List of usage examples for org.apache.hadoop.util IndexedSorter sort
void sort(IndexedSortable s, int l, int r);
From source file:com.ricemap.spateDB.core.GlobalIndex.java
License:Apache License
public int knn(final double qt, final double qx, final double qy, int k, ResultCollector2<S, Double> output) { double query_area = ((getMBR().t2 - getMBR().t1) * (getMBR().x2 - getMBR().x1) * (getMBR().y2 - getMBR().y1)) * k / size(); double query_radius = Math.sqrt(query_area / Math.PI); boolean result_correct; final Vector<Double> distances = new Vector<Double>(); final Vector<S> shapes = new Vector<S>(); // Find results in the range and increase this range if needed to ensure // correctness of the answer do {/*ww w. jav a 2 s . c o m*/ // Initialize result and query range distances.clear(); shapes.clear(); Prism queryRange = new Prism(); queryRange.t1 = qt - query_radius / 2; queryRange.x1 = qx - query_radius / 2; queryRange.y1 = qy - query_radius / 2; queryRange.t2 = qt + query_radius / 2; queryRange.x2 = qx + query_radius / 2; queryRange.y2 = qy + query_radius / 2; // Retrieve all results in range rangeQuery(queryRange, new ResultCollector<S>() { @Override public void collect(S shape) { distances.add(shape.distanceTo(qt, qx, qy)); shapes.add((S) shape.clone()); } }); if (shapes.size() <= k) { // Didn't find k elements in range, double the range to get more // items if (shapes.size() == size() || shapes.size() == k) { // Already returned all possible elements result_correct = true; } else { query_radius *= 2; result_correct = false; } } else { // Sort items by distance to get the kth neighbor IndexedSortable s = new IndexedSortable() { @Override public void swap(int i, int j) { double temp_distance = distances.elementAt(i); distances.set(i, distances.elementAt(j)); distances.set(j, temp_distance); S temp_shape = shapes.elementAt(i); shapes.set(i, shapes.elementAt(j)); shapes.set(j, temp_shape); } @Override public int compare(int i, int j) { // Note. Equality is not important to check because // items with the // same distance can be ordered anyway. if (distances.elementAt(i) < distances.elementAt(j)) return -1; return 1; } }; IndexedSorter sorter = new QuickSort(); sorter.sort(s, 0, shapes.size()); if (distances.elementAt(k - 1) > query_radius) { result_correct = false; query_radius = distances.elementAt(k); } else { result_correct = true; } } } while (!result_correct); int result_size = Math.min(k, shapes.size()); if (output != null) { for (int i = 0; i < result_size; i++) { output.collect(shapes.elementAt(i), distances.elementAt(i)); } } return result_size; }
From source file:com.ricemap.spateDB.core.RTree.java
License:Apache License
/** * Builds the RTree given a serialized list of elements. It uses the given * stockObject to deserialize these elements and build the tree. Also writes * the created tree to the disk directly. * //from w w w .j av a 2 s . c om * @param elements * - serialization of elements to be written * @param offset * - index of the first element to use in the elements array * @param len * - number of bytes to user from the elements array * @param bytesAvailable * - size available (in bytes) to store the tree structures * @param dataOut * - an output to use for writing the tree to * @param fast_sort * - setting this to <code>true</code> allows the method to run * faster by materializing the offset of each element in the list * which speeds up the comparison. However, this requires an * additional 16 bytes per element. So, for each 1M elements, the * method will require an additional 16 M bytes (approximately). */ public void bulkLoadWrite(final byte[] element_bytes, final int offset, final int len, final int degree, DataOutput dataOut, final boolean fast_sort, final boolean columnarStorage) { try { columnar = columnarStorage; //TODO: the order of fields should be stable under Oracle JVM, but not guaranteed Field[] fields = stockObject.getClass().getDeclaredFields(); // Count number of elements in the given text int i_start = offset; final Text line = new Text(); while (i_start < offset + len) { int i_end = skipToEOL(element_bytes, i_start); // Extract the line without end of line character line.set(element_bytes, i_start, i_end - i_start - 1); stockObject.fromText(line); elementCount++; i_start = i_end; } LOG.info("Bulk loading an RTree with " + elementCount + " elements"); // It turns out the findBestDegree returns the best degree when the // whole // tree is loaded to memory when processed. However, as current // algorithms // process the tree while it's on disk, a higher degree should be // selected // such that a node fits one file block (assumed to be 4K). // final int degree = findBestDegree(bytesAvailable, elementCount); LOG.info("Writing an RTree with degree " + degree); int height = Math.max(1, (int) Math.ceil(Math.log(elementCount) / Math.log(degree))); int leafNodeCount = (int) Math.pow(degree, height - 1); if (elementCount < 2 * leafNodeCount && height > 1) { height--; leafNodeCount = (int) Math.pow(degree, height - 1); } int nodeCount = (int) ((Math.pow(degree, height) - 1) / (degree - 1)); int nonLeafNodeCount = nodeCount - leafNodeCount; // Keep track of the offset of each element in the text final int[] offsets = new int[elementCount]; final