// make a lookup table from symbols and their encodings
private void buildCode(String[] code, Node n, String path) {
// If n isn't a leaf, we need to go deeper
if (!n.isLeaf()) {
buildCode(code, n.left, path + '0');
buildCode(code, n.right, path + '1');
} else // if it's a leaf, we fill the case in the code table that correspond to this letter with
// the equivalent code that represent it
{
code[n.c] = path;
}
}
private void writeCode(Node n) throws IOException {
if (n.isLeaf()) {
output.write(1);
System.out.print(1);
output.write((int) n.c);
System.out.print((int) n.c);
} else {
output.write(0);
System.out.print(0);
writeCode(n.left);
writeCode(n.right);
}
}
private void print(Stack<Node> stack, Node n) {
if (n.isLeaf()) {
assert n.value != null;
for (Node node : stack) {
System.out.print(node.key + "->");
}
System.out.print(n.value);
System.out.println();
return;
}
for (Map.Entry<Character, Node> e : n.children.entrySet()) {
stack.push(e.getValue());
print(stack, e.getValue());
stack.pop();
}
}
/**
* Remove the node if possible or set its count to 0 if it has children and it needs to be kept
* around
*/
private Node tryRemove(Node node) {
if (node == null) {
return null;
}
if (node.weightedCount >= ZERO_WEIGHT_THRESHOLD) {
--nonZeroNodeCount;
}
weightedCount -= node.weightedCount;
Node result = null;
if (node.isLeaf()) {
--totalNodeCount;
} else if (node.hasSingleChild()) {
result = node.getSingleChild();
--totalNodeCount;
} else {
node.weightedCount = 0;
result = node;
}
return result;
}
public void outputToJSON(String filePath, int numTopWords, double boundaryFactor) {
System.out.println("Writing down the model...");
JSONObject featureCollection = new JSONObject();
featureCollection.put("type", "FeatureCollection");
JSONArray features = new JSONArray();
Queue<Node> nodeQueue = new LinkedList<>();
nodeQueue.offer(root);
while (!nodeQueue.isEmpty()) {
Node currentNode = nodeQueue.poll();
JSONObject feature = new JSONObject();
feature.put("type", "Feature");
feature.put("id", currentNode.hashCode());
JSONObject properties = new JSONObject();
properties.put("level", currentNode.level + 1);
properties.put("num_documents", currentNode.numCustomers);
if (currentNode.parent != null) {
properties.put("parent", currentNode.parent.hashCode());
}
if (!currentNode.isLeaf()) {
JSONArray children = new JSONArray();
for (Node child : currentNode.children) {
children.put(child.hashCode());
}
properties.put("children", children);
}
JSONArray center = new JSONArray();
double longitude = currentNode.location.longitude;
center.put(longitude);
double latitude = currentNode.location.latitude;
center.put(latitude);
properties.put("center", center);
JSONArray deviation = new JSONArray();
double longitudeDeviation = Math.sqrt(currentNode.location.longitudeVariance);
deviation.put(longitudeDeviation);
double latitudeDeviation = Math.sqrt(currentNode.location.latitudeVariance);
deviation.put(latitudeDeviation);
properties.put("deviation", deviation);
JSONArray topWords = new JSONArray();
PriorityQueue<Map.Entry<Integer, Integer>> wordMinHeap =
new PriorityQueue<>(
numTopWords,
new Comparator<Map.Entry<Integer, Integer>>() {
@Override
public int compare(Map.Entry<Integer, Integer> o1, Map.Entry<Integer, Integer> o2) {
return o1.getValue() - o2.getValue();
}
});
for (Map.Entry<Integer, Integer> entry : currentNode.wordCounts.entrySet()) {
if (wordMinHeap.size() < numTopWords) {
wordMinHeap.offer(entry);
} else {
if (entry.getValue() > wordMinHeap.peek().getValue()) {
wordMinHeap.poll();
wordMinHeap.offer(entry);
}
}
}
while (!wordMinHeap.isEmpty()) {
topWords.put(this.id2Word.get(wordMinHeap.poll().getKey()));
}
properties.put("top_words", topWords);
feature.put("properties", properties);
JSONObject geometry = new JSONObject();
geometry.put("type", "Polygon");
JSONArray coordinates = new JSONArray();
JSONArray boundary = new JSONArray();
double east = longitude + boundaryFactor * longitudeDeviation;
double west = longitude - boundaryFactor * longitudeDeviation;
double north = latitude + boundaryFactor * latitudeDeviation;
double south = latitude - boundaryFactor * latitudeDeviation;
boundary.put(new JSONArray(new double[] {west, north}));
boundary.put(new JSONArray(new double[] {east, north}));
boundary.put(new JSONArray(new double[] {east, south}));
boundary.put(new JSONArray(new double[] {west, south}));
boundary.put(new JSONArray(new double[] {west, north}));
coordinates.put(boundary);
geometry.put("coordinates", coordinates);
feature.put("geometry", geometry);
features.put(feature);
for (Node child : currentNode.children) {
nodeQueue.offer(child);
}
}
featureCollection.put("features", features);
try {
PrintWriter writer = new PrintWriter(filePath);
featureCollection.write(writer);
writer.close();
} catch (FileNotFoundException e) {
System.err.println("Model JSON cannot be created.");
}
}