Improve Error Reporting of SemVer Parser

Author: vmanektalla

Changelog.md

@@ -15,6 +15,7 @@ Compiler Features:
  * Optimizer: Added optimization rule ``and(shl(X, Y), shl(X, Z)) => shl(X, and(Y, Z))``.
  * SMTChecker: Support Eldarica as a Horn solver for the CHC engine when using the CLI option ``--model-checker-solvers eld``. The binary `eld` must be available in the system.
  * SMTChecker: Make ``z3`` the default solver for the BMC and CHC engines instead of all solvers.
+ * Parser: More detailed error messages about invalid version pragmas.
 
 
 Bugfixes:

liblangutil/SemVerHandler.cpp

@@ -27,10 +27,12 @@
 
 #include <functional>
 #include <limits>
+#include <fmt/format.h>
 
 using namespace std;
 using namespace solidity;
 using namespace solidity::langutil;
+using namespace solidity::util;
 
 SemVerMatchExpressionParser::SemVerMatchExpressionParser(vector<Token> _tokens, vector<string> _literals):
 	m_tokens(std::move(_tokens)), m_literals(std::move(_literals))
@@ -52,7 +54,7 @@ SemVerVersion::SemVerVersion(string const& _versionString)
 		if (level < 2)
 		{
 			if (i == end || *i != '.')
-				BOOST_THROW_EXCEPTION(SemVerError());
+				solThrow(SemVerError, "Invalid versionString: "s + _versionString);
 			else
 				++i;
 		}
@@ -70,7 +72,7 @@ SemVerVersion::SemVerVersion(string const& _versionString)
 		build = string(buildStart, i);
 	}
 	if (i != end)
-		BOOST_THROW_EXCEPTION(SemVerError());
+		solThrow(SemVerError, "Invalid versionString "s + _versionString);
 }
 
 bool SemVerMatchExpression::MatchComponent::matches(SemVerVersion const& _version) const
@@ -156,12 +158,12 @@ bool SemVerMatchExpression::matches(SemVerVersion const& _version) const
 	return false;
 }
 
-optional<SemVerMatchExpression> SemVerMatchExpressionParser::parse()
+SemVerMatchExpression SemVerMatchExpressionParser::parse()
 {
 	reset();
 
 	if (m_tokens.empty())
-		return nullopt;
+		solThrow(SemVerError, "Empty version pragma.");
 
 	try
 	{
@@ -171,14 +173,19 @@ optional<SemVerMatchExpression> SemVerMatchExpressionParser::parse()
 			if (m_pos >= m_tokens.size())
 				break;
 			if (currentToken() != Token::Or)
-				BOOST_THROW_EXCEPTION(SemVerError());
+			{
+				solThrow(
+					SemVerError,
+					"You can only combine version ranges using the || operator."
+				);
+			}
 			nextToken();
 		}
 	}
-	catch (SemVerError const&)
+	catch (SemVerError const& e)
 	{
 		reset();
-		return nullopt;
+		throw e;
 	}
 
 	return m_expression;
@@ -265,14 +272,22 @@ unsigned SemVerMatchExpressionParser::parseVersionPart()
 		{
 			c = currentChar();
 			if (v * 10 < v || v * 10 + static_cast<unsigned>(c - '0') < v * 10)
-				BOOST_THROW_EXCEPTION(SemVerError());
+				solThrow(SemVerError, "Integer too large to be used in a version number.");
 			v = v * 10 + static_cast<unsigned>(c - '0');
 			nextChar();
 		}
 		return v;
 	}
+	else if (c == char(-1))
+		solThrow(SemVerError, "Expected version number but reached end of pragma.");
 	else
-		BOOST_THROW_EXCEPTION(SemVerError());
+		solThrow(
+			SemVerError, fmt::format(
+				"Expected the start of a version number but instead found character '{}'. "
+				"Version number is invalid or the pragma is not terminated with a semicolon.",
+				c
+			)
+		);
 }
 
 char SemVerMatchExpressionParser::currentChar() const

liblangutil/SemVerHandler.h

@@ -25,17 +25,18 @@
 
 #include <liblangutil/Token.h>
 #include <libsolutil/Assertions.h>
+#include <liblangutil/Exceptions.h>
 
