILE: return parseWhileStatement(); case Token::FOR: return parseForStatement(); case Token::LBRACE: return parseBlock(); // starting from here, all statements must be terminated by a semicolon case Token::CONTINUE: statement = ASTNodeFactory(*this).createNode(); m_scanner->next(); break; case Token::BREAK: statement = ASTNodeFactory(*this).createNode(); m_scanner->next(); break; case Token::RETURN: { ASTNodeFactory nodeFactory(*this); ASTPointer expression; if (m_scanner->next() != Token::SEMICOLON) { expression = parseExpression(); nodeFactory.setEndPositionFromNode(expression); } statement = nodeFactory.createNode(expression); break; } case Token::IDENTIFIER: if (m_insideModifier && m_scanner->getCurrentLiteral() == "_") { statement = ASTNodeFactory(*this).createNode(); m_scanner->next(); return statement; } // fall-through default: statement = parseVarDefOrExprStmt(); } expectToken(Token::SEMICOLON); return statement; } ASTPointer Parser::parseIfStatement() { ASTNodeFactory nodeFactory(*this); expectToken(Token::IF); expectToken(Token::LPAREN); ASTPointer condition = parseExpression(); expectToken(Token::RPAREN); ASTPointer trueBody = parseStatement(); ASTPointer falseBody; if (m_scanner->getCurrentToken() == Token::ELSE) { m_scanner->next(); falseBody = parseStatement(); nodeFactory.setEndPositionFromNode(falseBody); } else nodeFactory.setEndPositionFromNode(trueBody); return nodeFactory.createNode(condition, trueBody, falseBody); } ASTPointer Parser::parseWhileStatement() { ASTNodeFactory nodeFactory(*this); expectToken(Token::WHILE); expectToken(Token::LPAREN); ASTPointer condition = parseExpression(); expectToken(Token::RPAREN); ASTPointer body = parseStatement(); nodeFactory.setEndPositionFromNode(body); return nodeFactory.createNode(condition, body); } ASTPointer Parser::parseForStatement() { ASTNodeFactory nodeFactory(*this); ASTPointer initExpression; ASTPointer conditionExpression; ASTPointer loopExpression; expectToken(Token::FOR); expectToken(Token::LPAREN); // LTODO: Maybe here have some predicate like peekExpression() instead of checking for semicolon and RPAREN? if (m_scanner->getCurrentToken() != Token::SEMICOLON) initExpression = parseVarDefOrExprStmt(); expectToken(Token::SEMICOLON); if (m_scanner->getCurrentToken() != Token::SEMICOLON) conditionExpression = parseExpression(); expectToken(Token::SEMICOLON); if (m_scanner->getCurrentToken() != Token::RPAREN) loopExpression = parseExpressionStatement(); expectToken(Token::RPAREN); ASTPointer body = parseStatement(); nodeFactory.setEndPositionFromNode(body); return nodeFactory.createNode(initExpression, conditionExpression, loopExpression, body); } ASTPointer Parser::parseVarDefOrExprStmt() { if (peekVariableDefinition()) return parseVariableDefinition(); else return parseExpressionStatement(); } ASTPointer Parser::parseVariableDefinition() { ASTNodeFactory nodeFactory(*this); bool const allowVar = true; ASTPointer variable = parseVariableDeclaration(allowVar); ASTPointer value; if (m_scanner->getCurrentToken() == Token::ASSIGN) { m_scanner->next(); value = parseExpression(); nodeFactory.setEndPositionFromNode(value); } else nodeFactory.setEndPositionFromNode(variable); return nodeFactory.createNode(variable, value); } ASTPointer Parser::parseExpressionStatement() { ASTNodeFactory nodeFactory(*this); ASTPointer expression = parseExpression(); nodeFactory.setEndPositionFromNode(expression); return nodeFactory.createNode(expression); } ASTPointer Parser::parseExpression() { ASTNodeFactory nodeFactory(*this); ASTPointer expression = parseBinaryExpression(); if (!Token::isAssignmentOp(m_scanner->getCurrentToken())) return expression; Token::Value assignmentOperator = expectAssignmentOperator(); ASTPointer rightHandSide = parseExpression(); nodeFactory.setEndPositionFromNode(rightHandSide); return nodeFactory.createNode(expression, assignmentOperator, rightHandSide); } ASTPointer Parser::parseBinaryExpression(int _minPrecedence) { ASTNodeFactory nodeFactory(*this); ASTPointer 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(); ASTPointer right = parseBinaryExpression(precedence + 1); nodeFactory.setEndPositionFromNode(right); expression = nodeFactory.createNode(expression, op, right); } } return expression; } ASTPointer Parser::parseUnaryExpression() { ASTNodeFactory nodeFactory(*this); Token::Value token = m_scanner->getCurrentToken(); if (Token::isUnaryOp(token) || Token::isCountOp(token)) { // prefix expression m_scanner->next(); ASTPointer subExpression = parseUnaryExpression(); nodeFactory.setEndPositionFromNode(subExpression); return nodeFactory.createNode(token, subExpression, true); } else { // potential postfix expression ASTPointer subExpression = parseLeftHandSideExpression(); token = m_scanner->getCurrentToken(); if (!Token::isCountOp(token)) return subExpression; nodeFactory.markEndPosition(); m_scanner->next(); return nodeFactory.createNode(token, subExpression, false); } } ASTPointer Parser::parseLeftHandSideExpression() { ASTNodeFactory nodeFactory(*this); ASTPointer expression; if (m_scanner->getCurrentToken() == Token::NEW) { expectToken(Token::NEW); ASTPointer contractName(parseIdentifier()); nodeFactory.setEndPositionFromNode(contractName); expression = nodeFactory.createNode(contractName); } else expression = parsePrimaryExpression(); while (true) { switch (m_scanner->getCurrentToken()) { case Token::LBRACK: { m_scanner->next(); ASTPointer index = parseExpression(); nodeFactory.markEndPosition(); expectToken(Token::RBRACK); expression = nodeFactory.createNode(expression, index); } break; case Token::PERIOD: { m_scanner->next(); nodeFactory.markEndPosition(); expression = nodeFactory.createNode(expression, expectIdentifierToken()); } break; case Token::LPAREN: { m_scanner->next(); vector> arguments = parseFunctionCallArguments(); nodeFactory.markEndPosition(); expectToken(Token::RPAREN); expression = nodeFactory.createNode(expression, arguments); } break; default: return expression; } } } ASTPointer Parser::parsePrimaryExpression() { ASTNodeFactory nodeFactory(*this); Token::Value token = m_scanner->getCurrentToken(); ASTPointer expression; switch (token) { case Token::TRUE_LITERAL: case Token::FALSE_LITERAL: expression = nodeFactory.createNode(token, getLiteralAndAdvance()); break; case Token::NUMBER: case Token::STRING_LITERAL: nodeFactory.markEndPosition(); expression = nodeFactory.createNode(token, getLiteralAndAdvance()); break; case Token::IDENTIFIER: nodeFactory.markEndPosition(); expression = nodeFactory.createNode(getLiteralAndAdvance()); break; case Token::LPAREN: { m_scanner->next(); ASTPointer expression = parseExpression(); expectToken(Token::RPAREN); return expression; } default: if (Token::isElementaryTypeName(token)) { // used for casts expression = nodeFactory.createNode(token); m_scanner->next(); } else { BOOST_THROW_EXCEPTION(createParserError("Expected primary expression.")); return ASTPointer(); // this is not reached } break; } return expression; } vector> Parser::parseFunctionCallArguments() { vector> 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; } bool Parser::peekVariableDefinition() { // distinguish between variable definition (and potentially assignment) and expression statement // (which include assignments to other expressions and pre-declared variables) // We have a variable definition if we get a keyword that specifies a type name, or // in the case of a user-defined type, we have two identifiers following each other. return (m_scanner->getCurrentToken() == Token::MAPPING || m_scanner->getCurrentToken() == Token::VAR || ((Token::isElementaryTypeName(m_scanner->getCurrentToken()) || m_scanner->getCurrentToken() == Token::IDENTIFIER) && m_scanner->peekNextToken() == Token::IDENTIFIER)); } void Parser::expectToken(Token::Value _value) { if (m_scanner->getCurrentToken() != _value) BOOST_THROW_EXCEPTION(createParserError(string("Expected token ") + string(Token::getName(_value)))); m_scanner->next(); } Token::Value Parser::expectAssignmentOperator() { Token::Value op = m_scanner->getCurrentToken(); if (!Token::isAssignmentOp(op)) BOOST_THROW_EXCEPTION(createParserError("Expected assignment operator")); m_scanner->next(); return op; } ASTPointer Parser::expectIdentifierToken() { if (m_scanner->getCurrentToken() != Token::IDENTIFIER) BOOST_THROW_EXCEPTION(createParserError("Expected identifier")); return getLiteralAndAdvance(); } ASTPointer Parser::getLiteralAndAdvance() { ASTPointer identifier = make_shared(m_scanner->getCurrentLiteral()); m_scanner->next(); return identifier; } ParserError Parser::createParserError(string const& _description) const { return ParserError() << errinfo_sourceLocation(Location(getPosition(), getPosition(), getSourceName())) << errinfo_comment(_description); } } }