int[] ids = new int[elementCount]; final double[] ts = fast_sort ? new double[elementCount] : null; final double[] xs = fast_sort ? new double[elementCount] : null; final double[] ys = fast_sort ? new double[elementCount] : null; //initialize columnar data output ByteArrayOutputStream index_bos = new ByteArrayOutputStream(); DataOutputStream index_dos = new DataOutputStream(index_bos); ByteArrayOutputStream[] bos = new ByteArrayOutputStream[fields.length]; DataOutputStream[] dos = new DataOutputStream[fields.length]; for (int i = 0; i < bos.length; i++) { bos[i] = new ByteArrayOutputStream(); dos[i] = new DataOutputStream(bos[i]); } i_start = offset; line.clear(); for (int i = 0; i < elementCount; i++) { offsets[i] = i_start; ids[i] = i; int i_end = skipToEOL(element_bytes, i_start); if (xs != null) { // Extract the line with end of line character line.set(element_bytes, i_start, i_end - i_start - 1); stockObject.fromText(line); // Sample center of the shape ts[i] = (stockObject.getMBR().t1 + stockObject.getMBR().t2) / 2; xs[i] = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; ys[i] = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; //build columnar storage if (stockObject instanceof Point3d) { index_dos.writeDouble(ts[i]); index_dos.writeDouble(xs[i]); index_dos.writeDouble(ys[i]); } else { throw new RuntimeException("Indexing non-point shape with RTREE is not supported yet"); } for (int j = 0; j < fields.length; j++) { if (fields[j].getType().equals(Integer.TYPE)) { dos[j].writeInt(fields[j].getInt(stockObject)); } else if (fields[j].getType().equals(Double.TYPE)) { dos[j].writeDouble(fields[j].getDouble(stockObject)); } else if (fields[j].getType().equals(Long.TYPE)) { dos[j].writeLong(fields[j].getLong(stockObject)); } else { continue; //throw new RuntimeException("Field type is not supported yet"); } } } i_start = i_end; } index_dos.close(); for (int i = 0; i < dos.length; i++) { dos[i].close(); } /** A struct to store information about a split */ class SplitStruct extends Prism { /** Start and end index for this split */ int index1, index2; /** Direction of this split */ byte direction; /** Index of first element on disk */ int offsetOfFirstElement; static final byte DIRECTION_T = 0; static final byte DIRECTION_X = 1; static final byte DIRECTION_Y = 2; SplitStruct(int index1, int index2, byte direction) { this.index1 = index1; this.index2 = index2; this.direction = direction; } @Override public void write(DataOutput out) throws IOException { // if (columnarStorage) out.writeInt(index1); else out.writeInt(offsetOfFirstElement); super.write(out); } void partition(Queue<SplitStruct> toBePartitioned) { IndexedSortable sortableT; IndexedSortable sortableX; IndexedSortable sortableY; if (fast_sort) { // Use materialized xs[] and ys[] to do the comparisons sortableT = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap ts double tempt = ts[i]; ts[i] = ts[j]; ts[j] = tempt; // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; tempid = ids[i]; ids[i] = ids[j]; ids[j] = tempid; } @Override public int compare(int i, int j) { if (ts[i] < ts[j]) return -1; if (ts[i] > ts[j]) return 1; return 0; } }; sortableX = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap ts double tempt = ts[i]; ts[i] = ts[j]; ts[j] = tempt; // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; tempid = ids[i]; ids[i] = ids[j]; ids[j] = tempid; } @Override public int compare(int i, int j) { if (ts[i] < ts[j]) return -1; if (xs[i] < xs[j]) return -1; if (xs[i] > xs[j]) return 1; return 0; } }; sortableY = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap ts double tempt = ts[i]; ts[i] = ts[j]; ts[j] = tempt; // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; tempid = ids[i]; ids[i] = ids[j]; ids[j] = tempid; } @Override public int compare(int i, int j) { if (ys[i] < ys[j]) return -1; if (ys[i] > ys[j]) return 1; return 0; } }; } else { // No materialized xs and ys. Always deserialize objects // to compare sortableT = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; tempid = ids[i]; ids[i] = ids[j]; ids[j] = tempid; } @Override public int compare(int i, int j) { // Get end of line int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double ti = (stockObject.getMBR().t1 + stockObject.getMBR().t2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double tj = (stockObject.getMBR().t1 + stockObject.getMBR().t2) / 2; if (ti < tj) return -1; if (ti > tj) return 1; return 0; } }; sortableX = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; tempid = ids[i]; ids[i] = ids[j]; ids[j] = tempid; } @Override public int compare(int i, int j) { // Get end of line int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double xi = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double xj = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; if (xi < xj) return -1; if (xi > xj) return 1; return 0; } }; sortableY = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; tempid = ids[i]; ids[i] = ids[j]; ids[j] = tempid; } @Override public int compare(int i, int j) { int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double yi = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double yj = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; if (yi < yj) return -1; if (yi > yj) return 1; return 0; } }; } final IndexedSorter sorter = new QuickSort(); final IndexedSortable[] sortables = new IndexedSortable[3]; sortables[SplitStruct.DIRECTION_T] = sortableT; sortables[SplitStruct.DIRECTION_X] = sortableX; sortables[SplitStruct.DIRECTION_Y] = sortableY; sorter.sort(sortables[direction], index1, index2); // Partition into maxEntries partitions (equally) and // create a SplitStruct for each partition int i1 = index1; for (int iSplit = 0; iSplit < degree; iSplit++) { int i2 = index1 + (index2 - index1) * (iSplit + 1) / degree; SplitStruct newSplit; if (direction == 0) { newSplit = new SplitStruct(i1, i2, (byte) 1); } else if (direction == 1) { newSplit = new SplitStruct(i1, i2, (byte) 2); } else { newSplit = new SplitStruct(i1, i2, (byte) 0); } toBePartitioned.add(newSplit); i1 = i2; } } } // All nodes stored in level-order traversal Vector<SplitStruct> nodes = new Vector<SplitStruct>(); final Queue<SplitStruct> toBePartitioned = new LinkedList<SplitStruct>(); toBePartitioned.add(new SplitStruct(0, elementCount, SplitStruct.DIRECTION_X)); while (!toBePartitioned.isEmpty()) { SplitStruct split = toBePartitioned.poll(); if (nodes.size() < nonLeafNodeCount) { // This is a non-leaf split.partition(toBePartitioned); } nodes.add(split); } if (nodes.size() != nodeCount) { throw new RuntimeException( "Expected node count: " + nodeCount + ". Real node count: " + nodes.size()); } // Now we have our data sorted in the required order. Start building // the tree. // Store the offset of each leaf node in the tree FSDataOutputStream fakeOut = new FSDataOutputStream(new java.io.OutputStream() { // Null output stream @Override public void write(int b) throws IOException { // Do nothing } @Override public void write(byte[] b, int off, int len) throws IOException { // Do nothing } @Override public void write(byte[] b) throws IOException { // Do nothing } }, null, TreeHeaderSize + nodes.size() * NodeSize); for (int i_leaf = nonLeafNodeCount, i = 0; i_leaf < nodes.size(); i_leaf++) { nodes.elementAt(i_leaf).offsetOfFirstElement = (int) fakeOut.getPos(); if (i != nodes.elementAt(i_leaf).index1) throw new RuntimeException(); double t1, x1, y1, t2, x2, y2; // Initialize MBR to first object int eol = skipToEOL(element_bytes, offsets[i]); fakeOut.write(element_bytes, offsets[i], eol - offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); Prism mbr = stockObject.getMBR(); t1 = mbr.t1; x1 = mbr.x1; y1 = mbr.y1; t2 = mbr.t2; x2 = mbr.x2; y2 = mbr.y2; i++; while (i < nodes.elementAt(i_leaf).index2) { eol = skipToEOL(element_bytes, offsets[i]); fakeOut.write(element_bytes, offsets[i], eol - offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); mbr = stockObject.getMBR(); if (mbr.t1 < t1) t1 = mbr.t1; if (mbr.x1 < x1) x1 = mbr.x1; if (mbr.y1 < y1) y1 = mbr.y1; if (mbr.t2 > t2) t2 = mbr.t2; if (mbr.x2 > x2) x2 = mbr.x2; if (mbr.y2 > y2) y2 = mbr.y2; i++; } nodes.elementAt(i_leaf).set(t1, x1, y1, t2, x2, y2); } fakeOut.close(); fakeOut = null; // Calculate MBR and offsetOfFirstElement for non-leaves for (int i_node = nonLeafNodeCount - 1; i_node >= 0; i_node--) { int i_first_child = i_node * degree + 1; nodes.elementAt(i_node).offsetOfFirstElement = nodes.elementAt(i_first_child).offsetOfFirstElement; int i_child = 0; Prism mbr; mbr = nodes.elementAt(i_first_child + i_child); double t1 = mbr.t1; double x1 = mbr.x1; double y1 = mbr.y1; double t2 = mbr.t2; double x2 = mbr.x2; double y2 = mbr.y2; i_child++; while (i_child < degree) { mbr = nodes.elementAt(i_first_child + i_child); if (mbr.t1 < t1) t1 = mbr.t1; if (mbr.x1 < x1) x1 = mbr.x1; if (mbr.y1 < y1) y1 = mbr.y1; if (mbr.t2 > t2) t2 = mbr.t2; if (mbr.x2 > x2) x2 = mbr.x2; if (mbr.y2 > y2) y2 = mbr.y2; i_child++; } nodes.elementAt(i_node).set(t1, x1, y1, t2, x2, y2); } // Start writing the tree // write tree header (including size) // Total tree size. (== Total bytes written - 8 bytes for the size // itself) dataOut.writeInt(TreeHeaderSize + NodeSize * nodeCount + len); // Tree height dataOut.writeInt(height); // Degree dataOut.writeInt(degree); dataOut.writeInt(elementCount); //isColumnar dataOut.writeInt(columnarStorage ? 1 : 0); // write nodes for (SplitStruct node : nodes) { node.write(dataOut); } // write elements if (columnarStorage) { byte[] index_bs = index_bos.toByteArray(); byte[][] bss = new byte[bos.length][]; for (int i = 0; i < bss.length; i++) { bss[i] = bos[i].toByteArray(); } for (int element_i = 0; element_i < elementCount; element_i++) { //int eol = skipToEOL(element_bytes, offsets[element_i]); //dataOut.write(element_bytes, offsets[element_i], eol - offsets[element_i]); dataOut.write(index_bs, ids[element_i] * IndexUnitSize, IndexUnitSize); } for (int i = 0; i < fields.length; i++) { int fieldSize = 0; if (fields[i].getType().equals(Integer.TYPE)) { fieldSize = 4; } else if (fields[i].getType().equals(Long.TYPE)) { fieldSize = 8; } else if (fields[i].getType().equals(Double.TYPE)) { fieldSize = 8; } else { //throw new RuntimeException("Unsupported field type: " + fields[i].getType().getName()); continue; } for (int element_i = 0; element_i < elementCount; element_i++) { //int eol = skipToEOL(element_bytes, offsets[element_i]); //dataOut.write(element_bytes, offsets[element_i], eol - offsets[element_i]); dataOut.write(bss[i], ids[element_i] * fieldSize, fieldSize); } } } else { for (int element_i = 0; element_i < elementCount; element_i++) { int eol = skipToEOL(element_bytes, offsets[element_i]); dataOut.write(element_bytes, offsets[element_i], eol - offsets[element_i]); } } } catch (IOException e) { e.printStackTrace(); } catch (IllegalArgumentException e) { // TODO Auto-generated catch block e.printStackTrace(); } catch (IllegalAccessException e) { // TODO Auto-generated catch block e.printStackTrace(); } }
