[libevmasm] Add support to import evm assembly json.

libevmasm/Assembly.cpp

@@ -34,13 +34,15 @@
 #include <liblangutil/CharStream.h>
 #include <liblangutil/Exceptions.h>
 
-#include <json/json.h>
+#include <libsolutil/JSON.h>
+#include <libsolutil/StringUtils.h>
 
 #include <range/v3/algorithm/any_of.hpp>
 #include <range/v3/view/enumerate.hpp>
 
 #include <fstream>
 #include <limits>
+#include <iterator>
 
 using namespace solidity;
 using namespace solidity::evmasm;
@@ -73,6 +75,207 @@ unsigned Assembly::codeSize(unsigned subTagSize) const
 	}
 }
 
+void Assembly::importAssemblyItemsFromJSON(Json::Value const& _code)
+{
+	solAssert(m_items.empty());
+	solRequire(_code.isArray(), AssemblyImportException, "Supplied JSON is not an array.");
+	for (auto current = begin(_code); current != end(_code); ++current)
+	{
+		auto const& item = m_items.emplace_back(createAssemblyItemFromJSON(*current, m_sourceList));
+		if (item == Instruction::JUMPDEST)
+			solThrow(AssemblyImportException, "JUMPDEST instruction without a tag");
+		else if (item.type() == AssemblyItemType::Tag)
+		{
+			++current;
+			if (current != end(_code) && createAssemblyItemFromJSON(*current, m_sourceList) != Instruction::JUMPDEST)
+				solThrow(AssemblyImportException, "JUMPDEST expected after tag.");
+		}
+	}
+}
+
+AssemblyItem Assembly::createAssemblyItemFromJSON(Json::Value const& _json, std::vector<std::string> const& _sourceList)
+{
+	solRequire(_json.isObject(), AssemblyImportException, "Supplied JSON is not an object.");
+	static set<string> const validMembers{"name", "begin", "end", "source", "value", "modifierDepth", "jumpType"};
+	for (auto const& member: _json.getMemberNames())
+		solRequire(
+			validMembers.count(member),
+			AssemblyImportException,
+			"Unknown member '" + member + "'. Valid members are " +
+			solidity::util::joinHumanReadable(validMembers, ", ") + "."
+		);
+	solRequire(isOfType<string>(_json["name"]), AssemblyImportException, "Member 'name' missing or not of type string.");
+	solRequire(isOfTypeIfExists<int>(_json, "begin"), AssemblyImportException, "Optional member 'begin' not of type int.");
+	solRequire(isOfTypeIfExists<int>(_json, "end"), AssemblyImportException, "Optional member 'end' not of type int.");
+	solRequire(isOfTypeIfExists<int>(_json, "source"), AssemblyImportException, "Optional member 'source' not of type int.");
+	solRequire(isOfTypeIfExists<string>(_json, "value"), AssemblyImportException, "Optional member 'value' not of type string.");
+	solRequire(
+		isOfTypeIfExists<int>(_json, "modifierDepth"),
+		AssemblyImportException,
+		"Optional member 'modifierDepth' not of type int."
+	);
+	solRequire(
+		isOfTypeIfExists<string>(_json, "jumpType"),
+		AssemblyImportException,
+		"Optional member 'jumpType' not of type string."
+	);
+
+	string name = get<string>(_json["name"]);
+	solRequire(!name.empty(), AssemblyImportException, "Member 'name' was empty.");
+
+	SourceLocation location;
+	location.start = get<int>(_json["begin"]);
+	location.end = get<int>(_json["end"]);
+	int srcIndex = getOrDefault<int>(_json["source"], -1);
+	size_t modifierDepth = static_cast<size_t>(getOrDefault<int>(_json["modifierDepth"], 0));
+	string value = getOrDefault<string>(_json["value"], "");
+	string jumpType = getOrDefault<string>(_json["jumpType"], "");
+
+	auto updateUsedTags = [&](u256 const& data)
+	{
+		m_usedTags = max(m_usedTags, static_cast<unsigned>(data) + 1);
+		return data;
+	};
+
+	auto storeImmutableHash = [&](string const& _immutableName) -> h256
+	{
+		h256 hash(util::keccak256(_immutableName));
+		solAssert(m_immutables.count(hash) == 0 || m_immutables[hash] == _immutableName);
+		m_immutables[hash] = _immutableName;
+		return hash;
+	};
+
+	auto storeLibraryHash = [&](string const& _libraryName) -> h256
+	{
+		h256 hash(util::keccak256(_libraryName));
+		solAssert(m_libraries.count(hash) == 0 || m_libraries[hash] == _libraryName);
+		m_libraries[hash] = _libraryName;
+		return hash;
+	};
+
+	auto requireValueDefinedForInstruction = [&](string const& _name, string const& _value)
+	{
+		solRequire(
+			!_value.empty(),
+			AssemblyImportException,
+			"Member 'value' was not defined for instruction '" + _name + "', but the instruction needs a value."
+		);
+	};
+
+	auto requireValueUndefinedForInstruction = [&](string const& _name, string const& _value)
+	{
+		solRequire(
+			_value.empty(),
+			AssemblyImportException,
+			"Member 'value' defined for instruction '" + _name + "', but the instruction does not need a value."
+		);
+	};
+
+	solRequire(srcIndex >= -1 && srcIndex < static_cast<int>(_sourceList.size()), AssemblyImportException, "srcIndex out of bound.");
+	if (srcIndex != -1)
+		location.sourceName = std::make_shared<std::string>(_sourceList[static_cast<size_t>(srcIndex)]);
+
+	AssemblyItem result(0);
+
+	if (c_instructions.count(name))
+	{
+		AssemblyItem item{c_instructions.at(name), location};
+		if (!jumpType.empty())
+		{
+			if (item.instruction() == Instruction::JUMP || item.instruction() == Instruction::JUMPI)
+				item.setJumpType(jumpType);
+			else
+				solThrow(
+					AssemblyImportException,
+					"Member 'jumpType' set on instruction different from JUMP or JUMPI (was set on instruction '" + name + "')"
+				);
+		}
+		InstructionInfo info = instructionInfo(item.instruction(), m_evmVersion);
+		if (info.args == 0)
+			requireValueUndefinedForInstruction(name, value);
+		result = item;
+	}
+	else
+	{
+		solRequire(
+			jumpType.empty(),
+			AssemblyImportException,
+			"Member 'jumpType' set on instruction different from JUMP or JUMPI (was set on instruction '" + name + "')"
+		);
+		if (name == "PUSH")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::Push, u256("0x" + value)};
+		}
+		else if (name == "PUSH [ErrorTag]")
+		{
+			requireValueUndefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushTag, 0};
+		}
+		else if (name == "PUSH [tag]")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushTag, updateUsedTags(u256(value))};
+		}
+		else if (name == "PUSH [$]")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushSub, u256("0x" + value)};
+		}
+		else if (name == "PUSH #[$]")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushSubSize, u256("0x" + value)};
+		}
+		else if (name == "PUSHSIZE")
+		{
+			requireValueUndefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushProgramSize, 0};
+		}
+		else if (name == "PUSHLIB")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushLibraryAddress, storeLibraryHash(value)};
+		}
+		else if (name == "PUSHDEPLOYADDRESS")
+		{
+			requireValueUndefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushDeployTimeAddress, 0};
+		}
+		else if (name == "PUSHIMMUTABLE")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushImmutable, storeImmutableHash(value)};
+		}
+		else if (name == "ASSIGNIMMUTABLE")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::AssignImmutable, storeImmutableHash(value)};
+		}
+		else if (name == "tag")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::Tag, updateUsedTags(u256(value))};
+		}
+		else if (name == "PUSH data")
+		{
+			requireValueDefinedForInstruction(name, value);
+			result = {AssemblyItemType::PushData, u256("0x" + value)};
+		}
+		else if (name == "VERBATIM")
+		{
+			requireValueUndefinedForInstruction(name, value);
+			AssemblyItem item(fromHex(value), 0, 0);
+			result = item;
+		}
+		else
+			solThrow(InvalidOpcode, "Invalid opcode: " + name);
+	}
+	result.setLocation(location);
+	result.m_modifierDepth = modifierDepth;
+	return result;
+}
+
 namespace
 {
 
