/// <summary>Loads treebank grammar from first argument and prints it.</summary>
/// <remarks>
/// Loads treebank grammar from first argument and prints it.
/// Just a demonstration of functionality. <br />
/// <code>usage: java MemoryTreebank treebankFilesPath</code>
/// </remarks>
/// <param name="args">array of command-line arguments</param>
public static void Main(string[] args)
{
Timing.StartTime();
Treebank treebank = new DiskTreebank(null);
Treebank treebank2 = new MemoryTreebank(null);
treebank.LoadPath(args[0]);
treebank2.LoadPath(args[0]);
CompositeTreebank c = new CompositeTreebank(treebank, treebank2);
Timing.EndTime();
ITreeTransformer myTransformer = new TransformingTreebank.MyTreeTransformer();
ITreeTransformer myTransformer2 = new TransformingTreebank.MyTreeTransformer2();
ITreeTransformer myTransformer3 = new TransformingTreebank.MyTreeTransformer3();
Treebank tf1 = c.Transform(myTransformer).Transform(myTransformer2).Transform(myTransformer3);
Treebank tf2 = new Edu.Stanford.Nlp.Trees.TransformingTreebank(new Edu.Stanford.Nlp.Trees.TransformingTreebank(new Edu.Stanford.Nlp.Trees.TransformingTreebank(c, myTransformer), myTransformer2), myTransformer3);
ITreeTransformer[] tta = new ITreeTransformer[] { myTransformer, myTransformer2, myTransformer3 };
ITreeTransformer tt3 = new CompositeTreeTransformer(Arrays.AsList(tta));
Treebank tf3 = c.Transform(tt3);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("COMPOSITE (DISK THEN MEMORY REPEATED VERSION OF) INPUT TREEBANK");
System.Console.Out.WriteLine(c);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("SLOWLY TRANSFORMED TREEBANK, USING TransformingTreebank() CONSTRUCTOR");
Treebank tx1 = new Edu.Stanford.Nlp.Trees.TransformingTreebank(c, myTransformer);
System.Console.Out.WriteLine(tx1);
System.Console.Out.WriteLine("-----");
Treebank tx2 = new Edu.Stanford.Nlp.Trees.TransformingTreebank(tx1, myTransformer2);
System.Console.Out.WriteLine(tx2);
System.Console.Out.WriteLine("-----");
Treebank tx3 = new Edu.Stanford.Nlp.Trees.TransformingTreebank(tx2, myTransformer3);
System.Console.Out.WriteLine(tx3);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("TRANSFORMED TREEBANK, USING Treebank.transform()");
System.Console.Out.WriteLine(tf1);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("PRINTING AGAIN TRANSFORMED TREEBANK, USING Treebank.transform()");
System.Console.Out.WriteLine(tf1);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("TRANSFORMED TREEBANK, USING TransformingTreebank() CONSTRUCTOR");
System.Console.Out.WriteLine(tf2);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("TRANSFORMED TREEBANK, USING CompositeTreeTransformer");
System.Console.Out.WriteLine(tf3);
System.Console.Out.WriteLine("-------------------------");
System.Console.Out.WriteLine("COMPOSITE (DISK THEN MEMORY REPEATED VERSION OF) INPUT TREEBANK");
System.Console.Out.WriteLine(c);
System.Console.Out.WriteLine("-------------------------");
}
private static void CountTaggings(Treebank tb, PrintWriter pw)
{
TwoDimensionalCounter <string, string> wtc = new TwoDimensionalCounter <string, string>();
tb.Apply(null);
foreach (string key in wtc.FirstKeySet())
{
pw.Print(key);
pw.Print('\t');
ICounter <string> ctr = wtc.GetCounter(key);
foreach (string k2 in ctr.KeySet())
{
pw.Print(k2 + '\t' + ctr.GetCount(k2) + '\t');
}
pw.Println();
}
}
private static void RunTiming(Treebank treebank)
{
System.Console.Out.WriteLine();
Timing.StartTime();
int num = 0;