From source file:com.ricemap.spateDB.core.RTree.java
License:Apache License
/** * k nearest neighbor query Note: Current algorithm is approximate just for * simplicity. Writing an exact algorithm is on our TODO list * /*w ww .j a v a2 s. c o m*/ * @param qx * @param qy * @param k * @param output */ public int knn(final double qt, final double qx, final double qy, int k, final ResultCollector2<T, Double> output) { double query_area = ((getMBR().x2 - getMBR().x1) * (getMBR().y2 - getMBR().y1)) * k / getElementCount(); double query_radius = Math.sqrt(query_area / Math.PI); boolean result_correct; final Vector<Double> distances = new Vector<Double>(); final Vector<T> shapes = new Vector<T>(); // Find results in the range and increase this range if needed to ensure // correctness of the answer do { // Initialize result and query range distances.clear(); shapes.clear(); Prism queryRange = new Prism(); queryRange.x1 = qx - query_radius / 2; queryRange.y1 = qy - query_radius / 2; queryRange.x2 = qx + query_radius / 2; queryRange.y2 = qy + query_radius / 2; // Retrieve all results in range searchColumnar(queryRange, new ResultCollector<Writable>() { @Override public void collect(Writable shape) { distances.add(((T) shape).distanceTo(qt, qx, qy)); shapes.add((T) ((T) shape).clone()); } }, null); if (shapes.size() < k) { // Didn't find k elements in range, double the range to get more // items if (shapes.size() == getElementCount()) { // Already returned all possible elements result_correct = true; } else { query_radius *= 2; result_correct = false; } } else { // Sort items by distance to get the kth neighbor IndexedSortable s = new IndexedSortable() { @Override public void swap(int i, int j) { double temp_distance = distances.elementAt(i); distances.set(i, distances.elementAt(j)); distances.set(j, temp_distance); T temp_shape = shapes.elementAt(i); shapes.set(i, shapes.elementAt(j)); shapes.set(j, temp_shape); } @Override public int compare(int i, int j) { // Note. Equality is not important to check because // items with the // same distance can be ordered anyway. if (distances.elementAt(i) < distances.elementAt(j)) return -1; return 1; } }; IndexedSorter sorter = new QuickSort(); sorter.sort(s, 0, shapes.size()); if (distances.elementAt(k - 1) > query_radius) { result_correct = false; query_radius = distances.elementAt(k); } else { result_correct = true; } } } while (!result_correct); int result_size = Math.min(k, shapes.size()); if (output != null) { for (int i = 0; i < result_size; i++) { output.collect(shapes.elementAt(i), distances.elementAt(i)); } } return result_size; }
From source file:edu.umn.cs.spatialHadoop.core.GlobalIndex.java
License:Open Source License
public int knn(final double qx, final double qy, int k, ResultCollector2<S, Double> output) { double query_area = (getMBR().getWidth() * getMBR().getHeight()) * k / size(); double query_radius = Math.sqrt(query_area / Math.PI); boolean result_correct; final Vector<Double> distances = new Vector<Double>(); final Vector<S> shapes = new Vector<S>(); // Find results in the range and increase this range if needed to ensure // correctness of the answer do {//from w w w . j a v a2 s . com // Initialize result and query range distances.clear(); shapes.clear(); Rectangle queryRange = new Rectangle(); queryRange.x1 = qx - query_radius / 2; queryRange.y1 = qy - query_radius / 2; queryRange.x2 = qx + query_radius / 2; queryRange.y2 = qy + query_radius / 2; // Retrieve all results in range rangeQuery(queryRange, new ResultCollector<S>() { @Override public void collect(S shape) { distances.add(shape.distanceTo(qx, qy)); shapes.add((S) shape.clone()); } }); if (shapes.size() <= k) { // Didn't find k elements in range, double the range to get more items if (shapes.size() == size() || shapes.size() == k) { // Already returned all possible elements result_correct = true; } else { query_radius *= 2; result_correct = false; } } else { // Sort items by distance to get the kth neighbor IndexedSortable s = new IndexedSortable() { @Override public void swap(int i, int j) { double temp_distance = distances.elementAt(i); distances.set(i, distances.elementAt(j)); distances.set(j, temp_distance); S temp_shape = shapes.elementAt(i); shapes.set(i, shapes.elementAt(j)); shapes.set(j, temp_shape); } @Override public int compare(int i, int j) { // Note. Equality is not important to check because items with the // same distance can be ordered anyway. if (distances.elementAt(i) == distances.elementAt(j)) return 0; if (distances.elementAt(i) < distances.elementAt(j)) return -1; return 1; } }; IndexedSorter sorter = new QuickSort(); sorter.sort(s, 0, shapes.size()); if (distances.elementAt(k - 1) > query_radius) { result_correct = false; query_radius = distances.elementAt(k); } else { result_correct = true; } } } while (!result_correct); int result_size = Math.min(k, shapes.size()); if (output != null) { for (int i = 0; i < result_size; i++) { output.collect(shapes.elementAt(i), distances.elementAt(i)); } } return result_size; }