@@ -221,7 +424,7 @@ std::string Assembly::assemblyString(
 	return tmp.str();
 }
 
-Json::Value Assembly::assemblyJSON(std::map<std::string, unsigned> const& _sourceIndices, bool _includeSourceList) const
+Json::Value Assembly::assemblyJSON(bool _includeSourceList) const
 {
 	Json::Value root;
 	root[".code"] = Json::arrayValue;
@@ -231,9 +434,9 @@ Json::Value Assembly::assemblyJSON(std::map<std::string, unsigned> const& _sourc
 		int sourceIndex = -1;
 		if (item.location().sourceName)
 		{
-			auto iter = _sourceIndices.find(*item.location().sourceName);
-			if (iter != _sourceIndices.end())
-				sourceIndex = static_cast<int>(iter->second);
+			for (size_t index = 0; index < m_sourceList.size(); ++index)
+				if (m_sourceList[index] == *item.location().sourceName)
+					sourceIndex = static_cast<int>(index);
 		}
 
 		auto [name, data] = item.nameAndData(m_evmVersion);
@@ -271,8 +474,8 @@ Json::Value Assembly::assemblyJSON(std::map<std::string, unsigned> const& _sourc
 	{
 		root["sourceList"] = Json::arrayValue;
 		Json::Value& jsonSourceList = root["sourceList"];
-		for (auto const& [name, index]: _sourceIndices)
-			jsonSourceList[index] = name;
+		for (int index = 0; index < static_cast<int>(m_sourceList.size()); ++index)
+			jsonSourceList[index] = m_sourceList[static_cast<size_t>(index)];
 	}
 
 	if (!m_data.empty() || !m_subs.empty())
@@ -285,9 +488,10 @@ Json::Value Assembly::assemblyJSON(std::map<std::string, unsigned> const& _sourc
 
 		for (size_t i = 0; i < m_subs.size(); ++i)
 		{
-			std::stringstream hexStr;
-			hexStr << std::hex << i;
-			data[hexStr.str()] = m_subs[i]->assemblyJSON(_sourceIndices, /*_includeSourceList = */false);
+			stringstream hexStr;
+			hexStr << hex << i;
+			m_subs[i]->setSourceList(m_sourceList);
+			data[hexStr.str()] = m_subs[i]->assemblyJSON(false);
 		}
 	}
 
@@ -297,7 +501,84 @@ Json::Value Assembly::assemblyJSON(std::map<std::string, unsigned> const& _sourc
 	return root;
 }
 
-AssemblyItem Assembly::namedTag(std::string const& _name, size_t _params, size_t _returns, std::optional<uint64_t> _sourceID)
+shared_ptr<Assembly> Assembly::fromJSON(Json::Value const& _json, vector<string> const& _sourceList, int _level)
+{
+	solRequire(_json.isObject(), AssemblyImportException, "Supplied JSON is not an object.");
+	static set<string> const validMembers{".code", ".data", ".auxdata", "sourceList"};
+	for (auto const& attribute: _json.getMemberNames())
+		solRequire(validMembers.count(attribute), AssemblyImportException, "Unknown attribute '" + attribute + "'.");
+	solRequire(_json.isMember(".code"), AssemblyImportException, "Member '.code' does not exist.");
+	solRequire(_json[".code"].isArray(), AssemblyImportException, "Member '.code' is not an array.");
+	for (auto const& codeItem: _json[".code"])
+		solRequire(codeItem.isObject(), AssemblyImportException, "Item of '.code' array is not an object.");
+
+	if (_level == 0)
+	{
+		if (_json.isMember("sourceList"))
+		{
+			solRequire(_json["sourceList"].isArray(), AssemblyImportException, "Optional member 'sourceList' is not an array.");
+			for (auto const& sourceListItem: _json["sourceList"])
+				solRequire(sourceListItem.isString(), AssemblyImportException, "Item of 'sourceList' array is not of type string.");
+		}
+	} else
+		solRequire(
+			!_json.isMember("sourceList"),
+			AssemblyImportException,
+			"Member 'sourceList' is only allowed in root JSON object."
+		);
+
+	shared_ptr<Assembly> result = make_shared<Assembly>(langutil::EVMVersion(), _level == 0, "");
+	if (_json.isMember("sourceList"))
+	{
+		solAssert(_level == 0);
+		for (auto const& it: _json["sourceList"])
+			result->m_sourceList.emplace_back(it.asString());
+	} else
+		result->m_sourceList = _sourceList;
+
+	result->importAssemblyItemsFromJSON(_json[".code"]);
+	if (_json[".auxdata"])
+	{
+		solRequire(_json[".auxdata"].isString(), AssemblyImportException, "Optional member '.auxdata' is not of type string.");
+		bytes auxdata{fromHex(_json[".auxdata"].asString())};
+		solRequire(!auxdata.empty(), AssemblyImportException, "Optional member '.auxdata' is not a valid hexadecimal string.");
+		result->m_auxiliaryData = auxdata;
+	}
+
+	if (_json.isMember(".data"))
+	{
+		solRequire(_json[".data"].isObject(), AssemblyImportException, "Optional member '.data' is not an object.");
+		Json::Value const& data = _json[".data"];
+		for (Json::ValueConstIterator dataIter = data.begin(); dataIter != data.end(); dataIter++)
+		{
+			solRequire(dataIter.key().isString(), AssemblyImportException, "Key inside '.data' is not of type string.");
+			string dataItemID = dataIter.key().asString();
+			Json::Value const& code = data[dataItemID];
+			if (code.isString())
+			{
+				bytes data_value{fromHex(code.asString())};
+				solRequire(
+					!data_value.empty(),
+					AssemblyImportException,
+					"Member '.data' contains a value for '" + dataItemID + "' that is not a valid hexadecimal string.");
+				result->m_data[h256(fromHex(dataItemID))] = fromHex(code.asString());
+			}
+			else if (code.isObject())
+			{
+				shared_ptr<Assembly> subassembly(Assembly::fromJSON(code, result->m_sourceList, _level + 1));
+				solAssert(subassembly);
+				result->m_subs.emplace_back(make_shared<Assembly>(*subassembly));
+				// TODO: this shouldn't be enough for the general case.
+				result->encodeSubPath({0, 0});
+			}
+			else
+				solThrow(AssemblyImportException, "Key inside '.data' '" + dataItemID + "' can only be a valid hex-string or an object.");
+		}
+	}
+	return result;
+}
+
+AssemblyItem Assembly::namedTag(string const& _name, size_t _params, size_t _returns, optional<uint64_t> _sourceID)
 {
 	assertThrow(!_name.empty(), AssemblyException, "Empty named tag.");
 	if (m_namedTags.count(_name))

libevmasm/Assembly.h

@@ -151,10 +151,30 @@ public:
 
 	/// Create a JSON representation of the assembly.
 	Json::Value assemblyJSON(
-		std::map<std::string, unsigned> const& _sourceIndices = std::map<std::string, unsigned>(),
 		bool _includeSourceList = true
 	) const;
 
+	/// Loads the JSON representation of assembly.
+	/// @param _json JSON object containing assembly in the format produced by assemblyJSON().
+	/// @param _sourceList list of source names (used to supply the source-list from the root-assembly object to sub-assemblies).
+	/// @param _level this function might be called recursively, _level reflects the nesting level.
+	/// @returns Resulting Assembly object loaded from given json.
+	static std::shared_ptr<Assembly> fromJSON(
+		Json::Value const& _json,
+		std::vector<std::string> const& _sourceList = {},
+		int _level = 0
+	);
+
+	void setSourceList(std::vector<std::string> const& _sourceList)
+	{
+		m_sourceList = _sourceList;
+	}
+
+	std::vector<std::string> const& sourceList()
+	{
+		return m_sourceList;
+	}
+
 	/// Mark this assembly as invalid. Calling ``assemble`` on it will throw.
 	void markAsInvalid() { m_invalid = true; }
 
@@ -171,6 +191,18 @@ protected:
 
 	unsigned codeSize(unsigned subTagSize) const;
 
+	/// Add all assembly items from given JSON array. This function imports the items by iterating through
+	/// the code array. This method only works on clean Assembly objects that don't have any items defined yet.
+	/// @param _json JSON array that contains assembly items (e.g. json['.code'])
+	/// @param _sourceList list of source names.
+	void importAssemblyItemsFromJSON(Json::Value const& _code);
+
+	/// Creates an AssemblyItem from a given JSON representation.
+	/// @param _json JSON object that consists a single assembly item
+	/// @param _sourceList list of source names.
+	/// @returns AssemblyItem of _json argument.
+	AssemblyItem createAssemblyItemFromJSON(Json::Value const& _json, std::vector<std::string> const& _sourceList);
+
 private:
 	bool m_invalid = false;
 
@@ -217,7 +249,7 @@ protected:
 	/// Internal name of the assembly object, only used with the Yul backend
 	/// currently
 	std::string m_name;
-
+	std::vector<std::string> m_sourceList;
 	langutil::SourceLocation m_currentSourceLocation;
 
 public:

libevmasm/AssemblyItem.cpp

@@ -247,6 +247,18 @@ std::string AssemblyItem::getJumpTypeAsString() const
 	}
 }
 
