Merge pull request #584 from chriseth/sol_strings

String types.

Compiler.cpp

@@ -130,21 +130,17 @@ unsigned Compiler::appendCalldataUnpacker(FunctionDefinition const& _function, b
 {
 	// We do not check the calldata size, everything is zero-padded.
 	unsigned dataOffset = 1;
-	eth::Instruction load = _fromMemory ? eth::Instruction::MLOAD : eth::Instruction::CALLDATALOAD;
 
 	//@todo this can be done more efficiently, saving some CALLDATALOAD calls
 	for (ASTPointer<VariableDeclaration> const& var: _function.getParameters())
 	{
 		unsigned const numBytes = var->getType()->getCalldataEncodedSize();
-		if (numBytes == 0 || numBytes > 32)
+		if (numBytes > 32)
 			BOOST_THROW_EXCEPTION(CompilerError()
 								  << errinfo_sourceLocation(var->getLocation())
 								  << errinfo_comment("Type " + var->getType()->toString() + " not yet supported."));
-		if (numBytes == 32)
+		bool leftAligned = var->getType()->getCategory() == Type::Category::STRING;
-			m_context << u256(dataOffset) << load;
+		CompilerUtils(m_context).loadFromMemory(dataOffset, numBytes, leftAligned, !_fromMemory);
-		else
-			m_context << (u256(1) << ((32 - numBytes) * 8)) << u256(dataOffset)
-					  << load << eth::Instruction::DIV;
 		dataOffset += numBytes;
 	}
 	return dataOffset;
@@ -160,14 +156,13 @@ void Compiler::appendReturnValuePacker(FunctionDefinition const& _function)
 	{
 		Type const& paramType = *parameters[i]->getType();
 		unsigned numBytes = paramType.getCalldataEncodedSize();
-		if (numBytes == 0 || numBytes > 32)
+		if (numBytes > 32)
 			BOOST_THROW_EXCEPTION(CompilerError()
 								  << errinfo_sourceLocation(parameters[i]->getLocation())
 								  << errinfo_comment("Type " + paramType.toString() + " not yet supported."));
 		CompilerUtils(m_context).copyToStackTop(stackDepth, paramType);
-		if (numBytes != 32)
+		bool const leftAligned = paramType.getCategory() == Type::Category::STRING;
-			m_context << (u256(1) << ((32 - numBytes) * 8)) << eth::Instruction::MUL;
+		CompilerUtils(m_context).storeInMemory(dataOffset, numBytes, leftAligned);
-		m_context << u256(dataOffset) << eth::Instruction::MSTORE;
 		stackDepth -= paramType.getSizeOnStack();
 		dataOffset += numBytes;
 	}

CompilerUtils.cpp

@@ -31,6 +31,46 @@ namespace dev
 namespace solidity
 {
 
+void CompilerUtils::loadFromMemory(unsigned _offset, unsigned _bytes, bool _leftAligned, bool _fromCalldata)
+{
+	if (_bytes == 0)
+	{
+		m_context << u256(0);
+		return;
+	}
+	eth::Instruction load = _fromCalldata ? eth::Instruction::CALLDATALOAD : eth::Instruction::MLOAD;
+	if (asserts(_bytes <= 32))
+		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Memory load of more than 32 bytes requested."));
+	if (_bytes == 32)
+		m_context << u256(_offset) << load;
+	else
+	{
+		// load data and add leading or trailing zeros by dividing/multiplying depending on alignment
+		u256 shiftFactor = u256(1) << ((32 - _bytes) * 8);
+		m_context << shiftFactor;
+		if (_leftAligned)
+			m_context << eth::Instruction::DUP1;
+		m_context << u256(_offset) << load << eth::Instruction::DIV;
+		if (_leftAligned)
+			m_context << eth::Instruction::MUL;
+	}
+}
+
+void CompilerUtils::storeInMemory(unsigned _offset, unsigned _bytes, bool _leftAligned)
+{
+	if (_bytes == 0)
+	{
+		m_context << eth::Instruction::POP;
+		return;
+	}
+	if (asserts(_bytes <= 32))
+		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Memory store of more than 32 bytes requested."));
+	if (_bytes != 32 && !_leftAligned)
+		// shift the value accordingly before storing
+		m_context << (u256(1) << ((32 - _bytes) * 8)) << eth::Instruction::MUL;
+	m_context << u256(_offset) << eth::Instruction::MSTORE;
+}
+
 void CompilerUtils::moveToStackVariable(VariableDeclaration const& _variable)
 {
 	unsigned const stackPosition = m_context.baseToCurrentStackOffset(m_context.getBaseStackOffsetOfVariable(_variable));

CompilerUtils.h

@@ -35,6 +35,18 @@ class CompilerUtils
 public:
 	CompilerUtils(CompilerContext& _context): m_context(_context) {}
 
