Also output the known tags in a dataset after the dataset has been retagged in the srparser

Author: AngledLuffa

src/edu/stanford/nlp/parser/shiftreduce/ShiftReduceParser.java

@@ -537,6 +537,7 @@ private void train(List<Pair<String, FileFilter>> trainTreebankPath,
       log.info("Retagging with tagger model: " + op.testOptions.taggerSerializedFile);
       log.info("Known tags in the tagger model: " + tagger.tagSet());
       redoTags(binarizedTrees, tagger, nThreads);
+      log.info("Tags in training set: " + Trees.uniqueTags(binarizedTrees));
       retagTimer.done("Retagging");
     }
 

src/edu/stanford/nlp/trees/Trees.java

@@ -186,6 +186,27 @@ private static void preTerminals(Tree t, List<Tree> l) {
     }
   }
 
+  public static Set<String> uniqueTags(List<Tree> trees) {
+    Set<String> allTags = new HashSet<>();
+    for (Tree tree : trees) {
+      uniqueTags(tree, allTags);
+    }
+    return allTags;
+  }
+
+  public static Set<String> uniqueTags(Tree tree) {
+    List<Label> labels = tree.preTerminalYield();
+    return uniqueTags(tree, new HashSet<>());
+  }
+
+  public static Set<String> uniqueTags(Tree tree, Set<String> tags) {
+    List<Label> labels = tree.preTerminalYield();
+    for (Label label : labels) {
+      tags.add(label.value());
+    }
+    return tags;
+  }
+
 
   /**
    * returns the labels of the leaves in a Tree in the order that they're found.