Assembly output for Assembly object.

3 files changed, 105 insertions, 51 deletions

diff --git a/libevmasm/Assembly.cpp b/libevmasm/Assembly.cpp
index a9ca24dc..f50a38a6 100644
--- a/libevmasm/Assembly.cpp
+++ b/libevmasm/Assembly.cpp
@@ -117,69 +117,36 @@ string Assembly::locationFromSources(StringMap const& _sourceCodes, SourceLocati
 
 ostream& Assembly::streamAsm(ostream& _out, string const& _prefix, StringMap const& _sourceCodes) const
 {
-	_out << _prefix << ".code:" << endl;
-	for (AssemblyItem const& i: m_items)
+	for (size_t i = 0; i < m_items.size(); ++i)
 	{
-		_out << _prefix;
-		switch (i.type())
+		AssemblyItem const& item = m_items[i];
+		if (!item.location().isEmpty() && (i == 0 || m_items[i - 1].location() != item.location()))
 		{
-		case Operation:
-			_out << "  " << instructionInfo(i.instruction()).name  << "\t" << i.getJumpTypeAsString();
-			break;
-		case Push:
-			_out << "  PUSH" << dec << max<unsigned>(1, dev::bytesRequired(i.data())) << " 0x" << hex << i.data();
-			break;
-		case PushString:
-			_out << "  PUSH \"" << m_strings.at((h256)i.data()) << "\"";
-			break;
-		case PushTag:
-			if (i.data() == 0)
-				_out << "  PUSH [ErrorTag]";
-			else
-			{
-				size_t subId = i.splitForeignPushTag().first;
-				if (subId == size_t(-1))
-					_out << "  PUSH [tag" << dec << i.splitForeignPushTag().second << "]";
-				else
-					_out << "  PUSH [tag" << dec << subId << ":" << i.splitForeignPushTag().second << "]";
-			}
-			break;
-		case PushSub:
-			_out << "  PUSH [$" << size_t(i.data()) << "]";
-			break;
-		case PushSubSize:
-			_out << "  PUSH #[$" << size_t(i.data()) << "]";
-			break;
-		case PushProgramSize:
-			_out << "  PUSHSIZE";
-			break;
-		case PushLibraryAddress:
-			_out << "  PUSHLIB \"" << m_libraries.at(h256(i.data())) << "\"";
-			break;
-		case Tag:
-			_out << "tag" << dec << i.data() << ": " << endl << _prefix << "  JUMPDEST";
-			break;
-		case PushData:
-			_out << "  PUSH [" << hex << (unsigned)i.data() << "]";
-			break;
-		default:
-			BOOST_THROW_EXCEPTION(InvalidOpcode());
+			_out << _prefix << "    /*";
+			if (item.location().sourceName)
+				_out << " \"" + *item.location().sourceName + "\"";
+			if (!item.location().isEmpty())
+				_out << ":" << to_string(item.location().start) + ":" + to_string(item.location().end);
+			_out << " */" << endl;
 		}
-		_out << "\t\t" << locationFromSources(_sourceCodes, i.location()) << endl;
+		_out << _prefix << (item.type() == Tag ? "" : "  ") << item.toAssemblyText() << endl;
 	}
 
 	if (!m_data.empty() || !m_subs.empty())
 	{
-		_out << _prefix << ".data:" << endl;
+		_out << _prefix << "stop" << endl;
+		Json::Value data;
 		for (auto const& i: m_data)
-			if (u256(i.first) >= m_subs.size())
-				_out << _prefix << "  " << hex << (unsigned)(u256)i.first << ": " << dev::toHex(i.second) << endl;
+			assertThrow(u256(i.first) < m_subs.size(), AssemblyException, "Data not yet implemented.");
+
 		for (size_t i = 0; i < m_subs.size(); ++i)
 		{
-			_out << _prefix << "  " << hex << i << ": " << endl;
-			m_subs[i]->stream(_out, _prefix + "  ", _sourceCodes);
+			_out << endl << _prefix << "sub_" << i << ": assembly {\n";
+			m_subs[i]->streamAsm(_out, _prefix + "    ", _sourceCodes);
+			_out << _prefix << "}" << endl;
 		}
 	}
+
 	return _out;
 }
 
diff --git a/libevmasm/AssemblyItem.cpp b/libevmasm/AssemblyItem.cpp
index 54e38de8..8b3f920a 100644
--- a/libevmasm/AssemblyItem.cpp
+++ b/libevmasm/AssemblyItem.cpp
@@ -20,6 +20,7 @@
  */
 