From source file:edu.umn.cs.spatialHadoop.core.RTree.java
License:Open Source License
/** * Builds the RTree given a serialized list of elements. It uses the given * stockObject to deserialize these elements using * {@link TextSerializable#fromText(Text)} and build the tree. Also writes the * created tree to the disk directly.//from w w w . j a va 2 s .co m * * @param element_bytes * - serialization of all elements separated by new lines * @param offset * - offset of the first byte to use in elements_bytes * @param len * - number of bytes to use in elements_bytes * @param degree * - Degree of the R-tree to build in terms of number of children per * node * @param dataOut * - output stream to write the result to. * @param fast_sort * - setting this to <code>true</code> allows the method to run * faster by materializing the offset of each element in the list * which speeds up the comparison. However, this requires an * additional 16 bytes per element. So, for each 1M elements, the * method will require an additional 16 M bytes (approximately). */ public void bulkLoadWrite(final byte[] element_bytes, final int offset, final int len, final int degree, DataOutput dataOut, final boolean fast_sort) { try { // Count number of elements in the given text int i_start = offset; final Text line = new Text(); while (i_start < offset + len) { int i_end = skipToEOL(element_bytes, i_start); // Extract the line without end of line character line.set(element_bytes, i_start, i_end - i_start - 1); stockObject.fromText(line); elementCount++; i_start = i_end; } LOG.info("Bulk loading an RTree with " + elementCount + " elements"); // It turns out the findBestDegree returns the best degree when the whole // tree is loaded to memory when processed. However, as current algorithms // process the tree while it's on disk, a higher degree should be selected // such that a node fits one file block (assumed to be 4K). //final int degree = findBestDegree(bytesAvailable, elementCount); LOG.info("Writing an RTree with degree " + degree); int height = Math.max(1, (int) Math.ceil(Math.log(elementCount) / Math.log(degree))); int leafNodeCount = (int) Math.pow(degree, height - 1); if (elementCount < 2 * leafNodeCount && height > 1) { height--; leafNodeCount = (int) Math.pow(degree, height - 1); } int nodeCount = (int) ((Math.pow(degree, height) - 1) / (degree - 1)); int nonLeafNodeCount = nodeCount - leafNodeCount; // Keep track of the offset of each element in the text final int[] offsets = new int[elementCount]; final double[] xs = fast_sort ? new double[elementCount] : null; final double[] ys = fast_sort ? new double[elementCount] : null; i_start = offset; line.clear(); for (int i = 0; i < elementCount; i++) { offsets[i] = i_start; int i_end = skipToEOL(element_bytes, i_start); if (xs != null) { // Extract the line with end of line character line.set(element_bytes, i_start, i_end - i_start - 1); stockObject.fromText(line); // Sample center of the shape xs[i] = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; ys[i] = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; } i_start = i_end; } /**A struct to store information about a split*/ class SplitStruct extends Rectangle { /**Start and end index for this split*/ int index1, index2; /**Direction of this split*/ byte direction; /**Index of first element on disk*/ int offsetOfFirstElement; static final byte DIRECTION_X = 0; static final byte DIRECTION_Y = 1; SplitStruct(int index1, int index2, byte direction) { this.index1 = index1; this.index2 = index2; this.direction = direction; } @Override public void write(DataOutput out) throws IOException { out.writeInt(offsetOfFirstElement); super.write(out); } void partition(Queue<SplitStruct> toBePartitioned) { IndexedSortable sortableX; IndexedSortable sortableY; if (fast_sort) { // Use materialized xs[] and ys[] to do the comparisons sortableX = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { if (xs[i] < xs[j]) return -1; if (xs[i] > xs[j]) return 1; return 0; } }; sortableY = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { if (ys[i] < ys[j]) return -1; if (ys[i] > ys[j]) return 1; return 0; } }; } else { // No materialized xs and ys. Always deserialize objects to compare sortableX = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { // Get end of line int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double xi = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double xj = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; if (xi < xj) return -1; if (xi > xj) return 1; return 0; } }; sortableY = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double yi = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double yj = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; if (yi < yj) return -1; if (yi > yj) return 1; return 0; } }; } final IndexedSorter sorter = new QuickSort(); final IndexedSortable[] sortables = new IndexedSortable[2]; sortables[SplitStruct.DIRECTION_X] = sortableX; sortables[SplitStruct.DIRECTION_Y] = sortableY; sorter.sort(sortables[direction], index1, index2); // Partition