/* 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::StringLiteral) 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> enums; vector> stateVariables; vector> functions; vector> modifiers; vector> events; if (m_scanner->getCurrentToken() == Token::Is) do { m_scanner->next(); baseContracts.push_back(parseInheritanceSpecifier()); } while (m_scanner->getCurrentToken() == Token::Comma); expectToken(Token::LBrace); while (true) { Token::Value currentToken = m_scanner->getCurrentToken(); if (currentToken == Token::RBrace) break; else if (currentToken == Token::Function) functions.push_back(parseFunctionDefinition(name.get())); else if (currentToken == Token::Struct) structs.push_back(parseStructDefinition()); else if (currentToken == Token::Enum) enums.push_back(parseEnumDefinition()); else if (currentToken == Token::Identifier || currentToken == Token::Mapping || Token::isElementaryTypeName(currentToken)) { VarDeclParserOptions options; options.isStateVariable = true; stateVariables.push_back(parseVariableDeclaration(options)); expectToken(Token::Semicolon); } else if (currentToken == Token::Modifier) modifiers.push_back(parseModifierDefinition()); else if (currentToken == Token::Event) events.push_back(parseEventDefinition()); else BOOST_THROW_EXCEPTION(createParserError("Function, variable, struct or modifier declaration expected.")); } nodeFactory.markEndPosition(); expectToken(Token::RBrace); return nodeFactory.createNode(name, docString, baseContracts, structs, enums, stateVariables, functions, modifiers, events); } ASTPointer Parser::parseInheritanceSpecifier() { ASTNodeFactory nodeFactory(*this); ASTPointer name(parseIdentifier()); vector> arguments; if (m_scanner->getCurrentToken() == Token::LParen) { m_scanner->next(); arguments = parseFunctionCallListArguments(); nodeFactory.markEndPosition(); expectToken(Token::RParen); } else nodeFactory.setEndPositionFromNode(name); return nodeFactory.createNode(name, arguments); } Declaration::Visibility Parser::parseVisibilitySpecifier(Token::Value _token) { Declaration::Visibility visibility(Declaration::Visibility::Default); if (_token == Token::Public) visibility = Declaration::Visibility::Public; else if (_token == Token::Protected) visibility = Declaration::Visibility::Protected; else if (_token == Token::Private) visibility = Declaration::Visibility::Private; else if (_token == Token::External) visibility = Declaration::Visibility::External; else solAssert(false, "Invalid visibility specifier."); m_scanner->next(); return visibility; } ASTPointer Parser::parseFunctionDefinition(ASTString const* _contractName) { ASTNodeFactory nodeFactory(*this); ASTPointer docstring; if (m_scanner->getCurrentCommentLiteral() != "") docstring = make_shared(m_scanner->getCurrentCommentLiteral()); expectToken(Token::Function); ASTPointer name; if (m_scanner->getCurrentToken() == Token::LParen) name = make_shared(); // anonymous function else name = expectIdentifierToken(); ASTPointer parameters(parseParameterList()); bool isDeclaredConst = false; Declaration::Visibility visibility(Declaration::Visibility::Default); vector> modifiers; while (true) { Token::Value token = m_scanner->getCurrentToken(); if (token == Token::Const) { isDeclaredConst = true; m_scanner->next(); } else if (token == Token::Identifier) modifiers.push_back(parseModifierInvocation()); else if (Token::isVisibilitySpecifier(token)) { if (visibility != Declaration::Visibility::Default) BOOST_THROW_EXCEPTION(createParserError("Multiple visibility specifiers.")); visibility = parseVisibilitySpecifier(token); } else break; } ASTPointer returnParameters; if (m_scanner->getCurrentToken() == Token::Returns) { bool const permitEmptyParameterList = false; m_scanner->next(); returnParameters = parseParameterList(permitEmptyParameterList); } else returnParameters = createEmptyParameterList(); ASTPointer block = parseBlock(); nodeFactory.setEndPositionFromNode(block); bool const c_isConstructor = (_contractName && *name == *_contractName); return nodeFactory.createNode(name, visibility, 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) { members.push_back(parseVariableDeclaration()); expectToken(Token::Semicolon); } nodeFactory.markEndPosition(); expectToken(Token::RBrace); return nodeFactory.createNode(name, members); } ASTPointer Parser::parseEnumValue() { ASTNodeFactory nodeFactory(*this); nodeFactory.markEndPosition(); return nodeFactory.createNode(expectIdentifierToken()); } ASTPointer Parser::parseEnumDefinition() { ASTNodeFactory nodeFactory(*this); expectToken(Token::Enum); ASTPointer name = expectIdentifierToken(); vector> members; expectToken(Token::LBrace); while (m_scanner->getCurrentToken() != Token::RBrace) { members.push_back(parseEnumValue()); if (m_scanner->getCurrentToken() == Token::RBrace) break; expectToken(Token::Comma); if (m_scanner->getCurrentToken() != Token::Identifier) BOOST_THROW_EXCEPTION(createParserError("Expected Identifier after ','")); } nodeFactory.markEndPosition(); expectToken(Token::RBrace); return nodeFactory.createNode(name, members); } ASTPointer Parser::parseVariableDeclaration(VarDeclParserOptions const& _options) { ASTNodeFactory nodeFactory(*this); ASTPointer type = parseTypeName(_options.allowVar); if (type != nullptr) nodeFactory.setEndPositionFromNode(type); bool isIndexed = false; ASTPointer identifier; Token::Value token = m_scanner->getCurrentToken(); Declaration::Visibility visibility(Declaration::Visibility::Default); if (_options.isStateVariable && Token::isVariableVisibilitySpecifier(token)) visibility = parseVisibilitySpecifier(token); if (_options.allowIndexed && token == Token::Indexed) { isIndexed = true; m_scanner->next(); } nodeFactory.markEndPosition(); if (_options.allowEmptyName && m_scanner->getCurrentToken() != Token::Identifier) { identifier = make_shared(""); solAssert(type != nullptr, ""); nodeFactory.setEndPositionFromNode(type); } else identifier = expectIdentifierToken(); return nodeFactory.createNode(type, identifier, visibility, _options.isStateVariable, isIndexed); } 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 parameters = createEmptyParameterList(); ASTPointer block = parseBlock(); nodeFactory.setEndPositionFromNode(block); return nodeFactory.createNode(name, docstring, parameters, block); } ASTPointer Parser::parseEventDefinition() { ASTNodeFactory nodeFactory(*this); ASTPointer docstring; if (m_scanner->getCurrentCommentLiteral() != "") docstring = make_shared(m_scanner->getCurrentCommentLiteral()); expectToken(Token::Event); ASTPointer name(expectIdentifierToken()); ASTPointer parameters; if (m_scanner->getCurrentToken() == Token::LParen) parameters = parseParameterList(true, true); else parameters = createEmptyParameterList(); nodeFactory.markEndPosition(); expectToken(Token::Semicolon); return nodeFactory.createNode(name, docstring, parameters); } ASTPointer Parser::parseModifierInvocation() { ASTNodeFactory nodeFactory(*this); ASTPointer name(parseIdentifier()); vector> arguments; if (m_scanner->getCurrentToken() == Token::LParen) { m_scanner->next(); arguments = parseFunctionCallListArguments(); 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, bool _allowIndexed) { ASTNodeFactory nodeFactory(*this); vector> parameters; VarDeclParserOptions options; options.allowIndexed = _allowIndexed; options.allowEmptyName = true; expectToken(Token::LParen); if (!_allowEmpty || m_scanner->getCurrentToken() != Token::RParen) { parameters.push_back(parseVariableDeclaration(options)); while (m_scanner->getCurrentToken() != Token::RParen) { expectToken(Token::Comma); parameters.push_back(parseVariableDeclaration(options)); } } 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); VarDeclParserOptions options; options.allowVar = true; ASTPointer variable = parseVariableDeclaration(options); 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; vector> names; std::tie(arguments, names) = parseFunctionCallArguments(); nodeFactory.markEndPosition(); expectToken(Token::RParen); expression = nodeFactory.createNode(expression, arguments, names); } break; default: return expression; } } } ASTPointer Parser::parsePrimaryExpression() { ASTNodeFactory nodeFactory(*this); Token::Value token = m_scanner->getCurrentToken(); ASTPointer expression; switch (token) { case Token::TrueLiteral: case Token::FalseLiteral: expression = nodeFactory.createNode(token, getLiteralAndAdvance()); break; case Token::Number: if (Token::isEtherSubdenomination(m_scanner->peekNextToken())) { ASTPointer literal = getLiteralAndAdvance(); nodeFactory.markEndPosition(); Literal::SubDenomination subdenomination = static_cast(m_scanner->getCurrentToken()); m_scanner->next(); expression = nodeFactory.createNode(token, literal, subdenomination); break; } // fall-through case Token::StringLiteral: 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::parseFunctionCallListArguments() { 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; } pair>, vector>> Parser::parseFunctionCallArguments() { pair>, vector>> ret; Token::Value token = m_scanner->getCurrentToken(); if (token == Token::LBrace) { // call({arg1 : 1, arg2 : 2 }) expectToken(Token::LBrace); while (m_scanner->getCurrentToken() != Token::RBrace) { ret.second.push_back(expectIdentifierToken()); expectToken(Token::Colon); ret.first.push_back(parseExpression()); if (m_scanner->getCurrentToken() == Token::Comma) expectToken(Token::Comma); else break; } expectToken(Token::RBrace); } else ret.first = parseFunctionCallListArguments(); return ret; } 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; } ASTPointer Parser::createEmptyParameterList() { ASTNodeFactory nodeFactory(*this); nodeFactory.setLocationEmpty(); return nodeFactory.createNode(vector>()); } ParserError Parser::createParserError(string const& _description) const { return ParserError() << errinfo_sourceLocation(Location(getPosition(), getPosition(), getSourceName())) << errinfo_comment(_description); } } }