private Core.CharacterSet Parse()
{
Core.CharacterSet result = new Core.CharacterSet();
Expect('[');
if (LookingAt() == '^')
{
result.Negated = true;
_Position.Advance();
}
const int startState = 0;
const int charclassState = 1;
const int finalState = 99;
int state = startState;
while (state != finalState)
{
char lookAt = LookingAt();
switch (state)
{
case startState:
switch (lookAt)
{
case '[':
// character class
_Position.Advance();
if (LookingAt() == ':')
{
_Position.Advance();
state = charclassState;
}
else if (LookingAt() != '\0')
{
result.Add('[');
// don't advance
}
break;
case '\0':
// unexpected end
throw new Core.LanguagePlatformException(Core.ErrorCode.TokenizerInvalidCharacterSet, _Input);
case ']':
// right bracket - move on to final state
_Position.Advance();
state = finalState;
break;
default:
{
// a plain character (escaped, simple, or Unicode hex)
char lower = ScanChar();
char upper = '\0';
if (LookingAt() == '-')
{
_Position.Advance();
if (LookingAt() == ']')
{
// dash at end of input - add literal dash to charset, skip ], and go to final state
result.Add(lower);
result.Add('-');
_Position.Advance();
state = finalState;
}
else
{
upper = ScanChar();
result.Add(lower, upper);
}
}
else
{
result.Add(lower);
state = startState;
}
}
break;
}
break;
case charclassState:
// just got '[' followed by ':'
if (LookingAt() == '\0')
{
throw new Core.LanguagePlatformException(Core.ErrorCode.TokenizerInvalidCharacterSet, _Input);
}
else
{
StringBuilder className = new StringBuilder();
while (char.IsLetter(LookingAt()))
{
className.Append(LookingAt());
_Position.Advance();
}
Nullable <System.Globalization.UnicodeCategory> category;
if (className.Length == 0 ||
(category = Core.CharacterProperties.GetUnicodeCategoryFromName(className.ToString().ToLower())) == null ||
!category.HasValue)
{
throw new Core.LanguagePlatformException(Core.ErrorCode.TokenizerInvalidCharacterSet, _Input);
}
else
{
result.Add(category.Value);
}
Expect(':');
Expect(']');
state = startState;
}
break;
default:
// unexpected state
throw new Core.LanguagePlatformException(Core.ErrorCode.TokenizerInvalidCharacterSet, _Input);
}
}
return(result);
}