+void AssemblyItem::setJumpType(string const& _jumpType)
+{
+	if (_jumpType == "[in]")
+		m_jumpType = JumpType::IntoFunction;
+	else if (_jumpType == "[out]")
+		m_jumpType = JumpType::OutOfFunction;
+	else if (_jumpType.empty())
+		m_jumpType = JumpType::Ordinary;
+	else
+		solThrow(AssemblyImportException, "Invalid jump type.");
+}
+
 std::string AssemblyItem::toAssemblyText(Assembly const& _assembly) const
 {
 	std::string text;

libevmasm/AssemblyItem.h

@@ -174,6 +174,7 @@ public:
 	langutil::SourceLocation const& location() const { return m_location; }
 
 	void setJumpType(JumpType _jumpType) { m_jumpType = _jumpType; }
+	void setJumpType(std::string const& _jumpType);
 	JumpType getJumpType() const { return m_jumpType; }
 	std::string getJumpTypeAsString() const;
 

libevmasm/CMakeLists.txt

@@ -3,6 +3,8 @@ set(sources
 	Assembly.h
 	AssemblyItem.cpp
 	AssemblyItem.h
+	EVMAssemblyStack.cpp
+	EVMAssemblyStack.h
 	BlockDeduplicator.cpp
 	BlockDeduplicator.h
 	CommonSubexpressionEliminator.cpp

libevmasm/EVMAssemblyStack.cpp

@@ -0,0 +1,58 @@
+/*
+	This file is part of solidity.
+
+	solidity 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.
+
+	solidity 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 solidity.  If not, see <http://www.gnu.org/licenses/>.
+*/
+// SPDX-License-Identifier: GPL-3.0
+
+#include <libevmasm/EVMAssemblyStack.h>
+
+#include <libsolutil/JSON.h>
+#include <liblangutil/Exceptions.h>
+#include <libsolidity/codegen/CompilerContext.h>
+
+#include <utility>
+
+using namespace solidity::util;
+using namespace solidity::langutil;
+using namespace solidity::frontend;
+using namespace std;
+
+namespace solidity::evmasm
+{
+
+bool EVMAssemblyStack::parseAndAnalyze(std::string const& _sourceName, std::string const& _source)
+{
+	m_name = _sourceName;
+	if (jsonParseStrict(_source, m_json))
+	{
+		m_evmAssembly = evmasm::Assembly::fromJSON(m_json);
+		return m_evmAssembly != nullptr;
+	}
+	return false;
+}
+
+void EVMAssemblyStack::assemble()
+{
+	solAssert(m_evmAssembly->isCreation());
+	m_object = m_evmAssembly->assemble();
+	if (m_evmAssembly->numSubs() > 0)
+	{
+		m_evmRuntimeAssembly = make_shared<evmasm::Assembly>(m_evmAssembly->sub(0));
+		solAssert(m_evmRuntimeAssembly && !m_evmRuntimeAssembly->isCreation());
+		m_runtimeObject = m_evmRuntimeAssembly->assemble();
+	}
+}
+
+} // namespace solidity::evmasm

libevmasm/EVMAssemblyStack.h

@@ -0,0 +1,68 @@
+/*
+	This file is part of solidity.
+
+	solidity 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.
+
+	solidity 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 solidity.  If not, see <http://www.gnu.org/licenses/>.
+*/
+// SPDX-License-Identifier: GPL-3.0
+
+#pragma once
+
+#include <map>
+#include <string>
+
+#include <libsolutil/JSON.h>
+
+#include <libevmasm/Assembly.h>
+#include <libevmasm/LinkerObject.h>
+
+namespace solidity::evmasm
+{
+
+class EVMAssemblyStack
+{
+public:
+	explicit EVMAssemblyStack(langutil::EVMVersion _evmVersion): m_evmVersion(_evmVersion) {}
+
+	/// Runs parsing and analysis, returns false if input cannot be assembled.
+	/// Multiple calls overwrite the previous state.
+	bool parseAndAnalyze(std::string const& _sourceName, std::string const& _source);
+
+	void assemble();
+
+	std::string const& name() const { return m_name; }
+
+	evmasm::LinkerObject const& object() const { return m_object; }
+
+	std::shared_ptr<evmasm::Assembly> const& evmAssembly() const { return m_evmAssembly; }
+
+	evmasm::LinkerObject const& runtimeObject() const { return m_runtimeObject; }
+
+	std::shared_ptr<evmasm::Assembly> const& evmRuntimeAssembly() const { return m_evmRuntimeAssembly; }
+
+	Json::Value json() const
+	{
+		return m_json;
+	}
+
+private:
+	langutil::EVMVersion m_evmVersion;
+	std::string m_name;
+	Json::Value m_json;
+	std::shared_ptr<evmasm::Assembly> m_evmAssembly;
+	std::shared_ptr<evmasm::Assembly> m_evmRuntimeAssembly;
+	evmasm::LinkerObject m_object; ///< Deployment object (includes the runtime sub-object).
+	evmasm::LinkerObject m_runtimeObject; ///< Runtime object.
+};
+
+} // namespace solidity::evmasm

libevmasm/Exceptions.h

@@ -28,6 +28,7 @@ namespace solidity::evmasm
 {
 
 struct AssemblyException: virtual util::Exception {};
+struct AssemblyImportException: virtual AssemblyException {};
 struct OptimizerException: virtual AssemblyException {};
 struct StackTooDeepException: virtual OptimizerException {};
 struct ItemNotAvailableException: virtual OptimizerException {};

libsolidity/interface/CompilerStack.cpp

@@ -72,6 +72,7 @@
 #include <liblangutil/SemVerHandler.h>
 #include <liblangutil/SourceReferenceFormatter.h>
 
+#include <libevmasm/EVMAssemblyStack.h>
 #include <libevmasm/Exceptions.h>
 
 #include <libsolutil/SwarmHash.h>
@@ -438,6 +439,27 @@ void CompilerStack::importASTs(map<string, Json::Value> const& _sources)
 	storeContractDefinitions();
 }
 
+void CompilerStack::importFromEVMAssemblyStack(std::string const& _sourceName, std::string const& _source)
+{
+	solRequire(m_stackState == Empty, CompilerError, "");
+	m_evmAssemblyStack = std::make_unique<evmasm::EVMAssemblyStack>(m_evmVersion);
+	Json::Value evmAsmJson;
+	if (m_evmAssemblyStack->parseAndAnalyze(_sourceName, _source))
+	{
+		m_evmAssemblyStack->assemble();
+		string const name{m_evmAssemblyStack->name()};
+		Json::Value const& json = m_evmAssemblyStack->json();
+		m_sourceJsons[name] = json;
+		Contract& contract = m_contracts[name];
+		contract.evmAssembly = m_evmAssemblyStack->evmAssembly();
+		contract.evmRuntimeAssembly= m_evmAssemblyStack->evmRuntimeAssembly();
+		contract.object = m_evmAssemblyStack->object();
+		contract.runtimeObject = m_evmAssemblyStack->runtimeObject();
+
+		m_stackState = CompilationSuccessful;
+	}
+}
+
 bool CompilerStack::analyze()
 {
 	if (m_stackState != ParsedAndImported || m_stackState >= AnalysisPerformed)
@@ -674,6 +696,7 @@ bool CompilerStack::isRequestedContract(ContractDefinition const& _contract) con
 
 bool CompilerStack::compile(State _stopAfter)
 {
+	solAssert(m_compilationSourceType != CompilationSourceType::EvmAssemblyJSON);
 	m_stopAfter = _stopAfter;
 	if (m_stackState < AnalysisPerformed)
 		if (!parseAndAnalyze(_stopAfter))
@@ -957,7 +980,13 @@ Json::Value CompilerStack::assemblyJSON(string const& _contractName) const
 
