Merge pull request ethereum#11867 from ethereum/solidity-snippets-next-to-source-locations-in-ir

Solidity snippets next to source locations in IR

Author: chriseth

libsolidity/ast/ASTJsonConverter.cpp

@@ -26,7 +26,6 @@
 #include <libsolidity/ast/TypeProvider.h>
 
 #include <libyul/AsmJsonConverter.h>
-#include <libyul/AsmPrinter.h>
 #include <libyul/AST.h>
 #include <libyul/backends/evm/EVMDialect.h>
 

libsolidity/codegen/ir/Common.cpp

@@ -22,9 +22,13 @@
 
 #include <libsolutil/CommonIO.h>
 
+#include <libyul/AsmPrinter.h>
+
 using namespace std;
-using namespace solidity::util;
+using namespace solidity::langutil;
 using namespace solidity::frontend;
+using namespace solidity::util;
+using namespace solidity::yul;
 
 namespace solidity::frontend
 {
@@ -131,12 +135,12 @@ string dispenseLocationComment(langutil::SourceLocation const& _location, IRGene
 {
 	solAssert(_location.sourceName, "");
 	_context.markSourceUsed(*_location.sourceName);
-	return "/// @src "
-		+ to_string(_context.sourceIndices().at(*_location.sourceName))
-		+ ":"
-		+ to_string(_location.start)
-		+ ":"
-		+ to_string(_location.end);
+	return AsmPrinter::formatSourceLocationComment(
+		_location,
+		_context.sourceIndices(),
+		true /* _statement */,
+		_context.soliditySourceProvider()
+	);
 }
 
 string dispenseLocationComment(ASTNode const& _node, IRGenerationContext& _context)

libsolidity/codegen/ir/IRGenerationContext.h

@@ -29,6 +29,7 @@
 #include <libsolidity/codegen/MultiUseYulFunctionCollector.h>
 #include <libsolidity/codegen/ir/Common.h>
 
+#include <liblangutil/CharStreamProvider.h>
 #include <liblangutil/EVMVersion.h>
 
 #include <libsolutil/Common.h>
@@ -72,13 +73,15 @@ class IRGenerationContext
 		ExecutionContext _executionContext,
 		RevertStrings _revertStrings,
 		OptimiserSettings _optimiserSettings,
-		std::map<std::string, unsigned> _sourceIndices
+		std::map<std::string, unsigned> _sourceIndices,
+		langutil::CharStreamProvider const* _soliditySourceProvider
 	):
 		m_evmVersion(_evmVersion),
 		m_executionContext(_executionContext),
 		m_revertStrings(_revertStrings),
 		m_optimiserSettings(std::move(_optimiserSettings)),
-		m_sourceIndices(std::move(_sourceIndices))
+		m_sourceIndices(std::move(_sourceIndices)),
+		m_soliditySourceProvider(_soliditySourceProvider)
 	{}
 
 	MultiUseYulFunctionCollector& functionCollector() { return m_functions; }
@@ -171,6 +174,8 @@ class IRGenerationContext
 
 	bool immutableRegistered(VariableDeclaration const& _varDecl) const { return m_immutableVariables.count(&_varDecl); }
 
+	langutil::CharStreamProvider const* soliditySourceProvider() const { return m_soliditySourceProvider; }
+
 private:
 	langutil::EVMVersion m_evmVersion;
 	ExecutionContext m_executionContext;
@@ -214,6 +219,8 @@ class IRGenerationContext
 	std::map<int64_t, uint64_t> m_functionIDs;
 
 	std::set<ContractDefinition const*, ASTNode::CompareByID> m_subObjects;
+
+	langutil::CharStreamProvider const* m_soliditySourceProvider = nullptr;
 };
 
 }

libsolidity/codegen/ir/IRGenerator.cpp

@@ -45,8 +45,9 @@
 
 using namespace std;
 using namespace solidity;
-using namespace solidity::util;
 using namespace solidity::frontend;
+using namespace solidity::langutil;
+using namespace solidity::util;
 
 namespace
 {
@@ -115,7 +116,7 @@ pair<string, string> IRGenerator::run(
 		" *                !USE AT YOUR OWN RISK!               *\n"
 		" *=====================================================*/\n\n";
 
