/**
* 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;
}
/**
* Returns the index of <code>daughter</code> in <code>parent</code> by ==. Returns -1 if <code>
* daughter</code> not found.
*/
public static int objectEqualityIndexOf(Tree parent, Tree daughter) {
for (int i = 0; i < parent.children().length; i++) {
if (daughter == parent.children()[i]) {
return i;
}
}
return -1;
}
private List<Tree> helper(List<Tree> treeList, int start) {
List<Tree> newTreeList = new ArrayList<Tree>(treeList.size());
for (Tree tree : treeList) {
int end = start + tree.yield().size();
newTreeList.add(prune(tree, start));
start = end;
}
return newTreeList;
}
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 preTerminals(Tree t, List<Tree> l) {
if (t.isPreTerminal()) {
l.add(t);
} else {
Tree[] kids = t.children();
for (int j = 0, n = kids.length; j < n; j++) {
preTerminals(kids[j], l);
}
}
}
public static ArrayList<TaggedWord> StanfordParse(String sentence, LexicalizedParser lp) {
TokenizerFactory<CoreLabel> tokenizerFactory =
PTBTokenizer.factory(new CoreLabelTokenFactory(), "");
List<CoreLabel> rawWords2 =
tokenizerFactory.getTokenizer(new StringReader(sentence)).tokenize();
Tree parse = lp.apply(rawWords2);
ArrayList<TaggedWord> taggedWords = parse.taggedYield();
return taggedWords;
}
/**
* Gets the <i>i</i>th leaf of a tree from the left. The leftmost leaf is numbered 0.
*
* @return The <i>i</i><sup>th</sup> leaf as a Tree, or <code>null</code> if there is no such
* leaf.
*/
public static Tree getLeaf(Tree tree, int i) {
int count = -1;
for (Tree next : tree) {
if (next.isLeaf()) {
count++;
}
if (count == i) {
return next;
}
}
return null;
}
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);
}
}
}
/** replaces all instances (by ==) of node with node1. Doesn't affect the node t itself */
public static void replaceNode(Tree node, Tree node1, Tree t) {
if (t.isLeaf()) return;
Tree[] kids = t.children();
List<Tree> newKids = new ArrayList<Tree>(kids.length);
for (int i = 0, n = kids.length; i < n; i++) {
if (kids[i] != node) {
newKids.add(kids[i]);
replaceNode(node, node1, kids[i]);
} else {
newKids.add(node1);
}
}
t.setChildren(newKids);
}
/**
* returns the maximal projection of <code>head</code> in <code>root</code> given a {@link
* HeadFinder}
*/
public static Tree maximalProjection(Tree head, Tree root, HeadFinder hf) {
Tree projection = head;
if (projection == root) {
return root;
}
Tree parent = projection.parent(root);
while (hf.determineHead(parent) == projection) {
projection = parent;
if (projection == root) {
return root;
}
parent = projection.parent(root);
}
return projection;
}
private ArrayList<TaggedWord> parseSentenceTD(String text) {
System.out.println("Parsing sentence...");
ArrayList<TaggedWord> tw = new ArrayList<TaggedWord>();
Reader reader = new StringReader(text);
for (List<HasWord> sentence : new DocumentPreprocessor(reader)) {
Tree parse = lp.apply(sentence);
tw = parse.taggedYield();
}
return tw;
}
/** returns true iff <code>head</code> (transitively) heads <code>node</code> */
public static boolean heads(Tree head, Tree node, HeadFinder hf) {
if (node.isLeaf()) {
return false;
} else {
return heads(head, hf.determineHead(node), hf);
}
}
public static String treeToLatexEven(Tree t) {
StringBuilder connections = new StringBuilder();
StringBuilder hierarchy = new StringBuilder();
int maxDepth = t.depth();
treeToLatexEvenHelper(t, connections, hierarchy, 0, 1, 0, 0, maxDepth);
return "\\tree" + hierarchy + '\n' + connections + '\n';
}
static boolean rightEdge(Tree t, Tree t1, MutableInteger i) {
if (t == t1) {
return true;
} else if (t1.isLeaf()) {
int j = t1.yield().size(); // so that empties don't add size
i.set(i.intValue() - j);
return false;
} else {
Tree[] kids = t1.children();
for (int j = kids.length - 1; j >= 0; j--) {
if (rightEdge(t, kids[j], i)) {
return true;
}
}
return false;
}
}
public static ArrayList<ArrayList<TaggedWord>> getPhrases(Tree parse, int phraseSizeLimit) {
ArrayList<ArrayList<TaggedWord>> newList = new ArrayList<ArrayList<TaggedWord>>();
List<Tree> leaves = parse.getLeaves();
if (leaves.size() <= phraseSizeLimit) {
// ArrayList<TaggedWord> phraseElements = PreprocessPhrase(parse.taggedYield());
ArrayList<TaggedWord> phraseElements = Preprocess(parse.taggedYield());
if (phraseElements.size() > 0) newList.add(phraseElements);
} else {
Tree[] childrenNodes = parse.children();
for (int i = 0; i < childrenNodes.length; i++) {
Tree currentParse = childrenNodes[i];
newList.addAll(getPhrases(currentParse, phraseSizeLimit));
}
}
return newList;
}
/**
* 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();
}
/**
* Returns the positional index of the right edge of a tree <i>t</i> within a given root, as
* defined by the size of the yield of all material preceding <i>t</i> plus all the material
* contained in <i>t</i>.