 	Contract const& currentContract = contract(_contractName);
 	if (currentContract.evmAssembly)
-		return currentContract.evmAssembly->assemblyJSON(sourceIndices());
+	{
+		std::vector<std::string> sources = sourceNames();
+		if (std::find(sources.begin(), sources.end(), CompilerContext::yulUtilityFileName()) == sources.end())
+			sources.emplace_back(CompilerContext::yulUtilityFileName());
+		currentContract.evmAssembly->setSourceList(sources);
+		return currentContract.evmAssembly->assemblyJSON();
+	}
 	else
 		return Json::Value();
 }
@@ -971,9 +1000,20 @@ map<string, unsigned> CompilerStack::sourceIndices() const
 {
 	map<string, unsigned> indices;
 	unsigned index = 0;
+	if (m_evmAssemblyStack)
+	{
+		for (auto const& s: m_evmAssemblyStack->evmAssembly()->sourceList())
+			if (s != CompilerContext::yulUtilityFileName())
+				indices[s] = index++;
+	}
+	else
+	{
 		for (auto const& s: m_sources)
+			if (s.first != CompilerContext::yulUtilityFileName())
 				indices[s.first] = index++;
-	solAssert(!indices.count(CompilerContext::yulUtilityFileName()), "");
+	}
+
+	if (indices.find(CompilerContext::yulUtilityFileName()) == indices.end())
 		indices[CompilerContext::yulUtilityFileName()] = index++;
 	return indices;
 }
@@ -1589,6 +1629,11 @@ string CompilerStack::createMetadata(Contract const& _contract, bool _forIR) con
 	case CompilationSourceType::SolidityAST:
 		sourceType = "SolidityAST";
 		break;
+	case CompilationSourceType::EvmAssemblyJSON:
+		sourceType = "EvmAssemblyJson";
+		break;
+	default:
+		solAssert(false);
 	}
 	meta["language"] = sourceType;
 	meta["compiler"]["version"] = VersionStringStrict;

libsolidity/interface/CompilerStack.h

@@ -42,6 +42,7 @@
 #include <liblangutil/SourceLocation.h>
 
 #include <libevmasm/LinkerObject.h>
+#include <libevmasm/EVMAssemblyStack.h>
 
 #include <libsolutil/Common.h>
 #include <libsolutil/FixedHash.h>
@@ -121,7 +122,9 @@ public:
 		/// Regular compilation from Solidity source files.
 		Solidity,
 		/// Compilation from an imported Solidity AST.
-		SolidityAST
+		SolidityAST,
+		/// Compilation from an imported EVM Assembly JSON.
+		EvmAssemblyJSON
 	};
 
 	/// Creates a new compiler stack.
@@ -230,6 +233,8 @@ public:
 	/// Will throw errors if the import fails
 	void importASTs(std::map<std::string, Json::Value> const& _sources);
 
+	void importFromEVMAssemblyStack(std::string const& _sourceName, std::string const& _source);
+
 	/// Performs the analysis steps (imports, scopesetting, syntaxCheck, referenceResolving,
 	///  typechecking, staticAnalysis) on previously parsed sources.
 	/// @returns false on error.
@@ -524,6 +529,7 @@ private:
 	/// If this is true, the stack will refuse to generate code.
 	bool m_hasError = false;
 	MetadataFormat m_metadataFormat = defaultMetadataFormat();
+	std::unique_ptr<evmasm::EVMAssemblyStack> m_evmAssemblyStack;
 };
 
 }

scripts/ASTImportTest.sh

@@ -23,6 +23,18 @@
 # ast import/export tests:
 #   - first exporting a .sol file to JSON, then loading it into the compiler
 #     and exporting it again. The second JSON should be identical to the first.
+#
+# evm-assembly import/export tests:
+#   - first a .sol file will be exported to a combined json file, containing outputs
+#     for "asm" "bin" "bin-runtime" "opcodes" "srcmap" and "srcmap-runtime" (expected output).
+#     The "asm" output will then be used as import, where its output "bin" "bin-runtime"
+#     "opcodes" "srcmap" "srcmap-runtime" (obtained output) will be compared with the expected output.
+#     The expected output needs to be identical with the obtained output.
+#
+#     Additionally to this, the direct import/export is tested by importing an
+#     evm-assembly json with --import-asm-json and directly exporting it again with
+#     --asm-json using the same solc invocation. The original asm json file used for the
+#     import and the resulting exported asm json file need to be identical.
 
 set -euo pipefail
 
@@ -40,7 +52,7 @@ source "${REPO_ROOT}/scripts/common.sh"
 
 function print_usage
 {
-    echo "Usage: ${0} ast [--exit-on-error|--help]."
+    echo "Usage: ${0} ast|evm-assembly [--exit-on-error|--help]."
 }
 
 function print_used_commands
@@ -81,6 +93,7 @@ for PARAM in "$@"
 do
     case "$PARAM" in
         ast) check_import_test_type_unset ; IMPORT_TEST_TYPE="ast" ;;
+        evm-assembly) check_import_test_type_unset ; IMPORT_TEST_TYPE="evm-assembly" ;;
         --help) print_usage ; exit 0 ;;
         --exit-on-error) EXIT_ON_ERROR=1 ;;
         *) fail "Unknown option '$PARAM'. Aborting. $(print_usage)" ;;
@@ -89,6 +102,7 @@ done
 
 SYNTAXTESTS_DIR="${REPO_ROOT}/test/libsolidity/syntaxTests"
 ASTJSONTESTS_DIR="${REPO_ROOT}/test/libsolidity/ASTJSON"
+SEMANTICTESTS_DIR="${REPO_ROOT}/test/libsolidity/semanticTests"
 
 FAILED=0
 UNCOMPILABLE=0
@@ -153,6 +167,125 @@ function test_ast_import_export_equivalence
     TESTED=$((TESTED + 1))
 }
 
+function test_evmjson_import_export_equivalence
+{
+    local sol_file="$1"
+    local input_files=( "${@:2}" )
+    local outputs=( "asm" "bin" "bin-runtime" "opcodes" "srcmap" "srcmap-runtime" )
+    local export_command=("$SOLC" --combined-json "$(IFS=, ; echo "${outputs[*]}")" --pretty-json --json-indent 4 "${input_files[@]}")
+    local success=1
+    if ! "${export_command[@]}" > expected.json 2> expected.error
+    then
+        success=0
+        printError "ERROR: (export) EVM Assembly JSON reimport failed for ${sol_file}"
+        if (( EXIT_ON_ERROR == 1 ))
+        then
+            print_used_commands "$(pwd)" "${export_command[*]}" ""
+            return 1
+        fi
+    fi
+    if ! "${export_command[@]}" "--optimize" > expected.optimize.json 2> expected.optimize.error
+    then
+        success=0
+        printError "ERROR: (export with --optimize) EVM Assembly JSON reimport failed for ${sol_file}"
+        if (( EXIT_ON_ERROR == 1 ))
+        then
+            print_used_commands "$(pwd)" "${export_command[*]} --optimize" ""
+            return 1
+        fi
+    fi
+
+    for input_json in "expected.json" "expected.optimize.json"
+    do
+        local optimize_param=""
+        if [[ "$input_json" == "expected.optimize.json" ]]
+        then
+            optimize_param="--optimize"
+        fi
+
+        # Note that we have some test files, that only consist of free functions.
+        # Those files don't define any contracts, so the resulting JSON does not have any
+        # keys. In this case `jq` returns an error like `jq: error: null (null) has no keys`.
+        # To not get spammed by these errors, errors are redirected to /dev/null.
+        for contract in $(jq '.contracts | keys | .[]' "$input_json" 2> /dev/null)
+        do
+            for output in "${outputs[@]}"
+            do
+                jq --raw-output ".contracts.${contract}.\"${output}\"" "$input_json" > "expected.${output}.json"
+            done
+
+            assembly=$(cat expected.asm.json)
+            [[ $assembly != "" && $assembly != "null" ]] || continue
+
+            local import_command=("${SOLC}" --combined-json "bin,bin-runtime,opcodes,asm,srcmap,srcmap-runtime" --pretty-json --json-indent 4 --import-asm-json expected.asm.json)
+            if ! "${import_command[@]}" > obtained.json 2> obtained.error
+            then
+                success=0
+                printError "ERROR: (import) EVM Assembly JSON reimport failed for ${sol_file}"
+                if (( EXIT_ON_ERROR == 1 ))
+                then
+                    print_used_commands "$(pwd)" "${export_command[*]} ${optimize_param}" "${import_command[*]}"
+                    return 1
+                fi
+            fi
+
+            for output in "${outputs[@]}"
+            do
+                for obtained_contract in $(jq '.contracts | keys | .[]' obtained.json  2> /dev/null)
+                do
+                    jq --raw-output ".contracts.${obtained_contract}.\"${output}\"" obtained.json > "obtained.${output}.json"
+                    # compare expected and obtained evm assembly json
+                    if ! diff_files "expected.${output}.json" "obtained.${output}.json"
+                    then
+                        success=0
+                        printError "ERROR: (${output}) EVM Assembly JSON reimport failed for ${sol_file}"
+                        if (( EXIT_ON_ERROR == 1 ))
+                        then
+                            print_used_commands "$(pwd)" "${export_command[*]} ${optimize_param}" "${import_command[*]}"
+                            return 1
+                        fi
+                    fi
+                done
+            done
+
+            # direct export via --asm-json, if imported with --import-asm-json.
+            if ! "${SOLC}" --asm-json --import-asm-json expected.asm.json --pretty-json --json-indent 4 | tail -n+2 > obtained_direct_import_export.json 2> obtained_direct_import_export.error
+            then
+                success=0
+                printError "ERROR: (direct) EVM Assembly JSON reimport failed for ${sol_file}"
+                if (( EXIT_ON_ERROR == 1 ))
+                then
+                    print_used_commands "$(pwd)" "${SOLC} --asm-json --import-asm-json expected.asm.json --pretty-json --json-indent 4 | tail -n+4" ""
+                    return 1
+                fi
+            fi
+
+            # reformat json files using jq.
+            jq . expected.asm.json > expected.asm.json.pretty
+            jq . obtained_direct_import_export.json > obtained_direct_import_export.json.pretty
+
+            # compare expected and obtained evm assembly.
+            if ! diff_files expected.asm.json.pretty obtained_direct_import_export.json.pretty
+            then
+                success=0
+                printError "ERROR: EVM Assembly JSON reimport failed for ${sol_file}"
+                if (( EXIT_ON_ERROR == 1 ))
+                then
+                    print_used_commands "$(pwd)" "${export_command[*]} ${optimize_param}" "${import_command[*]}"
+                    return 1
+                fi
+            fi
+        done
+    done
+
+    if (( success == 1 ))
+    then
+        TESTED=$((TESTED + 1))
+    else
+        FAILED=$((FAILED + 1))
+    fi
+}
+
 # function tests whether exporting and importing again is equivalent.
 # Results are recorded by incrementing the FAILED or UNCOMPILABLE global variable.
 # Also, in case of a mismatch a diff is printed
@@ -168,6 +301,7 @@ function test_import_export_equivalence {
 
