Add stricter hex underscore rules

1 files changed, 19 insertions, 16 deletions

diff --git a/libsolidity/parsing/Scanner.cpp b/libsolidity/parsing/Scanner.cpp
index 65189b19..c223779e 100644
--- a/libsolidity/parsing/Scanner.cpp
+++ b/libsolidity/parsing/Scanner.cpp
@@ -726,21 +726,26 @@ Token::Value Scanner::scanHexString()
 
 void Scanner::scanDecimalDigits()
 {
-	if (!isDecimalDigit(m_char)) // avoid underscore at beginning
-		return;
-	while (isDecimalDigit(m_char) || m_char == '_') 
+	// Parse for regex [:digit:]+(_[:digit:]+)*
+
+	do
 	{
+		if (!isDecimalDigit(m_char))
+			return;
+		while (isDecimalDigit(m_char))
+			addLiteralCharAndAdvance();
+
 		if (m_char == '_') 
 		{
 			advance();
-			if (!isDecimalDigit(m_char)) // avoid trailing underscore
+			if (!isDecimalDigit(m_char)) // Trailing underscore. Rollback and allow next step to flag it as illegal
 			{
 				rollback(1);
-				break;
+				return;
 			}
 		}
-		addLiteralCharAndAdvance();
 	}
+	while (isDecimalDigit(m_char));
 }
 
 Token::Value Scanner::scanNumber(char _charSeen)
@@ -768,19 +773,17 @@ Token::Value Scanner::scanNumber(char _charSeen)
 				addLiteralCharAndAdvance();
 				if (!isHexDigit(m_char))
 					return Token::Illegal; // we must have at least one hex digit after 'x'/'X'
-				while (isHexDigit(m_char) || m_char == '_') // same logic as scanDecimalDigits
+				char last = m_char;
+				while (isHexDigit(m_char) || m_char == '_') // Unlike decimal digits, we keep the underscores for later validation
 				{
-					if (m_char == '_') 
-					{
-						advance();
-						if (!isHexDigit(m_char)) // avoid trailing underscore
-						{
-							rollback(1);
-							break;
-						}
-					}
+					if (m_char == '_' && last == '_')
+						return Token::Illegal; // Double underscore
+
+					last = m_char;
 					addLiteralCharAndAdvance();
 				}
+				if (last == '_')
+					return Token::Illegal; // Trailing underscore
 			}
 			else if (isDecimalDigit(m_char))
 				// We do not allow octal numbers