/**
* Return information about the objects in this Tree.
*
* @param t The tree to examine.
* @return A human-readable String
*/
public static String toDebugStructureString(Tree t) {
StringBuilder sb = new StringBuilder();
String tCl = StringUtils.getShortClassName(t);
String tfCl = StringUtils.getShortClassName(t.treeFactory());
String lCl = StringUtils.getShortClassName(t.label());
String lfCl = StringUtils.getShortClassName(t.label().labelFactory());
Set<String> otherClasses = new HashSet<String>();
for (Tree st : t) {
String stCl = StringUtils.getShortClassName(st);
String stfCl = StringUtils.getShortClassName(st.treeFactory());
String slCl = StringUtils.getShortClassName(st.label());
String slfCl = StringUtils.getShortClassName(st.label().labelFactory());
if (!tCl.equals(stCl)) {
otherClasses.add(stCl);
}
if (!tfCl.equals(stfCl)) {
otherClasses.add(stfCl);
}
if (!lCl.equals(slCl)) {
otherClasses.add(slCl);
}
if (!lfCl.equals(slfCl)) {
otherClasses.add(slfCl);
}
}
sb.append("Tree with root of class ").append(tCl).append(" and factory ").append(tfCl);
sb.append(" with label class ").append(lCl).append(" and factory ").append(lfCl);
if (!otherClasses.isEmpty()) {
sb.append(" with the following classes also found within the tree: ").append(otherClasses);
}
return sb.toString();
}
Tree prune(Tree tree, int start) {
if (tree.isLeaf() || tree.isPreTerminal()) {
return tree;
}
// check each node's children for deletion
List<Tree> children = helper(tree.getChildrenAsList(), start);
children = prune(children, tree.label(), start, start + tree.yield().size());
return tree.treeFactory().newTreeNode(tree.label(), children);
}
public static Tree normalizeTree(Tree tree, TreeNormalizer tn, TreeFactory tf) {
for (Tree node : 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);
}
/**
* returns a list of categories that is the path from Tree from to Tree to within Tree root. If
* either from or to is not in root, returns null. Otherwise includes both from and to in the
* list.
*/
public static List<String> pathNodeToNode(Tree from, Tree to, Tree root) {
List<Tree> fromPath = pathFromRoot(from, root);
// System.out.println(treeListToCatList(fromPath));
if (fromPath == null) return null;
List<Tree> toPath = pathFromRoot(to, root);
// System.out.println(treeListToCatList(toPath));
if (toPath == null) return null;
// System.out.println(treeListToCatList(fromPath));
// System.out.println(treeListToCatList(toPath));
int last = 0;
int min = fromPath.size() <= toPath.size() ? fromPath.size() : toPath.size();
Tree lastNode = null;
// while((! (fromPath.isEmpty() || toPath.isEmpty())) &&
// fromPath.get(0).equals(toPath.get(0))) {
// lastNode = (Tree) fromPath.remove(0);
// toPath.remove(0);
// }
while (last < min && fromPath.get(last).equals(toPath.get(last))) {
lastNode = fromPath.get(last);
last++;
}
// System.out.println(treeListToCatList(fromPath));
// System.out.println(treeListToCatList(toPath));
List<String> totalPath = new ArrayList<String>();
for (int i = fromPath.size() - 1; i >= last; i--) {
Tree t = fromPath.get(i);
totalPath.add("up-" + t.label().value());
}
if (lastNode != null) totalPath.add("up-" + lastNode.label().value());
for (Tree t : toPath) totalPath.add("down-" + t.label().value());
// for(ListIterator i = fromPath.listIterator(fromPath.size()); i.hasPrevious(); ){
// Tree t = (Tree) i.previous();
// totalPath.add("up-" + t.label().value());
// }
// if(lastNode != null)
// totalPath.add("up-" + lastNode.label().value());
// for(ListIterator j = toPath.listIterator(); j.hasNext(); ){
// Tree t = (Tree) j.next();
// totalPath.add("down-" + t.label().value());
// }
return totalPath;
}
List<Tree> prune(List<Tree> treeList, Label label, int start, int end) {
// get reference tree
if (treeList.size() == 1) {
return treeList;
}
Tree testTree = treeList.get(0).treeFactory().newTreeNode(label, treeList);
int goal = Numberer.getGlobalNumberer("states").number(label.value());
Tree tempTree = parser.extractBestParse(goal, start, end);
// parser.restoreUnaries(tempTree);
Tree pcfgTree = debinarizer.transformTree(tempTree);
Set<Constituent> pcfgConstituents =
pcfgTree.constituents(new LabeledScoredConstituentFactory());
// delete child labels that are not in reference but do not cross reference
List<Tree> prunedChildren = new ArrayList<Tree>();
int childStart = 0;
for (int c = 0, numCh = testTree.numChildren(); c < numCh; c++) {
Tree child = testTree.getChild(c);
boolean isExtra = true;
int childEnd = childStart + child.yield().size();
Constituent childConstituent =
new LabeledScoredConstituent(childStart, childEnd, child.label(), 0);
if (pcfgConstituents.contains(childConstituent)) {