foreach (Tree t in treebank)
{
num += t.Yield().Count;
}
Timing.EndTime("traversing corpus, counting words with iterator");
log.Info("There were " + num + " words in the treebank.");
treebank.Apply(new _ITreeVisitor_352());
// = 0;
log.Info();
Timing.EndTime("traversing corpus, counting words with TreeVisitor");
log.Info("There were " + num + " words in the treebank.");
log.Info();
Timing.StartTime();
log.Info("This treebank contains " + treebank.Count + " trees.");
Timing.EndTime("size of corpus");
}
// end main()
private static void PrintPunct(Treebank treebank, ITreebankLanguagePack tlp, PrintWriter pw)
{
if (tlp == null)
{
log.Info("The -punct option requires you to specify -tlp");
}
else
{
IPredicate <string> punctTagFilter = tlp.PunctuationTagAcceptFilter();
foreach (Tree t in treebank)
{
IList <TaggedWord> tws = t.TaggedYield();
foreach (TaggedWord tw in tws)
{
if (punctTagFilter.Test(tw.Tag()))
{
pw.Println(tw);
}
}
}
}
}
public FilteringTreebank(Treebank treebank, IPredicate <Tree> filter)
{
this.filter = filter;
this.treebank = treebank;
}
private static void SentenceLengths(Treebank treebank, string name, string range, PrintWriter pw)
{
int maxleng = 150;
int[] lengthCounts = new int[maxleng + 2];
int numSents = 0;
int longestSeen = 0;
int totalWords = 0;
string longSent = string.Empty;
double median = 0.0;
NumberFormat nf = new DecimalFormat("0.0");
bool foundMedian = false;
foreach (Tree t in treebank)
{
numSents++;
int len = t.Yield().Count;
if (len <= maxleng)
{
lengthCounts[len]++;
}
else
{
lengthCounts[maxleng + 1]++;
}
totalWords += len;
if (len > longestSeen)
{
longestSeen = len;
longSent = t.ToString();
}
}
System.Console.Out.Write("Files " + name + ' ');
if (range != null)
{
System.Console.Out.Write(range + ' ');
}
System.Console.Out.WriteLine("consists of " + numSents + " sentences");
int runningTotal = 0;
for (int i = 0; i <= maxleng; i++)
{
runningTotal += lengthCounts[i];
System.Console.Out.WriteLine(" " + lengthCounts[i] + " of length " + i + " (running total: " + runningTotal + ')');
if (!foundMedian && runningTotal > numSents / 2)
{
if (numSents % 2 == 0 && runningTotal == numSents / 2 + 1)
{
// right on the boundary
int j = i - 1;
while (j > 0 && lengthCounts[j] == 0)
{
j--;
}
median = ((double)i + j) / 2;
}
else
{
median = i;
}
foundMedian = true;
}
}
if (lengthCounts[maxleng + 1] > 0)
{
runningTotal += lengthCounts[maxleng + 1];
System.Console.Out.WriteLine(" " + lengthCounts[maxleng + 1] + " of length " + (maxleng + 1) + " to " + longestSeen + " (running total: " + runningTotal + ')');
}
System.Console.Out.WriteLine("Average length: " + nf.Format(((double)totalWords) / numSents) + "; median length: " + nf.Format(median));
System.Console.Out.WriteLine("Longest sentence is of length: " + longestSeen);
pw.Println(longSent);
}
/// <summary>
/// Create a new TransformingTreebank from a base Treebank that will
/// transform trees with the given TreeTransformer.
/// </summary>
/// <remarks>
/// Create a new TransformingTreebank from a base Treebank that will
/// transform trees with the given TreeTransformer.
/// This is the constructor that you should use.
/// </remarks>
/// <param name="tb">The base Treebank</param>
/// <param name="transformer">The TreeTransformer applied to each Tree.</param>
public TransformingTreebank(Treebank tb, ITreeTransformer transformer)
{
this.tb = tb;
this.transformer = transformer;
}
public CompositeTreebank(Treebank t1, Treebank t2)
{
this.t1 = t1;
this.t2 = t2;
}