     case "$IMPORT_TEST_TYPE" in
         ast) compile_test="--ast-compact-json" ;;
+        evm-assembly) compile_test="--bin" ;;
         *) assertFail "Unknown import test type '${IMPORT_TEST_TYPE}'. Aborting." ;;
     esac
 
@@ -181,6 +315,7 @@ function test_import_export_equivalence {
     then
         case "$IMPORT_TEST_TYPE" in
             ast) test_ast_import_export_equivalence "${sol_file}" "${input_files[@]}" ;;
+            evm-assembly) test_evmjson_import_export_equivalence "${sol_file}" "${input_files[@]}" ;;
             *) assertFail "Unknown import test type '${IMPORT_TEST_TYPE}'. Aborting." ;;
         esac
     else
@@ -190,10 +325,23 @@ function test_import_export_equivalence {
         # This should normally not happen, so we terminate the test execution here
         # and print some details about the corresponding solc invocation.
         if (( solc_return_code == 2 ))
+        then
+            # For the evm-assembly import/export tests, this script uses only the
+            # old code generator. Some semantic test can only be compiled with
+            # --via-ir (some need to be additionally compiled with --optimize).
+            # The tests that are meant to be compiled with --via-ir are throwing
+            # an UnimplementedFeatureError exception (e.g. Copying of type struct C.S
+            # memory[] memory to storage not yet supported, Copying nested calldata
+            # dynamic arrays to storage is not implemented in the old code generator.)
+            # We will just ignore these kind of exceptions for now.
+            # However, any other exception will be treated as a fatal error and the
+            # script execution will be terminated with an error.
+            if [[ "$output" != *"UnimplementedFeatureError"* ]]
             then
                 fail "\n\nERROR: Uncaught Exception while executing '$SOLC ${compile_test} ${input_files[*]}':\n${output}\n"
             fi
         fi
+    fi
 }
 
 WORKINGDIR=$PWD
@@ -203,6 +351,7 @@ command_available jq --version
 
 case "$IMPORT_TEST_TYPE" in
     ast) TEST_DIRS=("${SYNTAXTESTS_DIR}" "${ASTJSONTESTS_DIR}") ;;
+    evm-assembly) TEST_DIRS=("${SEMANTICTESTS_DIR}") ;;
     *) assertFail "Import test type not defined. $(print_usage)" ;;
 esac
 

solc/CommandLineInterface.cpp

@@ -93,6 +93,7 @@ namespace
 set<frontend::InputMode> const CompilerInputModes{
 	frontend::InputMode::Compiler,
 	frontend::InputMode::CompilerWithASTImport,
+	frontend::InputMode::EVMAssemblerJSON
 };
 
 } // anonymous namespace
@@ -167,28 +168,58 @@ static bool coloredOutput(CommandLineOptions const& _options)
 		(_options.formatting.coloredOutput.has_value() && _options.formatting.coloredOutput.value());
 }
 
+void CommandLineInterface::handleEVMAssembly(string const& _contract)
+{
+	solAssert(CompilerInputModes.count(m_options.input.mode) == 1);
+
+	if (m_options.compiler.outputs.asm_ || m_options.compiler.outputs.asmJson)
+	{
+		string assembly;
+		if (m_options.compiler.outputs.asmJson)
+			assembly = util::jsonPrint(removeNullMembers(m_compiler->assemblyJSON(_contract)), m_options.formatting.json);
+		else
+			assembly = m_compiler->assemblyString(_contract, m_fileReader.sourceUnits());
+
+		if (!m_options.output.dir.empty())
+			createFile(
+				m_compiler->filesystemFriendlyName(_contract) +
+				(m_options.compiler.outputs.asmJson ? "_evm.json" : ".evm"),
+				assembly
+			);
+		else
+			sout() << "EVM assembly:" << endl << assembly << endl;
+	}
+}
+
 void CommandLineInterface::handleBinary(string const& _contract)
 {
 	solAssert(CompilerInputModes.count(m_options.input.mode) == 1);
 
+	string binary;
+	string binaryRuntime;
+	if (m_options.compiler.outputs.binary)
+		binary = objectWithLinkRefsHex(m_compiler->object(_contract));
+	if (m_options.compiler.outputs.binaryRuntime)
+		binaryRuntime = objectWithLinkRefsHex(m_compiler->runtimeObject(_contract));
+
 	if (m_options.compiler.outputs.binary)
 	{
 		if (!m_options.output.dir.empty())
-			createFile(m_compiler->filesystemFriendlyName(_contract) + ".bin", objectWithLinkRefsHex(m_compiler->object(_contract)));
+			createFile(m_compiler->filesystemFriendlyName(_contract) + ".bin", binary);
 		else
 		{
 			sout() << "Binary:" << endl;
-			sout() << objectWithLinkRefsHex(m_compiler->object(_contract)) << endl;
+			sout() << binary << endl;
 		}
 	}
 	if (m_options.compiler.outputs.binaryRuntime)
 	{
 		if (!m_options.output.dir.empty())
-			createFile(m_compiler->filesystemFriendlyName(_contract) + ".bin-runtime", objectWithLinkRefsHex(m_compiler->runtimeObject(_contract)));
+			createFile(m_compiler->filesystemFriendlyName(_contract) + ".bin-runtime", binaryRuntime);
 		else
 		{
 			sout() << "Binary of the runtime part:" << endl;
-			sout() << objectWithLinkRefsHex(m_compiler->runtimeObject(_contract)) << endl;
+			sout() << binaryRuntime << endl;
 		}
 	}
 }
@@ -197,12 +228,14 @@ void CommandLineInterface::handleOpcode(string const& _contract)
 {
 	solAssert(CompilerInputModes.count(m_options.input.mode) == 1);
 
+	string opcodes{evmasm::disassemble(m_compiler->object(_contract).bytecode, m_options.output.evmVersion)};
+
 	if (!m_options.output.dir.empty())
-		createFile(m_compiler->filesystemFriendlyName(_contract) + ".opcode", evmasm::disassemble(m_compiler->object(_contract).bytecode, m_options.output.evmVersion));
+		createFile(m_compiler->filesystemFriendlyName(_contract) + ".opcode", opcodes);
 	else
 	{
 		sout() << "Opcodes:" << endl;
-		sout() << uppercase << evmasm::disassemble(m_compiler->object(_contract).bytecode, m_options.output.evmVersion);
+		sout() << uppercase << opcodes;
 		sout() << endl;
 	}
 }
@@ -702,6 +735,12 @@ void CommandLineInterface::processInput()
 		compile();
 		outputCompilationResults();
 		break;
+	case InputMode::EVMAssemblerJSON:
+		assembleFromEvmAssemblyJson();
+		handleCombinedJSON();
+		handleBytecode("");
+		handleEVMAssembly("");
+		break;
 	}
 }
 
