public Tree transformTree(Tree tree) {
Label lab = tree.label();
if (tree.isLeaf()) {
Tree leaf = tf.newLeaf(lab);
leaf.setScore(tree.score());
return leaf;
}
String s = lab.value();
s = treebankLanguagePack().basicCategory(s);
s = treebankLanguagePack().stripGF(s);
int numKids = tree.numChildren();
List<Tree> children = new ArrayList<Tree>(numKids);
for (int cNum = 0; cNum < numKids; cNum++) {
Tree child = tree.getChild(cNum);
Tree newChild = transformTree(child);
children.add(newChild);
}
CategoryWordTag newLabel = new CategoryWordTag(lab);
newLabel.setCategory(s);
if (lab instanceof HasTag) {
String tag = ((HasTag) lab).tag();
tag = treebankLanguagePack().basicCategory(tag);
tag = treebankLanguagePack().stripGF(tag);
newLabel.setTag(tag);
}
Tree node = tf.newTreeNode(newLabel, children);
node.setScore(tree.score());
return node;
}
public Tree transformTree(Tree tree) {
Label lab = tree.label();
if (tree.isLeaf()) {
Tree leaf = tf.newLeaf(lab);
leaf.setScore(tree.score());
return leaf;
}
String s = lab.value();
s = treebankLanguagePack().basicCategory(s);
int numKids = tree.numChildren();
List<Tree> children = new ArrayList<Tree>(numKids);
for (int cNum = 0; cNum < numKids; cNum++) {
Tree child = tree.getChild(cNum);
Tree newChild = transformTree(child);
// cdm 2007: for just subcategory stripping, null shouldn't happen
// if (newChild != null) {
children.add(newChild);
// }
}
// if (children.isEmpty()) {
// return null;
// }
CategoryWordTag newLabel = new CategoryWordTag(lab);
newLabel.setCategory(s);
if (lab instanceof HasTag) {
String tag = ((HasTag) lab).tag();
tag = treebankLanguagePack().basicCategory(tag);
newLabel.setTag(tag);
}
Tree node = tf.newTreeNode(newLabel, children);
node.setScore(tree.score());
return node;
}
public static Tree copyHelper(Tree t, Map<Tree, Tree> newToOld, Map<Tree, Tree> oldToNew) {
Tree[] kids = t.children();
Tree[] newKids = new Tree[kids.length];
for (int i = 0, n = kids.length; i < n; i++) {
newKids[i] = copyHelper(kids[i], newToOld, oldToNew);
}
TreeFactory tf = t.treeFactory();
if (kids.length == 0) {
Tree newLeaf = tf.newLeaf(t.label());
newToOld.put(newLeaf, t);
oldToNew.put(newLeaf, t);
return newLeaf;
}
Tree newNode = tf.newTreeNode(t.label(), Arrays.asList(newKids));
newToOld.put(newNode, t);
oldToNew.put(t, newNode);
return newNode;
}
public static Tree untransformTree(Tree tree) {
TreeFactory tf = tree.treeFactory();
if (tree.isPrePreTerminal()) {
if (tree.firstChild().label().value().matches(".*_.")) {
StringBuilder word = new StringBuilder();
for (int i = 0; i < tree.children().length; i++) {
Tree child = tree.children()[i];
word.append(child.firstChild().label().value());
}
Tree newChild = tf.newLeaf(word.toString());
tree.setChildren(Collections.singletonList(newChild));
}
} else {
for (int i = 0; i < tree.children().length; i++) {
Tree child = tree.children()[i];
untransformTree(child);
}
}
return tree;
}
public Tree transformTree(Tree tree) {
TreeFactory tf = tree.treeFactory();
String tag = tree.label().value();
if (tree.isPreTerminal()) {
String word = tree.firstChild().label().value();
List<Tree> newPreterms = new ArrayList<>();
for (int i = 0, size = word.length(); i < size; i++) {
String singleCharLabel = new String(new char[] {word.charAt(i)});
Tree newLeaf = tf.newLeaf(singleCharLabel);
String suffix;
if (useTwoCharTags) {
if (word.length() == 1 || i == 0) {
suffix = "_S";
} else {
suffix = "_M";
}
} else {
if (word.length() == 1) {
suffix = "_S";
} else if (i == 0) {
suffix = "_B";
} else if (i == word.length() - 1) {
suffix = "_E";
} else {
suffix = "_M";
}
}
newPreterms.add(tf.newTreeNode(tag + suffix, Collections.<Tree>singletonList(newLeaf)));
}
return tf.newTreeNode(tag, newPreterms);
} else {
List<Tree> newChildren = new ArrayList<>();
for (int i = 0; i < tree.children().length; i++) {
Tree child = tree.children()[i];
newChildren.add(transformTree(child));
}
return tf.newTreeNode(tag, newChildren);
}
}