 #include <string>
-#include <optional>
 #include <utility>
 #include <vector>
 
 namespace solidity::langutil
 {
 
-class SemVerError: public util::Exception
+struct SemVerError: public util::Exception
 {
+
 };
 
 #undef major
@@ -87,8 +88,8 @@ class SemVerMatchExpressionParser
 public:
 	SemVerMatchExpressionParser(std::vector<Token> _tokens, std::vector<std::string> _literals);
 
-	/// Returns an expression if it was parseable, or nothing otherwise.
-	std::optional<SemVerMatchExpression> parse();
+	/// Returns an expression if it was parsable, or throws a SemVerError otherwise.
+	SemVerMatchExpression parse();
 
 private:
 	void reset();

libsolidity/analysis/SyntaxChecker.cpp

@@ -153,22 +153,28 @@ bool SyntaxChecker::visit(PragmaDirective const& _pragma)
 	}
 	else if (_pragma.literals()[0] == "solidity")
 	{
-		vector<Token> tokens(_pragma.tokens().begin() + 1, _pragma.tokens().end());
-		vector<string> literals(_pragma.literals().begin() + 1, _pragma.literals().end());
-		SemVerMatchExpressionParser parser(tokens, literals);
-		auto matchExpression = parser.parse();
-		// An unparsable version pragma is an unrecoverable fatal error in the parser.
-		solAssert(matchExpression.has_value(), "");
-		static SemVerVersion const currentVersion{string(VersionString)};
-		if (!matchExpression->matches(currentVersion))
-			m_errorReporter.syntaxError(
-				3997_error,
-				_pragma.location(),
-				"Source file requires different compiler version (current compiler is " +
-				string(VersionString) + ") - note that nightly builds are considered to be "
-				"strictly less than the released version"
-			);
-		m_versionPragmaFound = true;
+		try
+		{
+			vector<Token> tokens(_pragma.tokens().begin() + 1, _pragma.tokens().end());
+			vector<string> literals(_pragma.literals().begin() + 1, _pragma.literals().end());
+			SemVerMatchExpressionParser parser(tokens, literals);
+			SemVerMatchExpression matchExpression = parser.parse();
+			static SemVerVersion const currentVersion{string(VersionString)};
+			if (!matchExpression.matches(currentVersion))
+				m_errorReporter.syntaxError(
+					3997_error,
+					_pragma.location(),
+					"Source file requires different compiler version (current compiler is " +
+					string(VersionString) + ") - note that nightly builds are considered to be "
+					"strictly less than the released version"
+				);
+			m_versionPragmaFound = true;
+		}
+		catch (SemVerError const&)
+		{
+			// An unparsable version pragma is an unrecoverable fatal error in the parser.
+			solAssert(false);
+		}
 	}
 	else
 		m_errorReporter.syntaxError(4936_error, _pragma.location(), "Unknown pragma \"" + _pragma.literals()[0] + "\"");

libsolidity/parsing/Parser.cpp

@@ -160,27 +160,31 @@ ASTPointer<SourceUnit> Parser::parse(CharStream& _charStream)
 void Parser::parsePragmaVersion(SourceLocation const& _location, vector<Token> const& _tokens, vector<string> const& _literals)
 {
 	SemVerMatchExpressionParser parser(_tokens, _literals);
-	auto matchExpression = parser.parse();
-	if (!matchExpression.has_value())
+	try
+	{
+		SemVerMatchExpression matchExpression = parser.parse();
+		static SemVerVersion const currentVersion{string(VersionString)};
+		// FIXME: only match for major version incompatibility
+		if (!matchExpression.matches(currentVersion))
+			// If m_parserErrorRecovery is true, the same message will appear from SyntaxChecker::visit(),
+			// so we don't need to report anything here.
+			if (!m_parserErrorRecovery)
+				m_errorReporter.fatalParserError(
+					5333_error,
+					_location,
+					"Source file requires different compiler version (current compiler is " +
+					string(VersionString) + ") - note that nightly builds are considered to be "
+					"strictly less than the released version"
+				);
+	}
+	catch (SemVerError const& matchError)
+	{
 		m_errorReporter.fatalParserError(
 			1684_error,
 			_location,
-			"Found version pragma, but failed to parse it. "
-			"Please ensure there is a trailing semicolon."
+			"Invalid version pragma. "s + matchError.what()
 		);
-	static SemVerVersion const currentVersion{string(VersionString)};
-	// FIXME: only match for major version incompatibility
-	if (!matchExpression->matches(currentVersion))
-		// If m_parserErrorRecovery is true, the same message will appear from SyntaxChecker::visit(),
-		// so we don't need to report anything here.
-		if (!m_parserErrorRecovery)
-			m_errorReporter.fatalParserError(
-				5333_error,
-				_location,
-				"Source file requires different compiler version (current compiler is " +
-				string(VersionString) + ") - note that nightly builds are considered to be "
-				"strictly less than the released version"
-			);
+	}
 }
 