@@ -718,6 +757,23 @@ void CommandLineInterface::printLicense()
 	sout() << licenseText << endl;
 }
 
+void CommandLineInterface::assembleFromEvmAssemblyJson()
+{
+	solAssert(m_options.input.mode == InputMode::EVMAssemblerJSON);
+	try
+	{
+		solAssert(m_fileReader.sourceUnits().size() == 1);
+		auto const iter = m_fileReader.sourceUnits().begin();
+		m_compiler = make_unique<CompilerStack>(m_universalCallback.callback());
+		m_compiler->importFromEVMAssemblyStack(iter->first, iter->second);
+	}
+	catch (evmasm::AssemblyImportException const& _exception)
+	{
+		solThrow(CommandLineExecutionError, "Assembly Import Error: "s + _exception.what());
+	}
+	solRequire(m_compiler != nullptr, CommandLineExecutionError, "Assembly Import Error: Could not create compiler object.");
+}
+
 void CommandLineInterface::compile()
 {
 	solAssert(CompilerInputModes.count(m_options.input.mode) == 1);
@@ -884,8 +940,7 @@ void CommandLineInterface::handleCombinedJSON()
 			);
 		if (m_options.compiler.combinedJsonRequests->funDebugRuntime && m_compiler->compilationSuccessful())
 			contractData[g_strFunDebugRuntime] = StandardCompiler::formatFunctionDebugData(
-				m_compiler->runtimeObject(contractName).functionDebugData
-			);
+				m_compiler->runtimeObject(contractName).functionDebugData);
 		if (m_options.compiler.combinedJsonRequests->signatureHashes)
 			contractData[g_strSignatureHashes] = m_compiler->interfaceSymbols(contractName)["methods"];
 		if (m_options.compiler.combinedJsonRequests->natspecDev)
@@ -894,8 +949,7 @@ void CommandLineInterface::handleCombinedJSON()
 			contractData[g_strNatspecUser] = m_compiler->natspecUser(contractName);
 	}
 
-	bool needsSourceList =
-		m_options.compiler.combinedJsonRequests->ast ||
+	bool needsSourceList = m_options.compiler.combinedJsonRequests->ast ||
 							m_options.compiler.combinedJsonRequests->srcMap ||
 							m_options.compiler.combinedJsonRequests->srcMapRuntime;
 	if (needsSourceList)
@@ -1152,6 +1206,53 @@ void CommandLineInterface::assembleYul(yul::YulStack::Language _language, yul::Y
 			else
 				serr() << "No text representation found." << endl;
 		}
+
+		if (m_options.compiler.outputs.asmJson)
+		{
+			shared_ptr<evmasm::Assembly> assembly{stack.assembleEVMWithDeployed().first};
+			if (assembly)
+			{
+				std::function<map<string, unsigned>(yul::Object const&)> collectSourceIndices =
+					[&](yul::Object const& _object) -> map<string, unsigned> {
+					map<string, unsigned> sourceIndices;
+					if (_object.debugData && _object.debugData->sourceNames.has_value())
+						for (auto const& iter: *_object.debugData->sourceNames)
+							sourceIndices[*iter.second] = iter.first;
+					for (auto const& sub: _object.subObjects)
+					{
+						auto subObject = dynamic_cast<yul::Object const*>(sub.get());
+						if (subObject)
+							for (auto const& [name, index]: collectSourceIndices(*subObject))
+								sourceIndices[name] = index;
+					}
+					return sourceIndices;
+				};
+				if (stack.parserResult() && stack.parserResult()->debugData)
+				{
+					map<string, unsigned> sourceIndices = collectSourceIndices(*stack.parserResult());
+					// if sourceIndices are empty here, there were no source locations annotated in the yul source.
+					// in this case, we just add the filename of the yul file itself here.
+					if (sourceIndices.empty())
+						sourceIndices[src.first] = 0;
+					size_t max_index = 0;
+					for (auto const& [name, index]: sourceIndices)
+						if (max_index < index)
+							max_index = index;
+					vector<string> sourceList(max_index + 1, "<unknown>");
+					for (auto const& [name, index]: sourceIndices)
+						sourceList[index] = name;
+					assembly->setSourceList(sourceList);
+					sout() << util::jsonPrint(
+						removeNullMembers(
+							assembly->assemblyJSON()
+						),
+						m_options.formatting.json
+					) << endl;
+					return;
+				}
+			}
+			serr() << "Could not create Assembly JSON representation." << endl;
+		}
 	}
 }
 
@@ -1180,19 +1281,7 @@ void CommandLineInterface::outputCompilationResults()
 			sout() << endl << "======= " << contract << " =======" << endl;
 
 		// do we need EVM assembly?
-		if (m_options.compiler.outputs.asm_ || m_options.compiler.outputs.asmJson)
-		{
-			string ret;
-			if (m_options.compiler.outputs.asmJson)
-				ret = util::jsonPrint(removeNullMembers(m_compiler->assemblyJSON(contract)), m_options.formatting.json);
-			else
-				ret = m_compiler->assemblyString(contract, m_fileReader.sourceUnits());
-
-			if (!m_options.output.dir.empty())
-				createFile(m_compiler->filesystemFriendlyName(contract) + (m_options.compiler.outputs.asmJson ? "_evm.json" : ".evm"), ret);
-			else
-				sout() << "EVM assembly:" << endl << ret << endl;
-		}
+		handleEVMAssembly(contract);
 
 		if (m_options.compiler.estimateGas)
 			handleGasEstimation(contract);

solc/CommandLineInterface.h

@@ -24,6 +24,7 @@
 
 #include <solc/CommandLineParser.h>
 
+#include <libevmasm/EVMAssemblyStack.h>
 #include <libsolidity/interface/CompilerStack.h>
 #include <libsolidity/interface/DebugSettings.h>
 #include <libsolidity/interface/FileReader.h>
@@ -84,6 +85,7 @@ private:
 	void printVersion();
 	void printLicense();
 	void compile();
+	void assembleFromEvmAssemblyJson();
 	void serveLSP();
 	void link();
 	void writeLinkedFiles();
@@ -98,6 +100,7 @@ private:
 
 	void handleCombinedJSON();
 	void handleAst();
+	void handleEVMAssembly(std::string const& _contract);
 	void handleBinary(std::string const& _contract);
 	void handleOpcode(std::string const& _contract);
 	void handleIR(std::string const& _contract);

solc/CommandLineParser.cpp

@@ -52,6 +52,7 @@ static string const g_strExperimentalViaIR = "experimental-via-ir";
 static string const g_strGas = "gas";
 static string const g_strHelp = "help";
 static string const g_strImportAst = "import-ast";
+static string const g_strImportEvmAssemblerJson = "import-asm-json";
 static string const g_strInputFile = "input-file";
 static string const g_strYul = "yul";
 static string const g_strYulDialect = "yul-dialect";
@@ -142,6 +143,7 @@ static map<InputMode, string> const g_inputModeName = {
 	{InputMode::StandardJson, "standard JSON"},
 	{InputMode::Linker, "linker"},
 	{InputMode::LanguageServer, "language server (LSP)"},
+	{InputMode::EVMAssemblerJSON, "EVM assembler (JSON format)"},
 };
 
 void CommandLineParser::checkMutuallyExclusive(vector<string> const& _optionNames)
@@ -468,8 +470,15 @@ void CommandLineParser::parseOutputSelection()
 			CompilerOutputs::componentName(&CompilerOutputs::binary),
 			CompilerOutputs::componentName(&CompilerOutputs::irOptimized),
 			CompilerOutputs::componentName(&CompilerOutputs::astCompactJson),
+			CompilerOutputs::componentName(&CompilerOutputs::asmJson),
+		};
+		static set<string> const evmAssemblyJsonImportModeOutputs = {
+			CompilerOutputs::componentName(&CompilerOutputs::asm_),
+			CompilerOutputs::componentName(&CompilerOutputs::binary),
+			CompilerOutputs::componentName(&CompilerOutputs::binaryRuntime),
+			CompilerOutputs::componentName(&CompilerOutputs::opcodes),
+			CompilerOutputs::componentName(&CompilerOutputs::asmJson),
 		};
-
 		switch (_mode)
 		{
 		case InputMode::Help:
@@ -480,6 +489,8 @@ void CommandLineParser::parseOutputSelection()
 		case InputMode::Compiler:
 		case InputMode::CompilerWithASTImport:
 			return util::contains(compilerModeOutputs, _outputName);
+		case InputMode::EVMAssemblerJSON:
+			return util::contains(evmAssemblyJsonImportModeOutputs, _outputName);
 		case InputMode::Assembler:
 			return util::contains(assemblerModeOutputs, _outputName);
 		case InputMode::StandardJson:
@@ -664,6 +675,10 @@ General Information)").c_str(),
 			"Supported Inputs is the output of the --" + g_strStandardJSON + " or the one produced by "
 			"--" + g_strCombinedJson + " " + CombinedJsonRequests::componentName(&CombinedJsonRequests::ast)).c_str()
 		)
+		(
+			g_strImportEvmAssemblerJson.c_str(),
+			"Import EVM assembly from JSON. Assumes input is in the format used by --asm-json."
+		)
 		(
 			g_strLSP.c_str(),
 			"Switch to language server mode (\"LSP\"). Allows the compiler to be used as an analysis backend "
@@ -932,6 +947,7 @@ void CommandLineParser::processArgs()
 		g_strYul,
 		g_strImportAst,
 		g_strLSP,
+		g_strImportEvmAssemblerJson,
 	});
 