-	return {warning + ir, warning + asmStack.print()};
+	return {warning + ir, warning + asmStack.print(m_context.soliditySourceProvider())};
 }
 
 string IRGenerator::generate(
@@ -1064,7 +1065,14 @@ void IRGenerator::resetContext(ContractDefinition const& _contract, ExecutionCon
 		m_context.internalDispatchClean(),
 		"Reset internal dispatch map without consuming it."
 	);
-	IRGenerationContext newContext(m_evmVersion, _context, m_context.revertStrings(), m_optimiserSettings, m_context.sourceIndices());
+	IRGenerationContext newContext(
+		m_evmVersion,
+		_context,
+		m_context.revertStrings(),
+		m_optimiserSettings,
+		m_context.sourceIndices(),
+		m_context.soliditySourceProvider()
+	);
 	newContext.copyFunctionIDsFrom(m_context);
 	m_context = move(newContext);
 

libsolidity/codegen/ir/IRGenerator.h

@@ -28,7 +28,10 @@
 #include <libsolidity/ast/CallGraph.h>
 #include <libsolidity/codegen/ir/IRGenerationContext.h>
 #include <libsolidity/codegen/YulUtilFunctions.h>
+
+#include <liblangutil/CharStreamProvider.h>
 #include <liblangutil/EVMVersion.h>
+
 #include <string>
 
 namespace solidity::frontend
@@ -45,11 +48,19 @@ class IRGenerator
 		langutil::EVMVersion _evmVersion,
 		RevertStrings _revertStrings,
 		OptimiserSettings _optimiserSettings,
-		std::map<std::string, unsigned> _sourceIndices
+		std::map<std::string, unsigned> _sourceIndices,
+		langutil::CharStreamProvider const* _soliditySourceProvider
 	):
 		m_evmVersion(_evmVersion),
 		m_optimiserSettings(_optimiserSettings),
-		m_context(_evmVersion, ExecutionContext::Creation, _revertStrings, std::move(_optimiserSettings), std::move(_sourceIndices)),
+		m_context(
+			_evmVersion,
+			ExecutionContext::Creation,
+			_revertStrings,
+			std::move(_optimiserSettings),
+			std::move(_sourceIndices),
+			_soliditySourceProvider
+		),
 		m_utils(_evmVersion, m_context.revertStrings(), m_context.functionCollector())
 	{}
 

libsolidity/interface/CompilerStack.cpp

@@ -1329,7 +1329,7 @@ void CompilerStack::generateIR(ContractDefinition const& _contract)
 	for (auto const& pair: m_contracts)
 		otherYulSources.emplace(pair.second.contract, pair.second.yulIR);
 
-	IRGenerator generator(m_evmVersion, m_revertStrings, m_optimiserSettings, sourceIndices());
+	IRGenerator generator(m_evmVersion, m_revertStrings, m_optimiserSettings, sourceIndices(), this);
 	tie(compiledContract.yulIR, compiledContract.yulIROptimized) = generator.run(
 		_contract,
 		createCBORMetadata(compiledContract),

libyul/AsmParser.cpp

@@ -135,7 +135,8 @@ void Parser::fetchSourceLocationFromComment()
 		regex_constants::ECMAScript | regex_constants::optimize
 	);
 	static regex const srcTagArgsRegex = regex(
-		R"~~(^(-1|\d+):(-1|\d+):(-1|\d+)(?:\s+|$))~~", // index and location, e.g.: 1:234:-1
+		R"~~(^(-1|\d+):(-1|\d+):(-1|\d+)(?:\s+|$))~~"  // index and location, e.g.: 1:234:-1
+		R"~~(("(?:[^"\\]|\\.)*"?)?)~~",                // optional code snippet, e.g.: "string memory s = \"abc\";..."
 		regex_constants::ECMAScript | regex_constants::optimize
 	);
 