isExtra = false;
}
if (childConstituent.crosses(pcfgConstituents)) {
isExtra = false;
}
if (child.isLeaf() || child.isPreTerminal()) {
isExtra = false;
}
if (pcfgTree.yield().size() != testTree.yield().size()) {
isExtra = false;
}
if (!label.value().startsWith("NP^NP")) {
isExtra = false;
}
if (isExtra) {
System.err.println(
"Pruning: "
+ child.label()
+ " from "
+ (childStart + start)
+ " to "
+ (childEnd + start));
System.err.println("Was: " + testTree + " vs " + pcfgTree);
prunedChildren.addAll(child.getChildrenAsList());
} else {
prunedChildren.add(child);
}
childStart = childEnd;
}
return prunedChildren;
}
protected Rule ltToRule(Tree lt) {
if (lt.children().length == 1) {
UnaryRule ur = new UnaryRule();
ur.parent = stateNumberer.number(lt.label().value());
ur.child = stateNumberer.number(lt.children()[0].label().value());
return ur;
} else {
BinaryRule br = new BinaryRule();
br.parent = stateNumberer.number(lt.label().value());
br.leftChild = stateNumberer.number(lt.children()[0].label().value());
br.rightChild = stateNumberer.number(lt.children()[1].label().value());
return br;
}
}
/**
* returns the syntactic category of the tree as a list of the syntactic categories of the mother
* and the daughters
*/
public static List<String> localTreeAsCatList(Tree t) {
List<String> l = new ArrayList<String>(t.children().length + 1);
l.add(t.label().value());
for (int i = 0; i < t.children().length; i++) {
l.add(t.children()[i].label().value());
}
return l;
}
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;
}
private static void leafLabels(Tree t, List<Label> l) {
if (t.isLeaf()) {
l.add(t.label());
} else {
Tree[] kids = t.children();
for (int j = 0, n = kids.length; j < n; j++) {
leafLabels(kids[j], l);
}
}
}
private static void taggedLeafLabels(Tree t, List<CoreLabel> l) {
if (t.isPreTerminal()) {
CoreLabel fl = (CoreLabel) t.getChild(0).label();
fl.set(TagLabelAnnotation.class, t.label());
l.add(fl);
} else {
Tree[] kids = t.children();
for (int j = 0, n = kids.length; j < n; j++) {
taggedLeafLabels(kids[j], l);
}
}
}
protected void tallyTree(Tree t, LinkedList<String> parents) {
// traverse tree, building parent list
String str = t.label().value();
boolean strIsPassive = (str.indexOf('@') == -1);
if (strIsPassive) {
parents.addFirst(str);
}
if (!t.isLeaf()) {
if (!t.children()[0].isLeaf()) {
tallyInternalNode(t, parents);
for (int c = 0; c < t.children().length; c++) {
Tree child = t.children()[c];
tallyTree(child, parents);
}
} else {
tagNumberer.number(t.label().value());
}
}
if (strIsPassive) {
parents.removeFirst();
}
}
protected void tallyInternalNode(Tree lt, List parents) {
// form base rule
String label = lt.label().value();
Rule baseR = ltToRule(lt);
ruleToLabel.put(baseR, label);
// act on each history depth
for (int depth = 0, maxDepth = Math.min(HISTORY_DEPTH(), parents.size());
depth <= maxDepth;
depth++) {
List history = new ArrayList(parents.subList(0, depth));
// tally each history level / rewrite pair
rulePairs.incrementCount(new Pair(baseR, history), 1);
labelPairs.incrementCount(new Pair(label, history), 1);
}
}
private static int treeToLatexHelper(
Tree t, StringBuilder c, StringBuilder h, int n, int nextN, int indent) {
StringBuilder sb = new StringBuilder();
for (int i = 0; i < indent; i++) sb.append(" ");
h.append('\n').append(sb);
h.append("{\\")
.append(t.isLeaf() ? "" : "n")
.append("tnode{z")
.append(n)
.append("}{")
.append(t.label())
.append('}');
if (!t.isLeaf()) {
for (int k = 0; k < t.children().length; k++) {
h.append(", ");
c.append("\\nodeconnect{z").append(n).append("}{z").append(nextN).append("}\n");
nextN = treeToLatexHelper(t.children()[k], c, h, nextN, nextN + 1, indent + 1);
}
}
h.append('}');
return nextN;
}
private static int treeToLatexEvenHelper(
Tree t,
StringBuilder c,
StringBuilder h,
int n,
int nextN,
int indent,
int curDepth,
int maxDepth) {
StringBuilder sb = new StringBuilder();
for (int i = 0; i < indent; i++) sb.append(" ");
h.append('\n').append(sb);
int tDepth = t.depth();
if (tDepth == 0 && tDepth + curDepth < maxDepth) {
for (int pad = 0; pad < maxDepth - tDepth - curDepth; pad++) {
h.append("{\\ntnode{pad}{}, ");
}
}
h.append("{\\ntnode{z").append(n).append("}{").append(t.label()).append('}');
if (!t.isLeaf()) {
for (int k = 0; k < t.children().length; k++) {
h.append(", ");
c.append("\\nodeconnect{z").append(n).append("}{z").append(nextN).append("}\n");
nextN =
treeToLatexEvenHelper(
t.children()[k], c, h, nextN, nextN + 1, indent + 1, curDepth + 1, maxDepth);
}
}
if (tDepth == 0 && tDepth + curDepth < maxDepth) {
for (int pad = 0; pad < maxDepth - tDepth - curDepth; pad++) {
h.append('}');
}
}
h.append('}');
return nextN;
}