+	/// Loads data from memory to the stack.
+	/// @param _offset offset in memory (or calldata)
+	/// @param _bytes number of bytes to load
+	/// @param _leftAligned if true, store left aligned on stack (otherwise right aligned)
+	/// @param _fromCalldata if true, load from calldata, not from memory
+	void loadFromMemory(unsigned _offset, unsigned _bytes = 32, bool _leftAligned = false, bool _fromCalldata = false);
+	/// Stores data from stack in memory.
+	/// @param _offset offset in memory
+	/// @param _bytes number of bytes to store
+	/// @param _leftAligned if true, data is left aligned on stack (otherwise right aligned)
+	void storeInMemory(unsigned _offset, unsigned _bytes = 32, bool _leftAligned = false);
+
 	/// Moves the value that is at the top of the stack to a stack variable.
 	void moveToStackVariable(VariableDeclaration const& _variable);
 	/// Copies a variable of type @a _type from a stack depth of @a _stackDepth to the top of the stack.

ExpressionCompiler.cpp

@@ -225,14 +225,14 @@ bool ExpressionCompiler::visit(FunctionCall const& _functionCall)
 					BOOST_THROW_EXCEPTION(CompilerError()
 										  << errinfo_sourceLocation(arguments[i]->getLocation())
 										  << errinfo_comment("Type " + type.toString() + " not yet supported."));
-				if (numBytes != 32)
+				bool const leftAligned = type.getCategory() == Type::Category::STRING;
-					m_context << (u256(1) << ((32 - numBytes) * 8)) << eth::Instruction::MUL;
+				CompilerUtils(m_context).storeInMemory(dataOffset, numBytes, leftAligned);
-				m_context << u256(dataOffset) << eth::Instruction::MSTORE;
 				dataOffset += numBytes;
 			}
 			//@todo only return the first return value for now
-			unsigned retSize = function.getReturnParameterTypes().empty() ? 0
+			Type const* firstType = function.getReturnParameterTypes().empty() ? nullptr :
-									: function.getReturnParameterTypes().front()->getCalldataEncodedSize();
+									function.getReturnParameterTypes().front().get();
+			unsigned retSize = firstType ? firstType->getCalldataEncodedSize() : 0;
 			// CALL arguments: outSize, outOff, inSize, inOff, value, addr, gas (stack top)
 			m_context << u256(retSize) << u256(0) << u256(dataOffset) << u256(0) << u256(0);
 			_functionCall.getExpression().accept(*this); // pushes addr and function index
@@ -240,11 +240,11 @@ bool ExpressionCompiler::visit(FunctionCall const& _functionCall)
 					  << u256(25) << eth::Instruction::GAS << eth::Instruction::SUB
 					  << eth::Instruction::CALL
 					  << eth::Instruction::POP; // @todo do not ignore failure indicator
-			if (retSize == 32)
+			if (retSize > 0)
-				m_context << u256(0) << eth::Instruction::MLOAD;
+			{
-			else if (retSize > 0)
+				bool const leftAligned = firstType->getCategory() == Type::Category::STRING;
-				m_context << (u256(1) << ((32 - retSize) * 8))
+				CompilerUtils(m_context).loadFromMemory(0, retSize, leftAligned);
-						  << u256(0) << eth::Instruction::MLOAD << eth::Instruction::DIV;
+			}
 			break;
 		}
 		case Location::SEND:
@@ -267,7 +267,8 @@ bool ExpressionCompiler::visit(FunctionCall const& _functionCall)
 			arguments.front()->accept(*this);
 			appendTypeConversion(*arguments.front()->getType(), *function.getParameterTypes().front(), true);
 			// @todo move this once we actually use memory
-			m_context << u256(0) << eth::Instruction::MSTORE << u256(32) << u256(0) << eth::Instruction::SHA3;
+			CompilerUtils(m_context).storeInMemory(0);
+			m_context << u256(32) << u256(0) << eth::Instruction::SHA3;
 			break;
 		case Location::ECRECOVER:
 		case Location::SHA256:
@@ -283,13 +284,13 @@ bool ExpressionCompiler::visit(FunctionCall const& _functionCall)
 				arguments[i]->accept(*this);
 				appendTypeConversion(*arguments[i]->getType(), *function.getParameterTypes()[i], true);
 				// @todo move this once we actually use memory
-				m_context << u256(i * 32) << eth::Instruction::MSTORE;
+				CompilerUtils(m_context).storeInMemory(i * 32);
 			}
 			m_context << u256(32) << u256(0) << u256(arguments.size() * 32) << u256(0) << u256(0)
 					  << contractAddress << u256(500) //@todo determine actual gas requirement
 					  << eth::Instruction::CALL
-					  << eth::Instruction::POP
+					  << eth::Instruction::POP;
-					  << u256(0) << eth::Instruction::MLOAD;
+			CompilerUtils(m_context).loadFromMemory(0);
 			break;
 		}
 		default:
@@ -373,7 +374,8 @@ bool ExpressionCompiler::visit(IndexAccess const& _indexAccess)
 						 *dynamic_cast<MappingType const&>(*_indexAccess.getBaseExpression().getType()).getKeyType(),
 						 true);
 	// @todo move this once we actually use memory
-	m_context << u256(32) << eth::Instruction::MSTORE << u256(0) << eth::Instruction::MSTORE;
+	CompilerUtils(m_context).storeInMemory(0);
+	CompilerUtils(m_context).storeInMemory(32);
 	m_context << u256(64) << u256(0) << eth::Instruction::SHA3;
 