@@ -164,9 +165,22 @@ void Parser::fetchSourceLocationFromComment()
 				return;
 			}
 
-			solAssert(srcTagArgsMatch.size() == 4, "");
+			solAssert(srcTagArgsMatch.size() == 5, "");
 			position += srcTagArgsMatch.position() + srcTagArgsMatch.length();
 
+			if (srcTagArgsMatch[4].matched && (
+				!boost::algorithm::ends_with(srcTagArgsMatch[4].str(), "\"") ||
+				boost::algorithm::ends_with(srcTagArgsMatch[4].str(), "\\\"")
+			))
+			{
+				m_errorReporter.syntaxError(
+					1544_error,
+					commentLocation,
+					"Invalid code snippet in source location mapping. Quote is not terminated."
+				);
+				return;
+			}
+
 			optional<int> const sourceIndex = toInt(srcTagArgsMatch[1].str());
 			optional<int> const start = toInt(srcTagArgsMatch[2].str());
 			optional<int> const end = toInt(srcTagArgsMatch[3].str());

libyul/AsmPrinter.cpp

@@ -27,6 +27,7 @@
 #include <libyul/Dialect.h>
 
 #include <libsolutil/CommonData.h>
+#include <libsolutil/StringUtils.h>
 
 #include <boost/algorithm/string.hpp>
 #include <boost/algorithm/string/replace.hpp>
@@ -38,6 +39,7 @@
 
 using namespace std;
 using namespace solidity;
+using namespace solidity::langutil;
 using namespace solidity::util;
 using namespace solidity::yul;
 
@@ -256,30 +258,63 @@ string AsmPrinter::appendTypeName(YulString _type, bool _isBoolLiteral) const
 		return ":" + _type.str();
 }
 
-string AsmPrinter::formatSourceLocationComment(shared_ptr<DebugData const> const& _debugData,  bool _statement)
+string AsmPrinter::formatSourceLocationComment(
+	SourceLocation const& _location,
+	map<string, unsigned> const& _nameToSourceIndex,
+	bool _statement,
+	CharStreamProvider const* _soliditySourceProvider
+)
 {
-	if (
-		!_debugData ||
-		m_lastLocation == _debugData->location ||
-		m_nameToSourceIndex.empty()
-	)
-		return "";
-
-	m_lastLocation = _debugData->location;
+	yulAssert(!_nameToSourceIndex.empty(), "");
 
 	string sourceIndex = "-1";
-	if (_debugData->location.sourceName)
-		sourceIndex = to_string(m_nameToSourceIndex.at(*_debugData->location.sourceName));
+	string solidityCodeSnippet = "";
+	if (_location.sourceName)
+	{
+		sourceIndex = to_string(_nameToSourceIndex.at(*_location.sourceName));
+
+		if (_soliditySourceProvider)
+		{
+			solidityCodeSnippet = escapeAndQuoteString(
+				_soliditySourceProvider->charStream(*_location.sourceName).singleLineSnippet(_location)
+			);
+
+			// On top of escaping quotes we also escape the slash inside any `*/` to guard against
+			// it prematurely terminating multi-line comment blocks. We do not escape all slashes
+			// because the ones without `*` are not dangerous and ignoring them reduces visual noise.
+			boost::replace_all(solidityCodeSnippet, "*/", "*\\/");
+		}
+	}
 
 	string sourceLocation =
 		"@src " +
 		sourceIndex +
 		":" +
-		to_string(_debugData->location.start) +
+		to_string(_location.start) +
 		":" +
-		to_string(_debugData->location.end);
+		to_string(_location.end);
+
 	return
 		_statement ?
-		"/// " + sourceLocation + "\n" :
-		"/** " + sourceLocation + " */ ";
+		"/// " + joinHumanReadable(vector<string>{sourceLocation, solidityCodeSnippet}, "  ") :
+		"/** " + joinHumanReadable(vector<string>{sourceLocation, solidityCodeSnippet}, "  ") + " */ ";
+}
+
+string AsmPrinter::formatSourceLocationComment(shared_ptr<DebugData const> const& _debugData, bool _statement)
+{
+	if (
+		!_debugData ||
+		m_lastLocation == _debugData->location ||
+		m_nameToSourceIndex.empty()
+	)
+		return "";
+
+	m_lastLocation = _debugData->location;
+
+	return formatSourceLocationComment(
+		_debugData->location,
+		m_nameToSourceIndex,
+		_statement,
+		m_soliditySourceProvider
+	) + (_statement ? "\n" : "");
 }

