Add --asm-json option in assembler mode.

Co-authored-by: r0qs <rodrigo.saramago@ethereum.org>
Co-authored-by: Kamil Śliwak <kamil.sliwak@codepoets.it>

Changelog.md

@@ -14,6 +14,7 @@ Compiler Features:
  * SMTChecker: Replace CVC4 as a possible BMC backend with cvc5.
  * Standard JSON Interface: Do not perform IR optimization when only unoptimized IR is requested.
  * Yul Optimizer: The optimizer now treats some previously unrecognized identical literals as identical.
+ * Commandline Interface: Allow the use of ``--asm-json`` output option in assembler mode to export EVM assembly of the contracts in JSON format.
 
 
 Bugfixes:

libevmasm/Assembly.cpp

@@ -479,8 +479,18 @@ Json Assembly::assemblyJSON(std::map<std::string, unsigned> const& _sourceIndice
 	{
 		root["sourceList"] = Json::array();
 		Json& jsonSourceList = root["sourceList"];
-		for (auto const& [name, index]: _sourceIndices)
-			jsonSourceList[index] = name;
+		unsigned maxSourceIndex = 0;
+		for (auto const& [sourceName, sourceIndex]: _sourceIndices)
+		{
+			maxSourceIndex = std::max(sourceIndex, maxSourceIndex);
+			jsonSourceList[sourceIndex] = sourceName;
+		}
+		solAssert(maxSourceIndex + 1 >= _sourceIndices.size());
+		solRequire(
+			_sourceIndices.size() == 0 || _sourceIndices.size() == maxSourceIndex + 1,
+			AssemblyImportException,
+			"The 'sourceList' array contains invalid 'null' item."
+		);
 	}
 
 	if (!m_data.empty() || !m_subs.empty())
@@ -522,7 +532,14 @@ std::pair<std::shared_ptr<Assembly>, std::vector<std::string>> Assembly::fromJSO
 		{
 			solRequire(_json["sourceList"].is_array(), AssemblyImportException, "Optional member 'sourceList' is not an array.");
 			for (Json const& sourceName: _json["sourceList"])
-				solRequire(sourceName.is_string(), AssemblyImportException, "The 'sourceList' array contains an item that is not a string.");
+			{
+				solRequire(!sourceName.is_null(), AssemblyImportException, "The 'sourceList' array contains invalid 'null' item.");
+				solRequire(
+					sourceName.is_string(),
+					AssemblyImportException,
+					"The 'sourceList' array contains an item that is not a string."
+				);
+			}
 		}
 	}
 	else

libsolidity/interface/StandardCompiler.cpp

@@ -1694,7 +1694,7 @@ Json StandardCompiler::compileYul(InputsAndSettings _inputsAndSettings)
 	if (isArtifactRequested(_inputsAndSettings.outputSelection, sourceName, contractName, "irOptimized", wildcardMatchesExperimental))
 		output["contracts"][sourceName][contractName]["irOptimized"] = stack.print();
 	if (isArtifactRequested(_inputsAndSettings.outputSelection, sourceName, contractName, "evm.assembly", wildcardMatchesExperimental))
-		output["contracts"][sourceName][contractName]["evm"]["assembly"] = object.assembly;
+		output["contracts"][sourceName][contractName]["evm"]["assembly"] = object.assembly->assemblyString(stack.debugInfoSelection());
 
 	return output;
 }

libyul/Object.cpp

@@ -160,3 +160,33 @@ std::vector<size_t> Object::pathToSubObject(std::string_view _qualifiedName) con
 
 	return path;
 }
+
+void Object::collectSourceIndices(std::map<std::string, unsigned>& _indices) const
+{
+	if (debugData && debugData->sourceNames.has_value())
+		for (auto const& [sourceIndex, sourceName]: debugData->sourceNames.value())
+		{
+			solAssert(_indices.count(*sourceName) == 0 || _indices[*sourceName] == sourceIndex);
+			_indices[*sourceName] = sourceIndex;
+		}
+	for (std::shared_ptr<ObjectNode> const& subNode: subObjects)
+		if (auto subObject = dynamic_cast<Object*>(subNode.get()))
+			subObject->collectSourceIndices(_indices);
+}
+
+bool Object::hasContiguousSourceIndices() const
+{
+	std::map<std::string, unsigned> sourceIndices;
+	collectSourceIndices(sourceIndices);
+
+	unsigned maxSourceIndex = 0;
+	std::set<unsigned> indices;
+	for (auto const& [sources, sourceIndex]: sourceIndices)
+	{
+		maxSourceIndex = std::max(sourceIndex, maxSourceIndex);
+		indices.insert(sourceIndex);
+	}
+
+	solAssert(maxSourceIndex + 1 >= indices.size());
+	return indices.size() == 0 || indices.size() == maxSourceIndex + 1;
+}

libyul/Object.h

@@ -129,6 +129,14 @@ struct Object: public ObjectNode
 
 	std::shared_ptr<ObjectDebugData const> debugData;
 
