public HyphenationTokenFilter(TokenStream input, Hyphenator hyphenator)
: base(input)
{
_hyphenator = hyphenator;
_termAtt = (TermAttribute)AddAttribute(typeof(TermAttribute));
_typeAtt = (TypeAttribute)AddAttribute(typeof(TypeAttribute));
_ofsAtt = (OffsetAttribute)AddAttribute(typeof(OffsetAttribute));
}
public Hyphenator Read(XDocument document)
{
if (document == null || document.Root == null) throw new ArgumentException("Invalid document supplied.", "document");
var hyphenator = new Hyphenator();
var minMax = document.Root.Element("hyphen-min");
hyphenator.MinimumPrefix = int.Parse(GetValue(minMax, "before", "2"));
hyphenator.MinimumSuffix = int.Parse(GetValue(minMax, "after", "2"));
hyphenator.Patterns = ReadPatterns(document.Root.Element("patterns")).ToList();
return hyphenator;
}
public override Hyphenator Load(TextReader source)
{
if (source == null) throw new ArgumentNullException("source");
string line;
bool started = false;
var patterns = new List<HyphenatorPattern>();
while((line = source.ReadLine()) != null)
{
if (!started && line.StartsWith("charset ")) continue; // no support for charset for now.
if(string.IsNullOrWhiteSpace(line)) continue;
started = true;
patterns.Add(new HyphenatorPattern(line));
}
var hyphenator = new Hyphenator();
hyphenator.MinimumPrefix = 2;
hyphenator.MinimumSuffix = 2;
hyphenator.Patterns = patterns;
return hyphenator;
}