 	if (m_args.count(g_strHelp) > 0)
@@ -950,6 +966,8 @@ void CommandLineParser::processArgs()
 		m_options.input.mode = InputMode::Linker;
 	else if (m_args.count(g_strImportAst) > 0)
 		m_options.input.mode = InputMode::CompilerWithASTImport;
+	else if (m_args.count(g_strImportEvmAssemblerJson) > 0)
+		m_options.input.mode = InputMode::EVMAssemblerJSON;
 	else
 		m_options.input.mode = InputMode::Compiler;
 
@@ -1010,9 +1028,35 @@ void CommandLineParser::processArgs()
 		if (option != CompilerOutputs::componentName(&CompilerOutputs::astCompactJson))
 			checkMutuallyExclusive({g_strStopAfter, option});
 
+	if (m_options.input.mode == InputMode::EVMAssemblerJSON)
+	{
+		static set<string> const supportedByEvmAsmJsonImport{
+			g_strImportEvmAssemblerJson,
+			CompilerOutputs::componentName(&CompilerOutputs::asm_),
+			CompilerOutputs::componentName(&CompilerOutputs::binary),
+			CompilerOutputs::componentName(&CompilerOutputs::binaryRuntime),
+			CompilerOutputs::componentName(&CompilerOutputs::asmJson),
+			CompilerOutputs::componentName(&CompilerOutputs::opcodes),
+			g_strCombinedJson,
+			g_strInputFile,
+			g_strJsonIndent,
+			g_strPrettyJson,
+			"srcmap",
+			"srcmap-runtime",
+		};
+
+		for (auto const& option: m_args)
+			if (!option.second.defaulted() && !supportedByEvmAsmJsonImport.count(option.first))
+				solThrow(
+					CommandLineValidationError,
+					"Option --" + option.first + " is not supported with --"+g_strImportEvmAssemblerJson+"."
+				);
+	}
+
 	if (
 		m_options.input.mode != InputMode::Compiler &&
 		m_options.input.mode != InputMode::CompilerWithASTImport &&
+		m_options.input.mode != InputMode::EVMAssemblerJSON &&
 		m_options.input.mode != InputMode::Assembler
 	)
 	{
@@ -1372,7 +1416,10 @@ void CommandLineParser::processArgs()
 	if (m_options.input.mode == InputMode::Compiler)
 		m_options.input.errorRecovery = (m_args.count(g_strErrorRecovery) > 0);
 
-	solAssert(m_options.input.mode == InputMode::Compiler || m_options.input.mode == InputMode::CompilerWithASTImport);
+	solAssert(
+		m_options.input.mode == InputMode::Compiler ||
+		m_options.input.mode == InputMode::CompilerWithASTImport ||
+		m_options.input.mode == InputMode::EVMAssemblerJSON);
 }
 
 void CommandLineParser::parseCombinedJsonOption()
@@ -1388,6 +1435,29 @@ void CommandLineParser::parseCombinedJsonOption()
 	m_options.compiler.combinedJsonRequests = CombinedJsonRequests{};
 	for (auto&& [componentName, component]: CombinedJsonRequests::componentMap())
 		m_options.compiler.combinedJsonRequests.value().*component = (requests.count(componentName) > 0);
+
+	if (m_options.input.mode == InputMode::EVMAssemblerJSON && m_options.compiler.combinedJsonRequests.has_value())
+	{
+		static bool CombinedJsonRequests::* invalidOptions[]{
+			&CombinedJsonRequests::abi,
+			&CombinedJsonRequests::ast,
+			&CombinedJsonRequests::funDebug,
+			&CombinedJsonRequests::funDebugRuntime,
+			&CombinedJsonRequests::generatedSources,
+			&CombinedJsonRequests::generatedSourcesRuntime,
+			&CombinedJsonRequests::metadata,
+			&CombinedJsonRequests::natspecDev,
+			&CombinedJsonRequests::natspecUser,
+			&CombinedJsonRequests::signatureHashes,
+			&CombinedJsonRequests::storageLayout
+		};
+
+		for (auto const invalidOption: invalidOptions)
+			if (m_options.compiler.combinedJsonRequests.value().*invalidOption)
+				solThrow(
+					CommandLineValidationError,
+					"Invalid option to --" + g_strCombinedJson + ": " + CombinedJsonRequests::componentName(invalidOption) + " for --" + g_strImportEvmAssemblerJson);
+	}
 }
 
 size_t CommandLineParser::countEnabledOptions(vector<string> const& _optionNames) const

solc/CommandLineParser.h

@@ -57,6 +57,7 @@ enum class InputMode
 	Linker,
 	Assembler,
 	LanguageServer,
+	EVMAssemblerJSON
 };
 
 struct CompilerOutputs

test/cmdlineTests/asm_json_import_all_valid_flags/args

@@ -0,0 +1 @@
+--pretty-json --json-indent 4 --combined-json bin,bin-runtime,opcodes,asm,srcmap,srcmap-runtime --asm --bin --bin-runtime --asm-json --import-asm-json -

test/cmdlineTests/asm_json_import_all_valid_flags/output

@@ -0,0 +1,91 @@
+{
+    "contracts":
+    {
+        "<stdin>":
+        {
+            "asm":
+            {
+                ".code":
+                [
+                    {
+                        "begin": 0,
+                        "end": 0,
+                        "name": "PUSH",
+                        "source": -1,
+                        "value": "0"
+                    }
+                ],
+                ".data":
+                {
+                    "0":
+                    {
+                        ".code":
+                        [
+                            {
+                                "begin": 0,
+                                "end": 0,
+                                "name": "PUSH",
+                                "source": -1,
+                                "value": "1"
+                            }
+                        ]
+                    }
+                },
+                "sourceList":
+                [
+                    "contract.sol",
+                    "#utility.yul"
+                ]
+            },
+            "bin": "6000fe",
+            "bin-runtime": "6001",
+            "opcodes": "PUSH1 0x0 INVALID ",
+            "srcmap": "0:0::-:0",
+            "srcmap-runtime": "0:0::-:0"
+        }
+    },
+    "sourceList":
+    [
+        "contract.sol",
+        "#utility.yul"
+    ],
+    "version": "<VERSION REMOVED>"
+}
+Binary:
+6000fe
+Binary of the runtime part:
+6001
+EVM assembly:
+{
+    ".code":
+    [
+        {
+            "begin": 0,
+            "end": 0,
+            "name": "PUSH",
+            "source": -1,
+            "value": "0"
+        }
+    ],
+    ".data":
+    {
+        "0":
+        {
+            ".code":
+            [
+                {
+                    "begin": 0,
+                    "end": 0,
+                    "name": "PUSH",
+                    "source": -1,
+                    "value": "1"
+                }
+            ]
+        }
+    },
+    "sourceList":
+    [
+        "contract.sol",
+        "#utility.yul"
+    ]
+}

test/cmdlineTests/asm_json_import_all_valid_flags/stdin

@@ -0,0 +1,22 @@
+{
+  ".code": [
+    {
+      "name": "PUSH",
+      "value": "0"
+    }
+  ],
+  ".data": {
+    "0": {
+      ".code": [
+        {
+          "name": "PUSH",
+          "value": "1"
+        }
+      ]
+    }
+  },
+  "sourceList": [
+    "contract.sol",
+    "#utility.yul"
+  ]
+}

test/cmdlineTests/asm_json_import_difficulty_prevrandao/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_difficulty_prevrandao/output

@@ -0,0 +1,6 @@
+Opcodes:
+PREVRANDAO PREVRANDAO 
+EVM assembly:
+    /*   */
+  prevrandao
+  prevrandao

test/cmdlineTests/asm_json_import_difficulty_prevrandao/stdin

@@ -0,0 +1,6 @@
+{
+  ".code": [
+    { "name": "DIFFICULTY" },
+    { "name": "PREVRANDAO" }
+  ]
+}

test/cmdlineTests/asm_json_import_invalid_data_no_hex/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_invalid_data_no_hex/err

