private IEnumerable <string> Tokenize(string text)
{
IList <string> tokens = new List <string>();
int tokenLen = 0;
int len = text.Length;
var buf = new char[255];
for (int i = 0; i <= len; i++)
{
char c = '\x00';
if (i != len)
{
c = TokenizeRules.Normalize(text[i]);
}
if (i != len && TokenizeRules.IsTokenChar(c))
{
buf[tokenLen] = c;
tokenLen++;
if (tokenLen > 255)
{
throw new InvalidOperationException("Token longer than 255 characters.");
}
}
else
{
// we're not interested in tokens shorter than 3 chars
if (tokenLen >= 3)
{
tokens.Add(new string(buf, 0, tokenLen));
}
tokenLen = 0;
}
}
return(tokens);
}
/// <summary>Converts char to lower case
/// <see cref="char.ToLower(char)" />.
/// </summary>
protected override char Normalize(char c)
{
return(TokenizeRules.Normalize(c));
}
/// <summary>Collects only characters which satisfy
/// <see cref="char.IsLetter(char)" />.
/// </summary>
protected override bool IsTokenChar(char c)
{
return(TokenizeRules.IsTokenChar(c));
}