 #include "AssemblyItem.h"
+#include <libevmasm/SemanticInformation.h>
 #include <fstream>
 
 using namespace std;
@@ -97,6 +98,28 @@ int AssemblyItem::deposit() const
 	return 0;
 }
 
+bool AssemblyItem::canBeFunctional() const
+{
+	switch (m_type)
+	{
+	case Operation:
+		return !SemanticInformation::isDupInstruction(*this) && !SemanticInformation::isSwapInstruction(*this);
+	case Push:
+	case PushString:
+	case PushTag:
+	case PushData:
+	case PushSub:
+	case PushSubSize:
+	case PushProgramSize:
+	case PushLibraryAddress:
+		return true;
+	case Tag:
+		return false;
+	default:;
+	}
+	return 0;
+}
+
 string AssemblyItem::getJumpTypeAsString() const
 {
 	switch (m_jumpType)
@@ -111,6 +134,65 @@ string AssemblyItem::getJumpTypeAsString() const
 	}
 }
 
+string AssemblyItem::toAssemblyText() const
+{
+	string text;
+	switch (type())
+	{
+	case Operation:
+	{
+		assertThrow(isValidInstruction(instruction()), AssemblyException, "Invalid instruction.");
+		string name = instructionInfo(instruction()).name;
+		transform(name.begin(), name.end(), name.begin(), [](unsigned char _c) { return tolower(_c); });
+		text = name;
+		break;
+	}
+	case Push:
+		text = toHex(toCompactBigEndian(data(), 1), 1, HexPrefix::Add);
+		break;
+	case PushString:
+		assertThrow(false, AssemblyException, "Push string assembly output not implemented.");
+		break;
+	case PushTag:
+		assertThrow(data() < 0x10000, AssemblyException, "Sub-assembly tags not yet implemented.");
+		text = string("tag_") + to_string(size_t(data()));
+		break;
+	case Tag:
+		assertThrow(data() < 0x10000, AssemblyException, "Sub-assembly tags not yet implemented.");
+		text = string("tag_") + to_string(size_t(data())) + ":";
+		break;
+	case PushData:
+		assertThrow(false, AssemblyException, "Push data not implemented.");
+		break;
+	case PushSub:
+		text = string("dataOffset(sub_") + to_string(size_t(data())) + ")";
+		break;
+	case PushSubSize:
+		text = string("dataSize(sub_") + to_string(size_t(data())) + ")";
+		break;
+	case PushProgramSize:
+		text = string("programSize");
+		break;
+	case PushLibraryAddress:
+		text = string("linkerSymbol(\"") + toHex(data()) + string("\")");
+		break;
+	case UndefinedItem:
+		assertThrow(false, AssemblyException, "Invalid assembly item.");
+		break;
+	default:
+		BOOST_THROW_EXCEPTION(InvalidOpcode());
+	}
+	if (m_jumpType == JumpType::IntoFunction || m_jumpType == JumpType::OutOfFunction)
+	{
+		text += "\t//";
+		if (m_jumpType == JumpType::IntoFunction)
+			text += " in";
+		else
+			text += " out";
+	}
+	return text;
+}
+
 ostream& dev::eth::operator<<(ostream& _out, AssemblyItem const& _item)
 {
 	switch (_item.type())
diff --git a/libevmasm/AssemblyItem.h b/libevmasm/AssemblyItem.h
index b5bd3ed8..cddfb17c 100644
--- a/libevmasm/AssemblyItem.h
+++ b/libevmasm/AssemblyItem.h
@@ -97,6 +97,9 @@ public:
 	unsigned bytesRequired(unsigned _addressLength) const;
 	int deposit() const;
 
+	/// @returns true if the assembly item can be used in a functional context.
+	bool canBeFunctional() const;
+
 	bool match(AssemblyItem const& _i) const { return _i.m_type == UndefinedItem || (m_type == _i.m_type && (m_type != Operation || m_data == _i.m_data)); }
 	void setLocation(SourceLocation const& _location) { m_location = _location; }
 	SourceLocation const& location() const { return m_location; }
@@ -108,6 +111,8 @@ public:
 	void setPushedValue(u256 const& _value) const { m_pushedValue = std::make_shared<u256>(_value); }
 	u256 const* pushedValue() const { return m_pushedValue.get(); }
 
+	std::string toAssemblyText() const;
+
 private:
 	AssemblyItemType m_type;
 	u256 m_data;