 ASTPointer<StructuredDocumentation> Parser::parseStructuredDocumentation()

test/libsolidity/SemVerMatcher.cpp

@@ -27,6 +27,7 @@
 #include <liblangutil/Scanner.h>
 #include <liblangutil/SemVerHandler.h>
 #include <test/Common.h>
+#include <test/libsolidity/util/SoltestErrors.h>
 
 #include <boost/test/unit_test.hpp>
 
@@ -58,15 +59,39 @@ SemVerMatchExpression parseExpression(string const& _input)
 		scanner.next();
 	}
 
-	auto expression = SemVerMatchExpressionParser(tokens, literals).parse();
-	BOOST_REQUIRE(expression.has_value());
-	BOOST_CHECK_MESSAGE(
-		expression->isValid(),
-		"Expression \"" + _input + "\" did not parse properly."
-	);
-	return *expression;
+	try
+	{
+		auto matchExpression = SemVerMatchExpressionParser(tokens, literals).parse();
+
+		BOOST_CHECK_MESSAGE(
+			matchExpression.isValid(),
+			"Expression \"" + _input + "\" did not parse properly."
+		);
+
+		return matchExpression;
+	}
+	catch (SemVerError const&)
+	{
+		// Ignored, since a test case should have a parsable version
+		soltestAssert(false);
+	}
+}
+
 }
 
+BOOST_AUTO_TEST_CASE(exception_on_invalid_version_in_semverversion_constructor)
+{
+	BOOST_CHECK_EXCEPTION(
+		SemVerVersion version("1.2"),
+		SemVerError,
+		[&](auto const& _exception) { BOOST_TEST(_exception.what() == "Invalid versionString: 1.2"); return true; }
+	);
+
+	BOOST_CHECK_EXCEPTION(
+		SemVerVersion version("-1.2.0"),
+		SemVerError,
+		[&](auto const& _exception) { BOOST_TEST(_exception.what() == "Invalid versionString: -1.2.0"); return true; }
+	);
 }
 
 BOOST_AUTO_TEST_CASE(positive_range)
@@ -159,9 +184,9 @@ BOOST_AUTO_TEST_CASE(positive_range)
 	for (auto const& t: tests)
 	{
 		SemVerVersion version(t.second);
-		SemVerMatchExpression expression = parseExpression(t.first);
+		SemVerMatchExpression matchExpression = parseExpression(t.first);
 		BOOST_CHECK_MESSAGE(
-			expression.matches(version),
+			matchExpression.matches(version),
 			"Version \"" + t.second + "\" did not satisfy expression \"" + t.first + "\""
 		);
 	}
@@ -235,9 +260,9 @@ BOOST_AUTO_TEST_CASE(negative_range)
 	for (auto const& t: tests)
 	{
 		SemVerVersion version(t.second);
-		SemVerMatchExpression expression = parseExpression(t.first);
+		auto matchExpression = parseExpression(t.first);
 		BOOST_CHECK_MESSAGE(
-			!expression.matches(version),
+			!matchExpression.matches(version),
 			"Version \"" + t.second + "\" did satisfy expression \"" + t.first + "\" " +
 			"(although it should not)"
 		);

test/libsolidity/syntaxTests/pragma/blank_space_version.sol

@@ -0,0 +1,3 @@
+pragma solidity ;
+// ----
+// ParserError 1684: (0-17): Invalid version pragma. Empty version pragma.