 	m_currentLValue = LValue(m_context, LValue::STORAGE, *_indexAccess.getType());
@@ -411,10 +413,11 @@ void ExpressionCompiler::endVisit(Literal const& _literal)
 	{
 	case Type::Category::INTEGER:
 	case Type::Category::BOOL:
+	case Type::Category::STRING:
 		m_context << _literal.getType()->literalValue(_literal);
 		break;
 	default:
-		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Only integer and boolean literals implemented for now."));
+		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Only integer, boolean and string literals implemented for now."));
 	}
 }
 
@@ -550,6 +553,11 @@ void ExpressionCompiler::appendTypeConversion(Type const& _typeOnStack, Type con
 		return;
 	if (_typeOnStack.getCategory() == Type::Category::INTEGER)
 		appendHighBitsCleanup(dynamic_cast<IntegerType const&>(_typeOnStack));
+	else if (_typeOnStack.getCategory() == Type::Category::STRING)
+	{
+		// nothing to do, strings are high-order-bit-aligned
+		//@todo clear lower-order bytes if we allow explicit conversion to shorter strings
+	}
 	else if (_typeOnStack != _targetType)
 		// All other types should not be convertible to non-equal types.
 		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Invalid type conversion requested."));

ExpressionCompiler.h

@@ -35,6 +35,7 @@ namespace solidity {
 class CompilerContext;
 class Type;
 class IntegerType;
+class StaticStringType;
 
 /**
  * Compiler for expressions, i.e. converts an AST tree whose root is an Expression into a stream
@@ -75,7 +76,7 @@ private:
 	/// @}
 
 	/// Appends an implicit or explicit type conversion. For now this comprises only erasing
-	/// higher-order bits (@see appendHighBitCleanup) when widening integer types.
+	/// higher-order bits (@see appendHighBitCleanup) when widening integer.
 	/// If @a _cleanupNeeded, high order bits cleanup is also done if no type conversion would be
 	/// necessary.
 	void appendTypeConversion(Type const& _typeOnStack, Type const& _targetType, bool _cleanupNeeded = false);

Token.h

@@ -269,6 +269,39 @@ namespace solidity
 	K(ADDRESS, "address", 0)                                           \
 	K(BOOL, "bool", 0)                                                 \
 	K(STRING_TYPE, "string", 0)                                        \
+	K(STRING0, "string0", 0)                                           \
+	K(STRING1, "string1", 0)                                           \
+	K(STRING2, "string2", 0)                                           \
+	K(STRING3, "string3", 0)                                           \
+	K(STRING4, "string4", 0)                                           \
+	K(STRING5, "string5", 0)                                           \
+	K(STRING6, "string6", 0)                                           \
+	K(STRING7, "string7", 0)                                           \
+	K(STRING8, "string8", 0)                                           \
+	K(STRING9, "string9", 0)                                           \
+	K(STRING10, "string10", 0)                                         \
+	K(STRING11, "string11", 0)                                         \
+	K(STRING12, "string12", 0)                                         \
+	K(STRING13, "string13", 0)                                         \
+	K(STRING14, "string14", 0)                                         \
+	K(STRING15, "string15", 0)                                         \
+	K(STRING16, "string16", 0)                                         \
+	K(STRING17, "string17", 0)                                         \
+	K(STRING18, "string18", 0)                                         \
+	K(STRING19, "string19", 0)                                         \
+	K(STRING20, "string20", 0)                                         \
+	K(STRING21, "string21", 0)                                         \
+	K(STRING22, "string22", 0)                                         \
+	K(STRING23, "string23", 0)                                         \
+	K(STRING24, "string24", 0)                                         \
+	K(STRING25, "string25", 0)                                         \
+	K(STRING26, "string26", 0)                                         \
+	K(STRING27, "string27", 0)                                         \
+	K(STRING28, "string28", 0)                                         \
+	K(STRING29, "string29", 0)                                         \
+	K(STRING30, "string30", 0)                                         \
+	K(STRING31, "string31", 0)                                         \
+	K(STRING32, "string32", 0)                                         \
 	K(TEXT, "text", 0)                                                 \
 	K(REAL, "real", 0)                                                 \
 	K(UREAL, "ureal", 0)                                               \