libyul/AsmPrinter.h

@@ -28,6 +28,7 @@
 
 #include <libsolutil/CommonData.h>
 
+#include <liblangutil/CharStreamProvider.h>
 #include <liblangutil/SourceLocation.h>
 
 #include <map>
@@ -46,9 +47,11 @@ class AsmPrinter
 public:
 	explicit AsmPrinter(
 		Dialect const* _dialect = nullptr,
-		std::optional<std::map<unsigned, std::shared_ptr<std::string const>>> _sourceIndexToName = {}
+		std::optional<std::map<unsigned, std::shared_ptr<std::string const>>> _sourceIndexToName = {},
+		langutil::CharStreamProvider const* _soliditySourceProvider = nullptr
 	):
-		m_dialect(_dialect)
+		m_dialect(_dialect),
+		m_soliditySourceProvider(_soliditySourceProvider)
 	{
 		if (_sourceIndexToName)
 			for (auto&& [index, name]: *_sourceIndexToName)
@@ -58,8 +61,9 @@ class AsmPrinter
 
 	explicit AsmPrinter(
 		Dialect const& _dialect,
-		std::optional<std::map<unsigned, std::shared_ptr<std::string const>>> _sourceIndexToName = {}
-	): AsmPrinter(&_dialect, _sourceIndexToName) {}
+		std::optional<std::map<unsigned, std::shared_ptr<std::string const>>> _sourceIndexToName = {},
+		langutil::CharStreamProvider const* _soliditySourceProvider = nullptr
+	): AsmPrinter(&_dialect, _sourceIndexToName, _soliditySourceProvider) {}
 
 	std::string operator()(Literal const& _literal);
 	std::string operator()(Identifier const& _identifier);
@@ -76,6 +80,13 @@ class AsmPrinter
 	std::string operator()(Leave const& _continue);
 	std::string operator()(Block const& _block);
 
+	static std::string formatSourceLocationComment(
+		langutil::SourceLocation const& _location,
+		std::map<std::string, unsigned> const& _nameToSourceIndex,
+		bool _statement,
+		langutil::CharStreamProvider const* m_soliditySourceProvider = nullptr
+	);
+
 private:
 	std::string formatTypedName(TypedName _variable);
 	std::string appendTypeName(YulString _type, bool _isBoolLiteral = false) const;
@@ -88,8 +99,9 @@ class AsmPrinter
 	}
 
 	Dialect const* const m_dialect = nullptr;
-	std::map<std::string const, unsigned> m_nameToSourceIndex;
+	std::map<std::string, unsigned> m_nameToSourceIndex;
 	langutil::SourceLocation m_lastLocation = {};
+	langutil::CharStreamProvider const* m_soliditySourceProvider = nullptr;
 };
 
 }

libyul/AssemblyStack.cpp

@@ -314,11 +314,11 @@ AssemblyStack::assembleEVMWithDeployed(optional<string_view> _deployName) const
 	return {make_shared<evmasm::Assembly>(assembly), {}};
 }
 
-string AssemblyStack::print() const
+string AssemblyStack::print(CharStreamProvider const* _soliditySourceProvider) const
 {
 	yulAssert(m_parserResult, "");
 	yulAssert(m_parserResult->code, "");
-	return m_parserResult->toString(&languageToDialect(m_language, m_evmVersion)) + "\n";
+	return m_parserResult->toString(&languageToDialect(m_language, m_evmVersion), _soliditySourceProvider) + "\n";
 }
 
 shared_ptr<Object> AssemblyStack::parserResult() const