/// <summary>
/// Called by determineHead and may be overridden in subclasses
/// if special treatment is necessary for particular categories.
/// </summary>
/// <param name="t">The tre to determine the head daughter of</param>
/// <param name="parent">The parent of t (or may be null)</param>
/// <returns>The head daughter of t</returns>
protected virtual Tree DetermineNonTrivialHead(Tree t, Tree parent)
{
Tree theHead = null;
string motherCat = Tlp.BasicCategory(t.Label().Value());
if (motherCat.StartsWith("@"))
{
motherCat = motherCat.Substring(1);
}
// We know we have nonterminals underneath
// (a bit of a Penn Treebank assumption, but).
// Look at label.
// a total special case....
// first look for POS tag at end
// this appears to be redundant in the Collins case since the rule already would do that
// Tree lastDtr = t.lastChild();
// if (tlp.basicCategory(lastDtr.label().value()).equals("POS")) {
// theHead = lastDtr;
// } else {
string[][] how = null;
var success = NonTerminalInfo.TryGetValue(motherCat, out how);
Tree[] kids = t.Children();
if (!success)
{
if (DefaultRule != null)
{
return(TraverseLocate(kids, DefaultRule, true));
}
else
{
throw new ArgumentException("No head rule defined for " + motherCat + " using "
/*+ this.getClass()*/ + " in " + t);
}
}
for (int i = 0; i < how.Length; i++)
{
bool lastResort = (i == how.Length - 1);
theHead = TraverseLocate(kids, how[i], lastResort);
if (theHead != null)
{
break;
}
}
return(theHead);
}
// only leaves NP-TMP and NP-ADV
protected string CleanUpLabel(string label)
{
if (label == null)
{
return("");
// This shouldn't really happen, but can happen if there are unlabeled nodes further down a tree, as apparently happens in at least the 20100730 era American National Corpus
}
bool nptemp = TmpPattern.IsMatch(label);
bool npadv = AdvPattern.IsMatch(label);
label = Tlp.BasicCategory(label);
if (nptemp)
{
label = label + "-TMP";
}
else if (npadv)
{
label = label + "-ADV";
}
return(label);
}
public string Apply(string input)
{
return(tlp.BasicCategory(input));
}