@@ -0,0 +1 @@
+Assembly Import Error: Member '.data' contains a value for '0' that is not a valid hexadecimal string.

test/cmdlineTests/asm_json_import_invalid_data_no_hex/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_invalid_data_no_hex/stdin

@@ -0,0 +1,11 @@
+{
+  ".code": [
+    {
+      "name": "PUSH",
+      "value": "0"
+    }
+  ],
+  ".data": {
+    "0": "no-hex-string"
+  }
+}

test/cmdlineTests/asm_json_import_invalid_data_no_object/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_invalid_data_no_object/err

@@ -0,0 +1 @@
+Assembly Import Error: Key inside '.data' '0' can only be a valid hex-string or an object.

test/cmdlineTests/asm_json_import_invalid_data_no_object/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_invalid_data_no_object/stdin

@@ -0,0 +1,16 @@
+{
+  ".code": [
+    {
+      "name": "PUSH",
+      "value": "0"
+    }
+  ],
+  ".data": {
+    "0": [
+      0,
+      1,
+      2,
+      3
+    ]
+  }
+}

test/cmdlineTests/asm_json_import_invalid_jumptype_instruction/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_invalid_jumptype_instruction/err

@@ -0,0 +1 @@
+Assembly Import Error: Member 'jumpType' set on instruction different from JUMP or JUMPI (was set on instruction 'PUSH')

test/cmdlineTests/asm_json_import_invalid_jumptype_instruction/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_invalid_jumptype_instruction/stdin

@@ -0,0 +1,9 @@
+{
+  ".code": [
+    {
+      "name": "PUSH",
+      "value": "0",
+      "jumpType": "[in]"
+    }
+  ]
+}

test/cmdlineTests/asm_json_import_invalid_value/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_invalid_value/err

@@ -0,0 +1 @@
+Assembly Import Error: Member 'value' defined for instruction 'DIFFICULTY', but the instruction does not need a value.

test/cmdlineTests/asm_json_import_invalid_value/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_invalid_value/stdin

@@ -0,0 +1,8 @@
+{
+  ".code": [
+    {
+      "name": "DIFFICULTY",
+      "value": "0"
+    }
+  ]
+}

test/cmdlineTests/asm_json_import_no_optimiser/args

@@ -0,0 +1 @@
+--optimize --import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_no_optimiser/err

@@ -0,0 +1 @@
+Option --optimize is not supported with --import-asm-json.

test/cmdlineTests/asm_json_import_no_optimiser/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_no_optimiser/stdin

@@ -0,0 +1,6 @@
+{
+  ".code": [
+    { "name": "DIFFICULTY" },
+    { "name": "PREVRANDAO" }
+  ]
+}

test/cmdlineTests/asm_json_import_no_value/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_no_value/err

@@ -0,0 +1 @@
+Assembly Import Error: Member 'value' was not defined for instruction 'PUSH', but the instruction needs a value.

test/cmdlineTests/asm_json_import_no_value/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_no_value/stdin

@@ -0,0 +1,5 @@
+{
+  ".code": [
+    { "name": "PUSH" }
+  ]
+}

test/cmdlineTests/asm_json_import_other_fields/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_other_fields/err

@@ -0,0 +1 @@
+Assembly Import Error: Unknown member '_name'. Valid members are begin, end, jumpType, modifierDepth, name, source, value.

test/cmdlineTests/asm_json_import_other_fields/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_other_fields/stdin

@@ -0,0 +1,5 @@
+{
+  ".code": [
+    { "_name": "DIFFICULTY" }
+  ]
+}

test/cmdlineTests/asm_json_import_untagged_jumpdest/args

@@ -0,0 +1 @@
+--import-asm-json - --opcodes --asm

test/cmdlineTests/asm_json_import_untagged_jumpdest/err

@@ -0,0 +1 @@
+Assembly Import Error: JUMPDEST instruction found that was not followed by tag.

test/cmdlineTests/asm_json_import_untagged_jumpdest/exit

@@ -0,0 +1 @@
+1

test/cmdlineTests/asm_json_import_untagged_jumpdest/stdin

@@ -0,0 +1,10 @@
+{
+  ".code": [
+    {
+      "name": "tag",
+      "value": "0x00"
+    },
+    { "name": "JUMPDEST" },
+    { "name": "JUMPDEST" }
+  ]
+}

test/cmdlineTests/strict_asm_output_selection_invalid/err

@@ -1 +1 @@
-The following outputs are not supported in assembler mode: --abi, --asm-json, --bin-runtime, --devdoc, --hashes, --ir, --metadata, --opcodes, --storage-layout, --userdoc.
+The following outputs are not supported in assembler mode: --abi, --bin-runtime, --devdoc, --hashes, --ir, --metadata, --opcodes, --storage-layout, --userdoc.

test/libevmasm/Assembler.cpp

@@ -55,13 +55,9 @@ BOOST_AUTO_TEST_SUITE(Assembler)
 
 BOOST_AUTO_TEST_CASE(all_assembly_items)
 {
-	map<string, unsigned> indices = {
-		{ "root.asm", 0 },
-		{ "sub.asm", 1 },
-		{ "verbatim.asm", 2 }
-	};
 	EVMVersion evmVersion = solidity::test::CommonOptions::get().evmVersion();
 	Assembly _assembly{evmVersion, false, {}};
+	_assembly.setSourceList({"root.asm", "sub.asm", "verbatim.asm"});
 	auto root_asm = make_shared<string>("root.asm");
 	_assembly.setSourceLocation({1, 3, root_asm});
 
@@ -212,7 +208,7 @@ BOOST_AUTO_TEST_CASE(all_assembly_items)
 	};
 	Json::Value jsonValue;
 	BOOST_CHECK(util::jsonParseStrict(json, jsonValue));
-	BOOST_CHECK_EQUAL(util::jsonCompactPrint(_assembly.assemblyJSON(indices)), util::jsonCompactPrint(jsonValue));
+	BOOST_CHECK_EQUAL(util::jsonCompactPrint(_assembly.assemblyJSON()), util::jsonCompactPrint(jsonValue));
 }
 
 BOOST_AUTO_TEST_CASE(immutables_and_its_source_maps)
@@ -301,12 +297,9 @@ BOOST_AUTO_TEST_CASE(immutables_and_its_source_maps)
 
 BOOST_AUTO_TEST_CASE(immutable)
 {
-	map<string, unsigned> indices = {
-		{ "root.asm", 0 },
-		{ "sub.asm", 1 }
-	};
 	EVMVersion evmVersion = solidity::test::CommonOptions::get().evmVersion();
 	Assembly _assembly{evmVersion, true, {}};
+	_assembly.setSourceList({"root.asm", "sub.asm"});
 	auto root_asm = make_shared<string>("root.asm");
 	_assembly.setSourceLocation({1, 3, root_asm});
 
@@ -383,7 +376,7 @@ BOOST_AUTO_TEST_CASE(immutable)
 		"}\n"
 	);
 	BOOST_CHECK_EQUAL(
-		util::jsonCompactPrint(_assembly.assemblyJSON(indices)),
+		util::jsonCompactPrint(_assembly.assemblyJSON()),
 		"{\".code\":["
 		"{\"begin\":1,\"end\":3,\"name\":\"PUSH\",\"source\":0,\"value\":\"2A\"},"
 		"{\"begin\":1,\"end\":3,\"name\":\"PUSH\",\"source\":0,\"value\":\"0\"},"

test/solc/CommandLineInterface.cpp

@@ -148,7 +148,7 @@ BOOST_AUTO_TEST_CASE(version)
 
 BOOST_AUTO_TEST_CASE(multiple_input_modes)
 {
-	array<string, 9> inputModeOptions = {
+	array<string, 10> inputModeOptions = {
 		"--help",
 		"--license",
 		"--version",
@@ -158,10 +158,11 @@ BOOST_AUTO_TEST_CASE(multiple_input_modes)
 		"--strict-assembly",
 		"--yul",
 		"--import-ast",
+		"--import-asm-json",
 	};
 	string expectedMessage =
 		"The following options are mutually exclusive: "
-		"--help, --license, --version, --standard-json, --link, --assemble, --strict-assembly, --yul, --import-ast, --lsp. "
+		"--help, --license, --version, --standard-json, --link, --assemble, --strict-assembly, --yul, --import-ast, --lsp, --import-asm-json. "
 		"Select at most one.";
 
 	for (string const& mode1: inputModeOptions)