test/libsolidity/syntaxTests/pragma/broken_version_2.sol

@@ -1,3 +1,3 @@
 pragma solidity pragma;
 // ----
-// ParserError 1684: (0-23): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-23): Invalid version pragma. Expected the start of a version number but instead found character 'p'. Version number is invalid or the pragma is not terminated with a semicolon.

test/libsolidity/syntaxTests/pragma/broken_version_4.sol

@@ -1,3 +1,3 @@
 pragma solidity (8.0.0);
 // ----
-// ParserError 1684: (0-24): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-24): Invalid version pragma. Expected the start of a version number but instead found character '('. Version number is invalid or the pragma is not terminated with a semicolon.

test/libsolidity/syntaxTests/pragma/broken_version_5.sol

@@ -1,3 +1,3 @@
 pragma solidity 88_;
 // ----
-// ParserError 1684: (0-20): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-20): Invalid version pragma. Expected the start of a version number but instead found character '_'. Version number is invalid or the pragma is not terminated with a semicolon.

test/libsolidity/syntaxTests/pragma/broken_version_6.sol

@@ -1,3 +1,3 @@
 pragma solidity v1.2.3;
 // ----
-// ParserError 1684: (0-23): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-23): Invalid version pragma. Expected the start of a version number but instead found character 'v'. Version number is invalid or the pragma is not terminated with a semicolon.

test/libsolidity/syntaxTests/pragma/broken_version_7.sol

@@ -1,3 +1,3 @@
 pragma solidity >0.5.0<;
 // ----
-// ParserError 1684: (0-24): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-24): Invalid version pragma. Expected version number but reached end of pragma.

test/libsolidity/syntaxTests/pragma/empty_version.sol

@@ -1,3 +1,3 @@
 pragma solidity;
 // ----
-// ParserError 1684: (0-16): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-16): Invalid version pragma. Empty version pragma.

test/libsolidity/syntaxTests/pragma/invalid_range_conjunction_range_first.sol

@@ -0,0 +1,3 @@
+pragma solidity 0.4.1 - 0.6.2 0.8.17;
+// ----
+// ParserError 1684: (0-37): Invalid version pragma. You can only combine version ranges using the || operator.

test/libsolidity/syntaxTests/pragma/invalid_range_conjunction_range_second.sol

@@ -0,0 +1,3 @@
+pragma solidity 0.8.17 0.4.1 - 0.6.2;
+// ----
+// ParserError 1684: (0-37): Invalid version pragma. Expected the start of a version number but instead found character '-'. Version number is invalid or the pragma is not terminated with a semicolon.

test/libsolidity/syntaxTests/pragma/invalid_range_conjunction_two_ranges.sol

@@ -0,0 +1,3 @@
+pragma solidity 0.4.0 - 0.4.1 0.4.1 - 0.6.2;
+// ----
+// ParserError 1684: (0-44): Invalid version pragma. You can only combine version ranges using the || operator.

test/libsolidity/syntaxTests/pragma/unterminated_version.sol

@@ -1,4 +1,4 @@
 pragma solidity 0.4.3
 pragma abicoder v2;
 // ----
-// ParserError 1684: (0-41): Found version pragma, but failed to parse it. Please ensure there is a trailing semicolon.
+// ParserError 1684: (0-41): Invalid version pragma. Expected the start of a version number but instead found character 'p'. Version number is invalid or the pragma is not terminated with a semicolon.

test/libsolidity/syntaxTests/pragma/valid_range_disjunction_range_second.sol

@@ -0,0 +1,2 @@
+pragma solidity 0.8.17 || 0.4.1 - 0.9.0;
+// ----

test/libsolidity/syntaxTests/pragma/valid_range_disjunction_two_ranges.sol

@@ -0,0 +1,2 @@
+pragma solidity 0.4.0 - 0.9.0 || 0.4.1 - 0.6.2;
+// ----

test/libsolidity/syntaxTests/pragma/version_number_too_large.sol

@@ -0,0 +1,3 @@
+pragma solidity 4294967296;
+// ----
+// ParserError 1684: (0-27): Invalid version pragma. Integer too large to be used in a version number.