Types.cpp

@@ -53,6 +53,8 @@ shared_ptr<Type const> Type::fromElementaryTypeName(Token::Value _typeToken)
 		return make_shared<IntegerType const>(0, IntegerType::Modifier::ADDRESS);
 	else if (_typeToken == Token::BOOL)
 		return make_shared<BoolType const>();
+	else if (Token::STRING0 <= _typeToken && _typeToken <= Token::STRING32)
+		return make_shared<StaticStringType const>(int(_typeToken) - int(Token::STRING0));
 	else
 		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Unable to convert elementary typename " +
 																		 std::string(Token::toString(_typeToken)) + " to type."));
@@ -91,7 +93,8 @@ shared_ptr<Type const> Type::forLiteral(Literal const& _literal)
 	case Token::NUMBER:
 		return IntegerType::smallestTypeForLiteral(_literal.getValue());
 	case Token::STRING_LITERAL:
-		return shared_ptr<Type const>(); // @todo add string literals
+		//@todo put larger strings into dynamic strings
+		return StaticStringType::smallestTypeForLiteral(_literal.getValue());
 	default:
 		return shared_ptr<Type const>();
 	}
@@ -194,6 +197,44 @@ const MemberList IntegerType::AddressMemberList =
 					{"send", make_shared<FunctionType const>(TypePointers({make_shared<IntegerType const>(256)}),
 															 TypePointers(), FunctionType::Location::SEND)}});
 
+shared_ptr<StaticStringType> StaticStringType::smallestTypeForLiteral(string const& _literal)
+{
+	if (_literal.length() <= 32)
+		return make_shared<StaticStringType>(_literal.length());
+	return shared_ptr<StaticStringType>();
+}
+
+StaticStringType::StaticStringType(int _bytes): m_bytes(_bytes)
+{
+	if (asserts(m_bytes >= 0 && m_bytes <= 32))
+		BOOST_THROW_EXCEPTION(InternalCompilerError() << errinfo_comment("Invalid byte number for static string type: " +
+																		 dev::toString(m_bytes)));
+}
+
+bool StaticStringType::isImplicitlyConvertibleTo(Type const& _convertTo) const
+{
+	if (_convertTo.getCategory() != getCategory())
+		return false;
+	StaticStringType const& convertTo = dynamic_cast<StaticStringType const&>(_convertTo);
+	return convertTo.m_bytes >= m_bytes;
+}
+
+bool StaticStringType::operator==(Type const& _other) const
+{
+	if (_other.getCategory() != getCategory())
+		return false;
+	StaticStringType const& other = dynamic_cast<StaticStringType const&>(_other);
+	return other.m_bytes == m_bytes;
+}
+
+u256 StaticStringType::literalValue(const Literal& _literal) const
+{
+	u256 value = 0;
+	for (char c: _literal.getValue())
+		value = (value << 8) | byte(c);
+	return value << ((32 - _literal.getValue().length()) * 8);
+}
+
 bool BoolType::isExplicitlyConvertibleTo(Type const& _convertTo) const
 {
 	// conversion to integer is fine, but not to address

Types.h

@@ -36,7 +36,7 @@ namespace dev
 namespace solidity
 {
 
-// @todo realMxN, string<N>
+// @todo realMxN, dynamic strings, text, arrays
 
 class Type; // forward
 using TypePointer = std::shared_ptr<Type const>;
@@ -178,6 +178,35 @@ private:
 	static const MemberList AddressMemberList;
 };
 
+/**
+ * String type with fixed length, up to 32 bytes.
+ */
+class StaticStringType: public Type
+{
+public:
+	virtual Category getCategory() const override { return Category::STRING; }
+
+	/// @returns the smallest string type for the given literal or an empty pointer
+	/// if no type fits.
+	static std::shared_ptr<StaticStringType> smallestTypeForLiteral(std::string const& _literal);
+
+	StaticStringType(int _bytes);
+
+	virtual bool isImplicitlyConvertibleTo(Type const& _convertTo) const override;
+	virtual bool operator==(Type const& _other) const override;
+
+	virtual unsigned getCalldataEncodedSize() const override { return m_bytes; }
+	virtual bool isValueType() const override { return true; }
+
+	virtual std::string toString() const override { return "string" + dev::toString(m_bytes); }
+	virtual u256 literalValue(Literal const& _literal) const override;
+
+	int getNumBytes() const { return m_bytes; }
+
+private:
+	int m_bytes;
+};
+
 /**
  * The boolean type.
  */