public (List <Token>, List <ScriptError>) Tokenize(string source)
{
int index = 0;
var tokens = new List <Token>();
var wsRegex = new Regex("[\t ]");
var errors = new List <ScriptError>();
while (index < source.Length)
{
Token token = null;
var match = wsRegex.Match(source, index);
if (match.Success && (match.Index - index) == 0)
{
index += match.Length;
continue;
}
foreach (var definition in TokenDefinitions)
{
match = definition.Regex.Match(source, index);
if (match.Success && (match.Index - index) == 0)
{
var value = source.Substring(index, match.Length);
if (definition.Type == TokenType.StringLiteral ||
definition.Type == TokenType.DateTimeLiteral ||
definition.Type == TokenType.Field)
{
value = value.Substring(1, value.Length - 2);
}
token = new Token
{
Type = definition.Type,
Contents = value,
Index = index,
};
index += match.Length;
break;
}
}
if (token == null)
{
var errorToken = new Token
{
Type = TokenType.Unknown,
Contents = source[index].ToString(),
Index = index,
};
errors.Add(ScriptError.UnrecognizedSymbol(errorToken, source[index].ToString()));
index++;
}
else
{
tokens.Add(token);
}
}
var eof = new Token
{
Type = TokenType.EOF,
Contents = "End of script",
Index = index,
};
tokens.Add(eof);
return(tokens, errors);
}