*/
public static int rightEdge(Tree t, Tree root) {
MutableInteger i = new MutableInteger(root.yield().size());
if (rightEdge(t, root, i)) {
return i.intValue();
} else {
throw new RuntimeException("Tree is not a descendent of root.");
// return root.yield().size() + 1;
}
}
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 void main(String args[]) {
// String sentence1 = "A large bird standing on a table picks up a plastic glass
// containing liquid and places it in a bowl of something.";
// String sentence2 = "A bird picks up a plastic cup containing a liquid with it's beak
// and puts the cup into a bowl.";
// LexicalizedParser lp = new LexicalizedParser("englishPCFG.ser.gz");
// LeskWSD tm = new LeskWSD(lp);
// WordNetSimilarity ws = new WordNetSimilarity();
//
// System.out.println(LexicalSimilarityScoreWordNet(sentence1, sentence2, tm, lp, ws));
String sentence =
"The broader Standard & Poor's 500 Index <.SPX> shed 2.38 points, or 0.24 percent, at 995.10.";
LexicalizedParser lp = new LexicalizedParser("englishPCFG.ser.gz");
Tree parse = lp.apply(sentence);
ArrayList<TaggedWord> taggedWords = parse.taggedYield();
taggedWords = Preprocess(taggedWords);
for (int i = 0; i < taggedWords.size(); i++) System.out.println(taggedWords.get(i).word());
}
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);
}
/* applies a TreeVisitor to all projections (including the node itself) of a node in a Tree.
* Does nothing if head is not in root.
* @return the maximal projection of head in root.
*/
public static Tree applyToProjections(TreeVisitor v, Tree head, Tree root, HeadFinder hf) {
Tree projection = head;
Tree parent = projection.parent(root);
if (parent == null && projection != root) {
return null;
}
v.visitTree(projection);
if (projection == root) {
return root;
}
while (hf.determineHead(parent) == projection) {
projection = parent;
v.visitTree(projection);
if (projection == root) {
return root;
}
parent = projection.parent(root);
}
return projection;
}
/**
* returns list of tree nodes to root from t. Includes root and t. Returns null if tree not found
* dominated by root
*/
public static List<Tree> pathFromRoot(Tree t, Tree root) {
if (t == root) {
// if (t.equals(root)) {
List<Tree> l = new ArrayList<Tree>(1);
l.add(t);
return l;
} else if (root == null) {
return null;
}
return root.dominationPath(t);
}
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;
}
static Tree getPreTerminal(Tree tree, MutableInteger i, int n) {
if (i.intValue() == n) {
if (tree.isPreTerminal()) {
return tree;
} else {
return getPreTerminal(tree.children()[0], i, n);
}
} else {
if (tree.isPreTerminal()) {
i.set(i.intValue() + tree.yield().size());
return null;
} else {
Tree[] kids = tree.children();
for (int j = 0; j < kids.length; j++) {
Tree result = getPreTerminal(kids[j], i, n);
if (result != null) {
return result;
}
}
return null;
}
}
}
/**
* traceTo() values that are contained in the tree are assigned to new objects. traceTo() values
* that are not contained in the tree are given the old value.
*/
public static void fixEmptyTreeLeafs(Tree t, Map<Tree, Tree> newToOld, Map<Tree, Tree> oldToNew) {
Tree[] kids = t.children();
for (int i = 0, n = kids.length; i < n; i++) {
fixEmptyTreeLeafs(kids[i], newToOld, oldToNew);
}
if (t instanceof EmptyTreeLeaf) {
EmptyTreeLeaf oldT = (EmptyTreeLeaf) newToOld.get(t);
((EmptyTreeLeaf) t).setEmptyType(oldT.emptyType());
Tree oldTraceTo = oldT.traceTo();
Tree newTraceTo = oldToNew.get(oldTraceTo);
if (newTraceTo != null) ((EmptyTreeLeaf) t).setTraceTo(newTraceTo);
else ((EmptyTreeLeaf) t).setTraceTo(oldTraceTo);
}
}
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 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;
}
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;
}
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;
}
}
