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 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;
}
/**
* Add -TMP when not present within an NP
*
* @param tree The tree to add temporal info to.
*/
private void addTMP9(final Tree tree) {
// do the head chain under it
Tree ht = headFinder.determineHead(tree);
// special fix for possessives! -- make noun before head
if (ht.value().equals("POS")) {
int j = tree.objectIndexOf(ht);
if (j > 0) {
ht = tree.getChild(j - 1);
}
}
// Note: this next bit changes the tree label, rather
// than creating a new tree node. Beware!
if (ht.isPreTerminal()
|| ht.value().startsWith("NP")
|| ht.value().startsWith("PP")
|| ht.value().startsWith("ADVP")) {
if (!TmpPattern.matcher(ht.value()).matches()) {
LabelFactory lf = ht.labelFactory();
// System.err.println("TMP: Changing " + ht.value() + " to " +
// ht.value() + "-TMP");
ht.setLabel(lf.newLabel(ht.value() + "-TMP"));
}
if (ht.value().startsWith("NP")
|| ht.value().startsWith("PP")
|| ht.value().startsWith("ADVP")) {
addTMP9(ht);
}
}
// do the NPs under it (which may or may not be the head chain
Tree[] kidlets = tree.children();
for (int k = 0; k < kidlets.length; k++) {
ht = kidlets[k];
LabelFactory lf;
if (tree.isPrePreTerminal() && !TmpPattern.matcher(ht.value()).matches()) {
// System.err.println("TMP: Changing " + ht.value() + " to " +
// ht.value() + "-TMP");
lf = ht.labelFactory();
// Note: this next bit changes the tree label, rather
// than creating a new tree node. Beware!
ht.setLabel(lf.newLabel(ht.value() + "-TMP"));
} else if (ht.value().startsWith("NP")) {
// don't add -TMP twice!
if (!TmpPattern.matcher(ht.value()).matches()) {
lf = ht.labelFactory();
// System.err.println("TMP: Changing " + ht.value() + " to " +
// ht.value() + "-TMP");
// Note: this next bit changes the tree label, rather
// than creating a new tree node. Beware!
ht.setLabel(lf.newLabel(ht.value() + "-TMP"));
}
addTMP9(ht);
}
}
}
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;
}
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);
}
}
}
/** 将决策树输出到标准输出 */
static void outputDecisionTree(Object obj, int level, Object from) {
for (int i = 0; i < level; i++) System.out.print("|-----");
if (from != null) System.out.printf("(%s):", from);
if (obj instanceof Tree) {
Tree tree = (Tree) obj;
String attrName = tree.getAttribute();
System.out.printf("[%s = ?]\n", attrName);
for (Object attrValue : tree.getAttributeValues()) {
Object child = tree.getChild(attrValue);
outputDecisionTree(child, level + 1, attrName + " = " + attrValue);
}
} else {
System.out.printf("[CATEGORY = %s]\n", obj);
}
}
// process a statement of the form: connect hostname/port
private void executeConnect(CommonTree ast) {
int portNumber = Integer.parseInt(ast.getChild(1).getText());
Tree idList = ast.getChild(0);
StringBuilder hostName = new StringBuilder();
int idCount = idList.getChildCount();
for (int idx = 0; idx < idCount; idx++) {
hostName.append(idList.getChild(idx).getText());
}
// disconnect current connection, if any.
// This is a no-op, if you aren't currently connected.
CliMain.disconnect();
// now, connect to the newly specified host name and port
css_.hostName = hostName.toString();
css_.thriftPort = portNumber;
CliMain.connect(css_.hostName, css_.thriftPort);
}
private boolean step(Tree code) {
if (code.getData().equals(FUN)) return false;
if (code.getData().equals("")
&& code.getChildren().size() >= 1
&& code.getChild(0).getData().equals(FUN)) {
// This is an APPLY -- make sure it works.
Tree fn = code.getChild(0);
Vector<Tree> fnargs = fn.getChildren();
Tree body = fnargs.get(1);
fnargs = fnargs.get(0).getChildren();
Vector<Tree> inargs = code.getChildren();
inargs.remove(0);
for (Tree t : inargs) {
if (step(t)) return true;
}
if (fnargs.size() != inargs.size()) {
return false;
}
for (int i = 0; i < fnargs.size(); i++) {
replace(body, fnargs.get(i).getData(), inargs.get(i));
}
code.removeChildren(false);
for (Tree t : body.getChildren()) {
code.addChild(t, false);
}
code.setData(body.getData());
return true;
} else if (code.getData().equals(IF)) {
// IF
Vector<Tree> ifargs = code.getChildren();
if (step(ifargs.get(0))) return true;
if (ifargs.get(0).getData().equals("0")) { // "0" is our only false value
System.out.println("FALSE IF");
code.removeChildren(false);
for (Tree kkid : ifargs.get(2).getChildren()) {
code.addChild(kkid, false);
}
code.setData(ifargs.get(2).getData());
} else {
System.out.println("TRUE IF: " + ifargs.get(0));
code.removeChildren(false);
for (Tree kkid : ifargs.get(1).getChildren()) {
code.addChild(kkid, false);
}
code.setData(ifargs.get(1).getData());
}
return true;
}
for (Tree kid : code.getChildren()) {
if (step(kid)) {
System.out.println("stepped " + kid);
return true;
}
}
if (code.getData().equals("<")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
if (left < right) {
code.setData("1");
} else {
code.setData("0");
}
return true;
} else if (code.getData().equals(">")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
if (left > right) {
code.setData("1");
} else {
code.setData("0");
}
return true;
} else if (code.getData().equals("=")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
if (left.equals(right)) {
code.setData("1");
} else {
code.setData("0");
}
return true;
} else if (code.getData().equals("+")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
code.setData("" + (left + right));
return true;
} else if (code.getData().equals("-")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
code.setData("" + (left - right));
return true;
} else if (code.getData().equals("*")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
code.setData("" + (left * right));
return true;
} else if (code.getData().equals("/")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
code.setData("" + (left / right));
return true;
} else if (code.getData().equals("^")) {
Vector<Tree> ltargs = code.getChildren();
Double left = Double.parseDouble(ltargs.get(0).getData());
Double right = Double.parseDouble(ltargs.get(1).getData());
code.removeChildren(false);
code.setData("" + Math.pow(left, right));
return true;
}
System.out.println("Couldn't step: " + code.getData());
return false;
}