Parse @ast-id annotation.

Author: chriseth

libyul/AST.h

@@ -38,14 +38,22 @@ using Type = YulString;
 
 struct DebugData
 {
-	explicit DebugData(langutil::SourceLocation _location): location(std::move(_location)) {}
+	explicit DebugData(langutil::SourceLocation _location, std::optional<int64_t> _astID = {}):
+		location(std::move(_location)),
+		astID(std::move(_astID))
+	{}
+
+	static std::shared_ptr<DebugData const> create(
+		langutil::SourceLocation _location = {},
+		std::optional<int64_t> _astID = {}
+	)
+	{
+		return std::make_shared<DebugData const>(std::move(_location), std::move(_astID));
+	}
+
 	langutil::SourceLocation location;
 	/// ID in the (Solidity) source AST.
 	std::optional<int64_t> astID;
-	static std::shared_ptr<DebugData const> create(langutil::SourceLocation _location = {})
-	{
-		return std::make_shared<DebugData const>(_location);
-	}
 };
 
 struct TypedName { std::shared_ptr<DebugData const> debugData; YulString name; Type type; };

libyul/AsmParser.cpp

@@ -103,7 +103,7 @@ unique_ptr<Block> Parser::parseInline(std::shared_ptr<Scanner> const& _scanner)
 	try
 	{
 		m_scanner = _scanner;
-		if (m_sourceNames)
+		if (m_useSourceLocationFrom == UseSourceLocationFrom::Comments)
 			fetchDebugDataFromComment();
 		return make_unique<Block>(parseBlock());
 	}
@@ -136,6 +136,8 @@ void Parser::fetchDebugDataFromComment()
 	match_results<string_view::const_iterator> match;
 
 	langutil::SourceLocation sourceLocation = m_debugDataOverride->location;
+	// Empty for each new node.
+	optional<int> astID;
 
 	while (regex_search(commentLiteral.cbegin(), commentLiteral.cend(), match, tagRegex))
 	{
@@ -149,12 +151,19 @@ void Parser::fetchDebugDataFromComment()
 			else
 				break;
 		}
+		else if (match[1] == "@ast-id")
+		{
+			if (auto parseResult = parseASTIDComment(commentLiteral, m_scanner->currentCommentLocation()))
+				tie(commentLiteral, astID) = *parseResult;
+			else
+				break;
+		}
 		else
 			// Ignore unrecognized tags.
 			continue;
 	}
 
-	m_debugDataOverride = DebugData::create(sourceLocation);
+	m_debugDataOverride = DebugData::create(sourceLocation, astID);
 }
 
 optional<pair<string_view, SourceLocation>> Parser::parseSrcComment(
@@ -222,6 +231,38 @@ optional<pair<string_view, SourceLocation>> Parser::parseSrcComment(
 	return {{tail, SourceLocation{}}};
 }
 
+optional<pair<string_view, optional<int>>> Parser::parseASTIDComment(
+	string_view _arguments,
+	langutil::SourceLocation const& _commentLocation
+)
+{
+	static regex const argRegex = regex(
+		R"~~(^(\d+)(?:\s|$))~~",
+		regex_constants::ECMAScript | regex_constants::optimize
+	);
+	match_results<string_view::const_iterator> match;
+	optional<int> astID;
+	bool matched = regex_search(_arguments.cbegin(), _arguments.cend(), match, argRegex);
+	string_view tail = _arguments;
+	if (matched)
+	{
+		solAssert(match.size() == 2, "");
+		tail = _arguments.substr(static_cast<size_t>(match.position() + match.length()));
+
+		astID = toInt(match[1].str());
+	}
+
+	if (!matched || !astID || *astID < 0 || static_cast<int64_t>(*astID) != *astID)
+	{
+		m_errorReporter.syntaxError(1749_error, _commentLocation, "Invalid argument for @ast-id.");
+		astID = nullopt;
+	}
+	if (matched)
+		return {{_arguments, astID}};
+	else
+		return nullopt;
+}
+
 Block Parser::parseBlock()
 {
 	RecursionGuard recursionGuard(*this);

libyul/AsmParser.h

@@ -108,8 +108,12 @@ class Parser: public langutil::ParserBase
 
 	void fetchDebugDataFromComment();
 
-	std::optional<std::pair<std::string_view, langutil::SourceLocation>>
-	parseSrcComment(
+	std::optional<std::pair<std::string_view, langutil::SourceLocation>> parseSrcComment(
+		std::string_view _arguments,
+		langutil::SourceLocation const& _commentLocation
+	);
+
+	std::optional<std::pair<std::string_view, std::optional<int>>> parseASTIDComment(
 		std::string_view _arguments,
 		langutil::SourceLocation const& _commentLocation
 	);

test/libyul/Parser.cpp

@@ -804,6 +804,51 @@ BOOST_AUTO_TEST_CASE(customSourceLocations_with_code_snippets_with_nested_locati
 	CHECK_LOCATION(literal128.debugData->location, "source1", 96, 165);
 }
 
+BOOST_AUTO_TEST_CASE(astid)
+{
+	ErrorList errorList;
+	ErrorReporter reporter(errorList);
+	auto const sourceText = R"(
+		/// @src -1:-1:-1 @ast-id 7
+		{
+			/** @ast-id 2 */
+			function f(x) -> y {}
+			mstore(1, 2)
+		}
+	)";
+	EVMDialectTyped const& dialect = EVMDialectTyped::instance(EVMVersion{});
+	shared_ptr<Block> result = parse(sourceText, dialect, reporter);
+	BOOST_REQUIRE(!!result);
+	BOOST_CHECK(result->debugData->astID == int64_t(7));
+	auto const& funDef = get<FunctionDefinition>(result->statements.at(0));
+	BOOST_CHECK(funDef.debugData->astID == int64_t(2));
+	BOOST_CHECK(funDef.parameters.at(0).debugData->astID == nullopt);
+	BOOST_CHECK(debugDataOf(result->statements.at(1))->astID == nullopt);
+}
+
+BOOST_AUTO_TEST_CASE(astid_reset)
+{
+	ErrorList errorList;
+	ErrorReporter reporter(errorList);
+	auto const sourceText = R"(
+		/// @src -1:-1:-1 @ast-id 7 @src 1:1:1
+		{
+			/** @ast-id 2 */
+			function f(x) -> y {}
+			mstore(1, 2)
+		}
+	)";
+	EVMDialectTyped const& dialect = EVMDialectTyped::instance(EVMVersion{});
+	shared_ptr<Block> result = parse(sourceText, dialect, reporter);
+	BOOST_REQUIRE(!!result);
+	BOOST_CHECK(result->debugData->astID == int64_t(7));
+	auto const& funDef = get<FunctionDefinition>(result->statements.at(0));
+	BOOST_CHECK(funDef.debugData->astID == int64_t(2));
+	BOOST_CHECK(funDef.parameters.at(0).debugData->astID == nullopt);
+	BOOST_CHECK(debugDataOf(result->statements.at(1))->astID == nullopt);
+}
+
+
 BOOST_AUTO_TEST_CASE(customSourceLocations_multiple_src_tags_on_one_line)
 {
 	ErrorList errorList;