public static Tree NormalizeTree(Tree tree, TreeNormalizer tn, ITreeFactory tf)
{
foreach (Tree node in tree)
{
if (node.IsLeaf())
{
node.Label().SetValue(tn.NormalizeTerminal(node.Label().Value()));
}
else
{
node.Label().SetValue(tn.NormalizeNonterminal(node.Label().Value()));
}
}
return(tn.NormalizeWholeTree(tree, tf));
}
/// <summary>Read parse trees from a Reader.</summary>
/// <param name="in">Reader</param>
/// <param name="tf">TreeFactory -- factory to create some kind of Tree</param>
/// <param name="tn">the method of normalizing trees</param>
/// <param name="st">Tokenizer that divides up Reader</param>
public PennTreeReader(Reader @in, ITreeFactory tf, TreeNormalizer tn, ITokenizer <string> st)
{
// misuse a list as a stack, since we want to avoid the synchronized and old Stack, but don't need the power and JDK 1.6 dependency of a Deque
reader = @in;
treeFactory = tf;
treeNormalizer = tn;
tokenizer = st;
// check for whacked out headers still present in Brown corpus in Treebank 3
string first = (st.MoveNext() ? st.Peek() : null);
if (first != null && first.StartsWith("*x*x*x"))
{
int foundCount = 0;
while (foundCount < 4 && st.MoveNext())
{
first = st.Current;
if (first != null && first.StartsWith("*x*x*x"))
{
foundCount++;
}
}
}
}
public LabeledScoredTreeReaderFactory(ILabelFactory lf, TreeNormalizer tm)
{
this.lf = lf;
this.tm = tm;
}
public LabeledScoredTreeReaderFactory(TreeNormalizer tm)
{
lf = CoreLabel.Factory();
this.tm = tm;
}
public LabeledScoredTreeReaderFactory(ILabelFactory lf)
{
this.lf = lf;
tm = new BobChrisTreeNormalizer();
}
/// <summary>Create a new TreeReaderFactory with CoreLabel labels.</summary>
public LabeledScoredTreeReaderFactory()
{
lf = CoreLabel.Factory();
tm = new BobChrisTreeNormalizer();
}
/// <summary>
/// Adds the given tree normalizer to this combination; the tree normalizers
/// are applied in the order they were added, with the first to be added being
/// the first to be applied.
/// </summary>
public virtual void AddTreeNormalizer(TreeNormalizer tn)
{
this.tns.Add(tn);
}
/// <summary>Read parse trees from a Reader.</summary>
/// <param name="in">Reader</param>
/// <param name="tf">TreeFactory -- factory to create some kind of Tree</param>
/// <param name="tn">the method of normalizing trees</param>
public PennTreeReader(Reader @in, ITreeFactory tf, TreeNormalizer tn)
: this(@in, tf, tn, new PennTreebankTokenizer(@in))
{
}
/// <summary>Create a new tree bank, using a specific TreeNormalizer.</summary>
/// <remarks>
/// Create a new tree bank, using a specific TreeNormalizer.
/// The trees are made with a <code>LabeledScoredTreeReaderFactory</code>.
/// <p/>
/// <i>Compatibility note: Until Sep 2004, this used to create a Treebank
/// with a SimpleTreeReaderFactory, but this was changed as the old
/// default wasn't very useful, especially to naive users.</i>
/// </remarks>
public MemoryTreebank(TreeNormalizer tm)
: this(new LabeledScoredTreeReaderFactory(tm))
{
}
/// <summary>
/// Specify your own
/// <see cref="ITreeFactory"/>
/// ;
/// uses a
/// <see cref="PennTreebankTokenizer"/>
/// , and a
/// <see cref="TreeNormalizer"/>
/// .
/// </summary>
/// <param name="tf">The TreeFactory to use in building Tree objects to return.</param>
/// <param name="tn">The TreeNormalizer to use</param>
public PennTreeReaderFactory(ITreeFactory tf, TreeNormalizer tn)
{
this.tf = tf;
this.tn = tn;
}
/// <summary>
/// Specify your own
/// <see cref="TreeNormalizer"/>
/// ;
/// uses a
/// <see cref="PennTreebankTokenizer"/>
/// , and a
/// <see cref="LabeledScoredTreeFactory"/>
/// .
/// </summary>
/// <param name="tn">The TreeNormalizer to use in building Tree objects to return.</param>
public PennTreeReaderFactory(TreeNormalizer tn)
: this(new LabeledScoredTreeFactory(), tn)
{
}