into maxEntries partitions (equally) and // create a SplitStruct for each partition int i1 = index1; for (int iSplit = 0; iSplit < degree; iSplit++) { int i2 = index1 + (index2 - index1) * (iSplit + 1) / degree; SplitStruct newSplit = new SplitStruct(i1, i2, (byte) (1 - direction)); toBePartitioned.add(newSplit); i1 = i2; } } } // All nodes stored in level-order traversal Vector<SplitStruct> nodes = new Vector<SplitStruct>(); final Queue<SplitStruct> toBePartitioned = new LinkedList<SplitStruct>(); toBePartitioned.add(new SplitStruct(0, elementCount, SplitStruct.DIRECTION_X)); while (!toBePartitioned.isEmpty()) { SplitStruct split = toBePartitioned.poll(); if (nodes.size() < nonLeafNodeCount) { // This is a non-leaf split.partition(toBePartitioned); } nodes.add(split); } if (nodes.size() != nodeCount) { throw new RuntimeException( "Expected node count: " + nodeCount + ". Real node count: " + nodes.size()); } // Now we have our data sorted in the required order. Start building // the tree. // Store the offset of each leaf node in the tree FSDataOutputStream fakeOut = null; try { fakeOut = new FSDataOutputStream(new java.io.OutputStream() { // Null output stream @Override public void write(int b) throws IOException { // Do nothing } @Override public void write(byte[] b, int off, int len) throws IOException { // Do nothing } @Override public void write(byte[] b) throws IOException { // Do nothing } }, null, TreeHeaderSize + nodes.size() * NodeSize); for (int i_leaf = nonLeafNodeCount, i = 0; i_leaf < nodes.size(); i_leaf++) { nodes.elementAt(i_leaf).offsetOfFirstElement = (int) fakeOut.getPos(); if (i != nodes.elementAt(i_leaf).index1) throw new RuntimeException(); double x1, y1, x2, y2; // Initialize MBR to first object int eol = skipToEOL(element_bytes, offsets[i]); fakeOut.write(element_bytes, offsets[i], eol - offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); Rectangle mbr = stockObject.getMBR(); x1 = mbr.x1; y1 = mbr.y1; x2 = mbr.x2; y2 = mbr.y2; i++; while (i < nodes.elementAt(i_leaf).index2) { eol = skipToEOL(element_bytes, offsets[i]); fakeOut.write(element_bytes, offsets[i], eol - offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); mbr = stockObject.getMBR(); if (mbr.x1 < x1) x1 = mbr.x1; if (mbr.y1 < y1) y1 = mbr.y1; if (mbr.x2 > x2) x2 = mbr.x2; if (mbr.y2 > y2) y2 = mbr.y2; i++; } nodes.elementAt(i_leaf).set(x1, y1, x2, y2); } } finally { if (fakeOut != null) fakeOut.close(); } // Calculate MBR and offsetOfFirstElement for non-leaves for (int i_node = nonLeafNodeCount - 1; i_node >= 0; i_node--) { int i_first_child = i_node * degree + 1; nodes.elementAt(i_node).offsetOfFirstElement = nodes.elementAt(i_first_child).offsetOfFirstElement; int i_child = 0; Rectangle mbr; mbr = nodes.elementAt(i_first_child + i_child); double x1 = mbr.x1; double y1 = mbr.y1; double x2 = mbr.x2; double y2 = mbr.y2; i_child++; while (i_child < degree) { mbr = nodes.elementAt(i_first_child + i_child); if (mbr.x1 < x1) x1 = mbr.x1; if (mbr.y1 < y1) y1 = mbr.y1; if (mbr.x2 > x2) x2 = mbr.x2; if (mbr.y2 > y2) y2 = mbr.y2; i_child++; } nodes.elementAt(i_node).set(x1, y1, x2, y2); } // Start writing the tree // write tree header (including size) // Total tree size. (== Total bytes written - 8 bytes for the size itself) dataOut.writeInt(TreeHeaderSize + NodeSize * nodeCount + len); // Tree height dataOut.writeInt(height); // Degree dataOut.writeInt(degree); dataOut.writeInt(elementCount); // write nodes for (SplitStruct node : nodes) { node.write(dataOut); } // write elements for (int element_i = 0; element_i < elementCount; element_i++) { int eol = skipToEOL(element_bytes, offsets[element_i]); dataOut.write(element_bytes, offsets[element_i], eol - offsets[element_i]); } } catch (IOException e) { e.printStackTrace(); } }
From source file:edu.umn.cs.spatialHadoop.core.RTree.java
License:Open Source License
/** * k nearest neighbor query/* w ww.j a va 2 s .c om*/ * @param qx * @param qy * @param k * @param output */ public int knn(final double qx, final double qy, int k, final ResultCollector2<T, Double> output) { double query_area = ((getMBR().x2 - getMBR().x1) * (getMBR().y2 - getMBR().y1)) * k / getElementCount(); double query_radius = Math.sqrt(query_area / Math.PI); boolean result_correct; final Vector<Double> distances = new Vector<Double>(); final Vector<T> shapes = new Vector<T>(); // Find results in the range and increase this range if needed to ensure // correctness of the answer do { // Initialize result and query range distances.clear(); shapes.clear(); Rectangle queryRange = new Rectangle(); queryRange.x1 = qx - query_radius; queryRange.y1 = qy - query_radius; queryRange.x2 = qx + query_radius; queryRange.y2 = qy + query_radius; // Retrieve all results in range search(queryRange, new ResultCollector<T>() { @Override public void collect(T shape) { distances.add(shape.distanceTo(qx, qy)); shapes.add((T) shape.clone()); } }); if (shapes.size() < k) { // Didn't find k elements in range, double the range to get more items if (shapes.size() == getElementCount()) { // Already returned all possible elements result_correct = true; } else { query_radius *= 2; result_correct = false; } } else { // Sort items