+	/// Collects names of all Solidity source units present in the debug data
+	/// of the Yul object (including sub-objects) and their assigned indices.
+	/// @param _indices map that will be filled with source indices of the current Yul object & its sub-objects.
+	void collectSourceIndices(std::map<std::string, unsigned>& _indices) const;
+
+	/// @returns true, if the range of source indices starts at zero and is contiguous, false otherwise.
+	bool hasContiguousSourceIndices() const;
+
 	/// @returns the name of the special metadata data object.
 	static std::string metadataName() { return ".metadata"; }
 };

libyul/YulStack.cpp

@@ -15,10 +15,6 @@
 	along with solidity.  If not, see <http://www.gnu.org/licenses/>.
 */
 // SPDX-License-Identifier: GPL-3.0
-/**
- * Full assembly stack that can support EVM-assembly and Yul as input and EVM, EVM1.5
- */
-
 
 #include <libyul/YulStack.h>
 
@@ -64,7 +60,6 @@ Dialect const& languageToDialect(YulStack::Language _language, EVMVersion _versi
 
 }
 
-
 CharStream const& YulStack::charStream(std::string const& _sourceName) const
 {
 	yulAssert(m_charStream, "");
@@ -274,7 +269,7 @@ YulStack::assembleWithDeployed(std::optional<std::string_view> _deployName)
 	{
 		creationObject.bytecode = std::make_shared<evmasm::LinkerObject>(creationAssembly->assemble());
 		yulAssert(creationObject.bytecode->immutableReferences.empty(), "Leftover immutables.");
-		creationObject.assembly = creationAssembly->assemblyString(m_debugInfoSelection);
+		creationObject.assembly = creationAssembly;
 		creationObject.sourceMappings = std::make_unique<std::string>(
 			evmasm::AssemblyItem::computeSourceMapping(
 				creationAssembly->items(),
@@ -285,7 +280,7 @@ YulStack::assembleWithDeployed(std::optional<std::string_view> _deployName)
 		if (deployedAssembly)
 		{
 			deployedObject.bytecode = std::make_shared<evmasm::LinkerObject>(deployedAssembly->assemble());
-			deployedObject.assembly = deployedAssembly->assemblyString(m_debugInfoSelection);
+			deployedObject.assembly = deployedAssembly;
 			deployedObject.sourceMappings = std::make_unique<std::string>(
 				evmasm::AssemblyItem::computeSourceMapping(
 					deployedAssembly->items(),

libyul/YulStack.h

@@ -16,7 +16,7 @@
 */
 // SPDX-License-Identifier: GPL-3.0
 /**
- * Full assembly stack that can support EVM-assembly and Yul as input and EVM.
+ * Full assembly stack that can support Yul as input.
  */
 
 #pragma once
@@ -55,7 +55,7 @@ class AbstractAssembly;
 struct MachineAssemblyObject
 {
 	std::shared_ptr<evmasm::LinkerObject> bytecode;
-	std::string assembly;
+	std::shared_ptr<evmasm::Assembly> assembly;
 	std::unique_ptr<std::string> sourceMappings;
 };
 
@@ -137,9 +137,12 @@ class YulStack: public langutil::CharStreamProvider
 		langutil::CharStreamProvider const* _soliditySourceProvider = nullptr
 	) const;
 	Json astJson() const;
+
 	/// Return the parsed and analyzed object.
 	std::shared_ptr<Object> parserResult() const;
 
+	langutil::DebugInfoSelection debugInfoSelection() const { return m_debugInfoSelection; }
+
 private:
 	bool parse(std::string const& _sourceName, std::string const& _source);
 	bool analyzeParsed();
@@ -163,6 +166,8 @@ class YulStack: public langutil::CharStreamProvider
 	std::shared_ptr<yul::Object> m_parserResult;
 	langutil::ErrorList m_errors;
 	langutil::ErrorReporter m_errorReporter;
+
+	std::unique_ptr<std::string> m_sourceMappings;
 };
 
 }

solc/CommandLineInterface.cpp

@@ -177,7 +177,7 @@ void CommandLineInterface::handleEVMAssembly(std::string const& _contract)
 
 	std::string assembly;
 	if (m_options.compiler.outputs.asmJson)
-		assembly = util::jsonPrint(removeNullMembers(m_assemblyStack->assemblyJSON(_contract)), m_options.formatting.json);
+		assembly = util::jsonPrint(m_assemblyStack->assemblyJSON(_contract), m_options.formatting.json);
 	else
 		assembly = m_assemblyStack->assemblyString(_contract, m_fileReader.sourceUnits());
 
