Add stricter hex underscore rules

2 files changed, 44 insertions, 0 deletions

diff --git a/libsolidity/analysis/SyntaxChecker.cpp b/libsolidity/analysis/SyntaxChecker.cpp
index 77492499..dc2e35e5 100644
--- a/libsolidity/analysis/SyntaxChecker.cpp
+++ b/libsolidity/analysis/SyntaxChecker.cpp
@@ -24,6 +24,9 @@
 #include <libsolidity/interface/Version.h>
 #include <boost/algorithm/cxx11/all_of.hpp>
 
+#include <boost/algorithm/string.hpp>
+#include <string>
+
 using namespace std;
 using namespace dev;
 using namespace dev::solidity;
@@ -183,6 +186,46 @@ bool SyntaxChecker::visit(Throw const& _throwStatement)
 	return true;
 }
 
+bool SyntaxChecker::visit(Literal const& _literal)
+{
+	if (!_literal.isHexNumber())
+		return true;
+	// We have a hex literal. Do underscore validation
+	solAssert(_literal.value().substr(0, 2) == "0x", "");
+	ASTString value = _literal.value().substr(2); // Skip the 0x
+	vector<ASTString> parts;
+	boost::split(parts, value, boost::is_any_of("_"));
+
+	if (parts.size() == 1) // no underscores
+		return true;
+	// Everything except first and last part must be 4 chars in length
+	for (size_t i = 1; i + 1 < parts.size(); ++i)
+	{
+		if (parts[i].size() != 4)
+			m_errorReporter.syntaxError(_literal.location(), "Invalid use of underscores in hex literal. Found inner part with " + to_string(parts[i].size()) + " digits (has to be 4 digits).");
+	}
+
+	// Validate rightmost block
+	if (parts.back().size() == 4) // If ends with 4 digits, then no need to validate first block
+		return true;
+
+	// Validate leftmost block
+	// If first part is 4 digits then last part's length has to be even to avoid ambiguity over zero padding
+	if (parts.front().size() == 4)
+	{
+		if (parts.back().size() % 2 == 0)
+			return true;
+		m_errorReporter.syntaxError(_literal.location(), "Invalid use of underscores in hex literal. If the first part has 4 digits, it is assumed to be a byte sequence instead of a number and thus the last part should have an even number of digits.");
+	}
+	else
+	{
+		// Both first and last part is invalid
+		m_errorReporter.syntaxError(_literal.location(), "Invalid use of underscores in hex literal. First or last part must have 4 digits.");
+	}
+
+	return true;
+}
+
 bool SyntaxChecker::visit(UnaryOperation const& _operation)
 {
 	if (_operation.getOperator() == Token::Add)
diff --git a/libsolidity/analysis/SyntaxChecker.h b/libsolidity/analysis/SyntaxChecker.h
index 28a0f66e..897df676 100644
--- a/libsolidity/analysis/SyntaxChecker.h
+++ b/libsolidity/analysis/SyntaxChecker.h
@@ -73,6 +73,7 @@ private:
 	virtual bool visit(VariableDeclarationStatement const& _statement) override;
 
 	virtual bool visit(StructDefinition const& _struct) override;
+	virtual bool visit(Literal const& _literal) override;
 
 	ErrorReporter& m_errorReporter;