by distance to get the kth neighbor IndexedSortable s = new IndexedSortable() { @Override public void swap(int i, int j) { double temp_distance = distances.elementAt(i); distances.set(i, distances.elementAt(j)); distances.set(j, temp_distance); T temp_shape = shapes.elementAt(i); shapes.set(i, shapes.elementAt(j)); shapes.set(j, temp_shape); } @Override public int compare(int i, int j) { // Note. Equality is not important to check because items with the // same distance can be ordered anyway. if (distances.elementAt(i) == distances.elementAt(j)) return 0; if (distances.elementAt(i) < distances.elementAt(j)) return -1; return 1; } }; IndexedSorter sorter = new QuickSort(); sorter.sort(s, 0, shapes.size()); if (distances.elementAt(k - 1) > query_radius) { result_correct = false; query_radius = distances.elementAt(k); } else { result_correct = true; } } } while (!result_correct); int result_size = Math.min(k, shapes.size()); if (output != null) { for (int i = 0; i < result_size; i++) { output.collect(shapes.elementAt(i), distances.elementAt(i)); } } return result_size; }
From source file:edu.umn.cs.spatialHadoop.indexing.RTree.java
License:Open Source License
/** * Builds the RTree given a serialized list of elements. It uses the given * stockObject to deserialize these elements using * {@link TextSerializable#fromText(Text)} and build the tree. Also writes the * created tree to the disk directly./* w w w . j a v a 2 s . c om*/ * * @param element_bytes * - serialization of all elements separated by new lines * @param offset * - offset of the first byte to use in elements_bytes * @param len * - number of bytes to use in elements_bytes * @param degree * - Degree of the R-tree to build in terms of number of children per * node * @param dataOut * - output stream to write the result to. * @param fast_sort * - setting this to <code>true</code> allows the method to run * faster by materializing the offset of each element in the list * which speeds up the comparison. However, this requires an * additional 16 bytes per element. So, for each 1M elements, the * method will require an additional 16 M bytes (approximately). */ public static void bulkLoadWrite(final byte[] element_bytes, final int offset, final int len, final int degree, DataOutput dataOut, final Shape stockObject, final boolean fast_sort) { try { int elementCount = 0; // Count number of elements in the given text int i_start = offset; final Text line = new Text(); while (i_start < offset + len) { int i_end = skipToEOL(element_bytes, i_start); // Extract the line without end of line character line.set(element_bytes, i_start, i_end - i_start - 1); stockObject.fromText(line); elementCount++; i_start = i_end; } LOG.info("Bulk loading an RTree with " + elementCount + " elements"); // It turns out the findBestDegree returns the best degree when the whole // tree is loaded to memory when processed. However, as current algorithms // process the tree while it's on disk, a higher degree should be selected // such that a node fits one file block (assumed to be 4K). //final int degree = findBestDegree(bytesAvailable, elementCount); int height = Math.max(1, (int) Math.ceil(Math.log(elementCount) / Math.log(degree))); int leafNodeCount = (int) Math.pow(degree, height - 1); if (elementCount < 2 * leafNodeCount && height > 1) { height--; leafNodeCount = (int) Math.pow(degree, height - 1); } int nodeCount = (int) ((Math.pow(degree, height) - 1) / (degree - 1)); int nonLeafNodeCount = nodeCount - leafNodeCount; // Keep track of the offset of each element in the text final int[] offsets = new int[elementCount]; final double[] xs = fast_sort ? new double[elementCount] : null; final double[] ys = fast_sort ? new double[elementCount] : null; i_start = offset; line.clear(); for (int i = 0; i < elementCount; i++) { offsets[i] = i_start; int i_end = skipToEOL(element_bytes, i_start); if (xs != null) { // Extract the line with end of line character line.set(element_bytes, i_start, i_end - i_start - 1); stockObject.fromText(line); // Sample center of the shape xs[i] = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; ys[i] = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; } i_start = i_end; } /**A struct to store information about a split*/ class SplitStruct extends Rectangle { /**Start and end index for this split*/ int index1, index2; /**Direction of this split*/ byte direction; /**Index of first element on disk*/ int offsetOfFirstElement; static final byte DIRECTION_X = 0; static final byte DIRECTION_Y = 1; SplitStruct(int index1, int index2, byte direction) { this.index1 = index1; this.index2 = index2; this.direction = direction; } @Override public void write(DataOutput out) throws IOException { out.writeInt(offsetOfFirstElement); super.write(out); } void partition(Queue<SplitStruct> toBePartitioned) { IndexedSortable sortableX; IndexedSortable sortableY; if (fast_sort) { // Use materialized xs[] and ys[] to do the comparisons sortableX = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { if (xs[i] < xs[j]) return -1; if (xs[i] > xs[j]) return 1; return 0; } }; sortableY = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap xs double tempx = xs[i]; xs[i] = xs[j]; xs[j] = tempx; // Swap ys double tempY = ys[i]; ys[i] = ys[j]; ys[j] = tempY; // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { if (ys[i] < ys[j]) return -1; if (ys[i] > ys[j]) return 1; return 0; } }; } else { // No materialized xs and ys. Always deserialize objects to compare sortableX = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { // Get end of line int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double xi = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double xj = (stockObject.getMBR().x1 + stockObject.getMBR().x2) / 2; if (xi < xj) return -1; if (xi > xj) return 1; return 0; } }; sortableY = new IndexedSortable() { @Override public void swap(int i, int j) { // Swap id int tempid = offsets[i]; offsets[i] = offsets[j]; offsets[j] = tempid; } @Override public int compare(int i, int j) { int eol = skipToEOL(element_bytes, offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); double yi = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; eol = skipToEOL(element_bytes, offsets[j]); line.set(element_bytes, offsets[j], eol - offsets[j] - 1); stockObject.fromText(line); double yj = (stockObject.getMBR().y1 + stockObject.getMBR().y2) / 2; if (yi < yj) return -1; if (yi > yj) return 1; return 0; } }; } final IndexedSorter sorter = new QuickSort(); final IndexedSortable[] sortables = new IndexedSortable[2]; sortables[SplitStruct.DIRECTION_X] = sortableX; sortables[SplitStruct.DIRECTION_Y] = sortableY; sorter.sort(sortables[direction], index1, index2); // Partition into maxEntries partitions (equally) and // create a SplitStruct for each partition int i1 = index1; for (int iSplit = 0; iSplit < degree; iSplit++) { int i2 = index1 + (index2 - index1) * (iSplit + 1) / degree; SplitStruct newSplit = new SplitStruct(i1, i2, (byte) (1 - direction)); toBePartitioned.add(newSplit); i1 = i2; } } } // All nodes stored in level-order traversal Vector<SplitStruct> nodes = new Vector<SplitStruct>(); final Queue<SplitStruct> toBePartitioned = new LinkedList<SplitStruct>(); toBePartitioned.add(new SplitStruct(0, elementCount, SplitStruct.DIRECTION_X)); while (!toBePartitioned.isEmpty()) { SplitStruct split = toBePartitioned.poll(); if (nodes.size() < nonLeafNodeCount) { // This is a non-leaf split.partition(toBePartitioned); } nodes.add(split); } if (nodes.size() != nodeCount) { throw new RuntimeException( "Expected node count: " + nodeCount + ". Real node count: " + nodes.size()); } // Now we have our data sorted in the required order. Start building // the tree. // Store the offset of each leaf node in the tree FSDataOutputStream fakeOut = null; try { fakeOut = new FSDataOutputStream(new java.io.OutputStream() { // Null output stream @Override public void write(int b) throws IOException { // Do nothing } @Override public void write(byte[] b, int off, int len) throws IOException { // Do nothing } @Override public void write(byte[] b) throws IOException { // Do nothing } }, null, TreeHeaderSize + nodes.size() * NodeSize); for (int i_leaf = nonLeafNodeCount, i = 0; i_leaf < nodes.size(); i_leaf++) { nodes.elementAt(i_leaf).offsetOfFirstElement = (int) fakeOut.getPos(); if (i != nodes.elementAt(i_leaf).index1) throw new RuntimeException(); double x1, y1, x2, y2; // Initialize MBR to first object int eol = skipToEOL(element_bytes, offsets[i]); fakeOut.write(element_bytes, offsets[i], eol - offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); Rectangle mbr = stockObject.getMBR(); x1 = mbr.x1; y1 = mbr.y1; x2 = mbr.x2; y2 = mbr.y2; i++; while (i < nodes.elementAt(i_leaf).index2) { eol = skipToEOL(element_bytes, offsets[i]); fakeOut.write(element_bytes, offsets[i], eol - offsets[i]); line.set(element_bytes, offsets[i], eol - offsets[i] - 1); stockObject.fromText(line); mbr = stockObject.getMBR(); if (mbr.x1 < x1) x1 = mbr.x1; if (mbr.y1 < y1) y1 = mbr.y1; if (mbr.x2 > x2) x2 = mbr.x2; if (mbr.y2 > y2) y2 = mbr.y2; i++; } nodes.elementAt(i_leaf).set(x1, y1, x2, y2); } } finally { if (fakeOut != null) fakeOut.close(); } // Calculate MBR and offsetOfFirstElement for non-leaves for (int i_node = nonLeafNodeCount - 1; i_node >= 0; i_node--) { int i_first_child = i_node * degree + 1; nodes.elementAt(i_node).offsetOfFirstElement = nodes.elementAt(i_first_child).offsetOfFirstElement; int i_child = 0; Rectangle mbr; mbr = nodes.elementAt(i_first_child + i_child); double x1 = mbr.x1; double y1 = mbr.y1; double x2 = mbr.x2; double y2 = mbr.y2; i_child++; while (i_child < degree) { mbr = nodes.elementAt(i_first_child + i_child); if (mbr.x1 < x1) x1 = mbr.x1; if (mbr.y1 < y1) y1 = mbr.y1; if (mbr.x2 > x2) x2 = mbr.x2; if (mbr.y2 > y2) y2 = mbr.y2; i_child++; } nodes.elementAt(i_node).set(x1, y1, x2, y2); } // Start writing the tree // write tree header (including size) // Total tree size. (== Total bytes written - 8 bytes for the size itself) dataOut.writeInt(TreeHeaderSize + NodeSize * nodeCount + len); // Tree height dataOut.writeInt(height); // Degree dataOut.writeInt(degree); dataOut.writeInt(elementCount); // write nodes for (SplitStruct node : nodes) { node.write(dataOut); } // write elements for (int element_i = 0; element_i < elementCount; element_i++) { int eol = skipToEOL(element_bytes, offsets[element_i]); dataOut.write(element_bytes, offsets[element_i], eol - offsets[element_i]); } } catch (IOException e) { e.printStackTrace(); } }