@@ -1197,6 +1197,17 @@ void CommandLineInterface::assembleYul(yul::YulStack::Language _language, yul::Y
 			successful = false;
 		else
 			stack.optimize();
+
+		if (successful && m_options.compiler.outputs.asmJson)
+		{
+			std::shared_ptr<yul::Object> result = stack.parserResult();
+			if (result && !result->hasContiguousSourceIndices())
+				solThrow(
+					CommandLineExecutionError,
+					"Generating the assembly JSON output was not possible. "
+					"Source indices provided in the @use-src annotation in the Yul input do not start at 0 or are not contiguous."
+				);
+		}
 	}
 
 	for (auto const& sourceAndStack: yulStacks)
@@ -1256,11 +1267,22 @@ void CommandLineInterface::assembleYul(yul::YulStack::Language _language, yul::Y
 		if (m_options.compiler.outputs.asm_)
 		{
 			sout() << std::endl << "Text representation:" << std::endl;
-			if (!object.assembly.empty())
-				sout() << object.assembly << std::endl;
+			std::string assemblyText{object.assembly->assemblyString(stack.debugInfoSelection())};
+			if (!assemblyText.empty())
+				sout() << assemblyText << std::endl;
 			else
 				report(Error::Severity::Info, "No text representation found.");
 		}
+		if (m_options.compiler.outputs.asmJson)
+		{
+			sout() << std::endl << "EVM assembly:" << std::endl;
+			std::map<std::string, unsigned> sourceIndices;
+			stack.parserResult()->collectSourceIndices(sourceIndices);
+			sout() << util::jsonPrint(
+				object.assembly->assemblyJSON(sourceIndices),
+				m_options.formatting.json
+			) << std::endl;
+		}
 	}
 }
 

solc/CommandLineParser.cpp

@@ -472,6 +472,7 @@ void CommandLineParser::parseOutputSelection()
 			CompilerOutputs::componentName(&CompilerOutputs::binary),
 			CompilerOutputs::componentName(&CompilerOutputs::irOptimized),
 			CompilerOutputs::componentName(&CompilerOutputs::astCompactJson),
+			CompilerOutputs::componentName(&CompilerOutputs::asmJson),
 		};
 		static std::set<std::string> const evmAssemblyJsonImportModeOutputs = {
 			CompilerOutputs::componentName(&CompilerOutputs::asm_),

test/cmdlineTests/asm_json_import_sourcelist_with_null_elements/args

@@ -0,0 +1 @@
+--pretty-json --json-indent 4 --import-asm-json -

test/cmdlineTests/asm_json_import_sourcelist_with_null_elements/err

@@ -0,0 +1 @@
+Error: Assembly Import Error: The 'sourceList' array contains invalid 'null' item.

test/cmdlineTests/asm_json_import_sourcelist_with_null_elements/exit

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

test/cmdlineTests/asm_json_import_sourcelist_with_null_elements/stdin

@@ -0,0 +1,17 @@
+{
+    ".code":
+    [
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "STOP",
+            "source": -1
+        }
+    ],
+    "sourceList":
+    [
+        null,
+        null,
+        "L0.sol"
+    ]
+}

test/cmdlineTests/asm_json_yul_export_evm_asm_import/args

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

test/cmdlineTests/asm_json_yul_export_evm_asm_import/output

@@ -0,0 +1,39 @@
+Opcodes:
+PUSH1 0x5 PUSH0 SSTORE STOP 
+Binary:
+60055f5500
+EVM assembly:
+{
+    ".code": [
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "PUSHSIZE",
+            "source": -1
+        },
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "PUSH",
+            "source": -1,
+            "value": "0"
+        },
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "SSTORE",
+            "source": -1
+        },
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "STOP",
+            "source": -1
+        }
+    ],
+    "sourceList": [
+        "L0.sol",
+        "L1.sol",
+        "L2.sol"
+    ]
+}

test/cmdlineTests/asm_json_yul_export_evm_asm_import/stdin

@@ -0,0 +1,36 @@
+{
+    ".code":
+    [
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "PUSHSIZE",
+            "source": -1
+        },
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "PUSH",
+            "source": -1,
+            "value": "0"
+        },
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "SSTORE",
+            "source": -1
+        },
+        {
+            "begin": -1,
+            "end": -1,
+            "name": "STOP",
+            "source": -1
+        }
+    ],
+    "sourceList":
+    [
+        "L0.sol",
+        "L1.sol",
+        "L2.sol"
+    ]
+}

test/cmdlineTests/strict_asm_asm_json_arbitrary_source_index/args

@@ -0,0 +1 @@
+--strict-assembly - --asm-json --pretty-json --json-indent 4

test/cmdlineTests/strict_asm_asm_json_arbitrary_source_index/err

@@ -0,0 +1 @@
+Error: Generating the assembly JSON output was not possible. Source indices provided in the @use-src annotation in the Yul input do not start at 0 or are not contiguous.

test/cmdlineTests/strict_asm_asm_json_arbitrary_source_index/exit

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

test/cmdlineTests/strict_asm_asm_json_arbitrary_source_index/stdin

@@ -0,0 +1,6 @@
+/// @use-src 2:"L0.sol"
+object "L0" {
+    code {
+        sstore(0, datasize("L0"))
+    }
+}

test/cmdlineTests/strict_asm_asm_json_subobjects_with_debug/args

@@ -0,0 +1 @@
+--strict-assembly - --asm-json --pretty-json --json-indent 4