the-qa-companyGH-528 improve binary search in super blocks by keeping…

… track of the estimated location of values in the underlying long array. This assumes that the values are ordered.

qendpoint-core/src/main/java/com/the_qa_company/qendpoint/core/compact/bitmap/Bitmap375Big.java

@@ -23,6 +23,8 @@
 import com.the_qa_company.qendpoint.core.util.io.CloseSuppressPath;
 import com.the_qa_company.qendpoint.core.util.io.Closer;
 import com.the_qa_company.qendpoint.core.util.io.IOUtil;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 import java.io.Closeable;
 import java.io.IOException;
@@ -39,14 +41,31 @@
  * @author mario.arias
  */
 public class Bitmap375Big extends Bitmap64Big {
+
+	private static final Logger logger = LoggerFactory.getLogger(Bitmap375Big.class);
+
+	private static final boolean oldBinarySearch;
+
+	static {
+		// check if the system property "useOldBinarySeearch" is set to true
+		String useOldBinarySearch = System.getProperty("useOldBinarySearch");
+		if (useOldBinarySearch != null && useOldBinarySearch.equalsIgnoreCase("true")) {
+			oldBinarySearch = true;
+			logger.debug("Using old binary search");
+		} else {
+			logger.debug("Using new binary search");
+			oldBinarySearch = false;
+		}
+
+	}
+
 	/**
 	 * create disk version bitmap with in memory super index
 	 *
 	 * @param location location
 	 * @param nbits    number of bits
 	 * @return bitmap
 	 */
-
 	public static Bitmap375Big disk(Path location, long nbits) {
 		return disk(location, nbits, false);
 	}
@@ -181,6 +200,7 @@ public void updateIndex() {
 		}
 		pop = countSuperBlock + countBlock;
 		indexUpToDate = true;
+		superBlocks.recalculateEstimatedValueLocation();
 	}
 
 	/*
@@ -189,8 +209,9 @@ public void updateIndex() {
 	 */
 	@Override
 	public boolean access(long bitIndex) {
-		if (bitIndex < 0)
+		if (bitIndex < 0) {
 			throw new IndexOutOfBoundsException("bitIndex < 0: " + bitIndex);
+		}
 
 		long wordIndex = wordIndex(bitIndex);
 		if (wordIndex >= words.length()) {
@@ -324,15 +345,14 @@ public long select1(long x) {
 			return 0;
 		}
 		// Search superblock (binary Search)
-		long superBlockIndex = binarySearch(superBlocks, x);
+		long superBlockIndex = oldBinarySearch ? binarySearch(superBlocks, x) : binarySearchNew(superBlocks, x);
 
 		// If there is a run of many zeros, two correlative superblocks may have
 		// the same value,
 		// We need to position at the first of them.
 
 		while (superBlockIndex > 0 && (superBlocks.get(superBlockIndex) >= x)) {
 			superBlockIndex--;
-
 		}
 
 		long countdown = x - superBlocks.get(superBlockIndex);
@@ -444,6 +464,7 @@ public static long binarySearch0(LongArray arr, long fromIndex, long toIndex, lo
 	 * @param val val
 	 * @return index
 	 */
+
 	public static long binarySearch(LongArray arr, long val) {
 		long min = 0, max = arr.length(), mid;
 
@@ -460,11 +481,52 @@ public static long binarySearch(LongArray arr, long val) {
 		return min;
 	}
 
+	public static long binarySearchNew(LongArray arr, long val) {
+
+		long min = arr.getEstimatedLocationLowerBound(val);
+		long max = arr.getEstimatedLocationUpperBound(val);
+		long mid = arr.getEstimatedLocation(val, min, max);
+
+		int i = 0;
+		while (min + 1 < max) {
+			// After the first iteration, the value that we are looking for is
+			// typically very close to the min value. Using linear search for
+			// the next two iterations improves the chances that we find the
+			// value faster than with binary search.
+			if (i == 1 || i == 2) {
+				long v = arr.get(min + 1);
+				if (v >= val) {
+					max = min + 1;
+				} else {
+					min = min + 1;
+				}
+			} else {
+				long v = arr.get(mid);
+				if (v >= val) {
+					max = mid;
+				} else {
+					min = mid;
+				}
+			}
+			mid = (min + max) / 2;
+			i++;
+		}
+
+		arr.updateEstimatedValueLocation(val, min);
+
+		return min;
+	}
+
 	public CloseSuppressPath getBlocksPath() {
 		return blocksPath;
 	}
 
 	public CloseSuppressPath getSuperBlocksPath() {
 		return superBlocksPath;
 	}
+
+	@Override
+	public String toString() {
+		return "Bitmap375Big{}";
+	}
 }

qendpoint-core/src/main/java/com/the_qa_company/qendpoint/core/util/disk/AbstractLongArray.java

@@ -0,0 +1,94 @@
+package com.the_qa_company.qendpoint.core.util.disk;
+
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+public abstract class AbstractLongArray implements LongArray {
+
+	private final Logger logger = LoggerFactory.getLogger(getClass());
+
+	private final int ESTIMATED_LOCATION_ARRAY_SIZE = 1024 * 128;
+
+	// should take about 1MB per array when PREV_FOUND_SIZE is 1024 * 128
+	private final long[] estimatedLocationMax = new long[ESTIMATED_LOCATION_ARRAY_SIZE];
+	private final long[] estimatedLocationMin = new long[ESTIMATED_LOCATION_ARRAY_SIZE];
+	private final long[] estimatedLocation = new long[ESTIMATED_LOCATION_ARRAY_SIZE];
+
+	private int estimatedLocationBucketSize;
+
+	long maxValue = 1;
+
+	@Override
+	public int getEstimatedLocationArrayBucketSize() {
+		return estimatedLocationBucketSize;
+	}
+
+	private void updateEstimatedLocationArrayBucketSize() {
+		int minBucketSize = (int) (maxValue / ESTIMATED_LOCATION_ARRAY_SIZE);
+		// we want to have the next power of 2
+		int next = 1;
+		while (next < minBucketSize) {
+			next <<= 1;
+		}
+		this.estimatedLocationBucketSize = next;
+	}
+
+	@Override
+	public long[] getEstimatedLocationArray() {
+		return estimatedLocation;
+	}
+
+	@Override
+	public long[] getEstimatedLocationArrayMin() {
+		return estimatedLocationMin;
+	}
+
+	@Override
+	public long[] getEstimatedLocationArrayMax() {
+		return estimatedLocationMax;
+	}
+
+	@Override
+	public void recalculateEstimatedValueLocation() {
+		updateEstimatedLocationArrayBucketSize();
+		int estimatedLocationBucketSize = getEstimatedLocationArrayBucketSize();
+		long len = length();
+		boolean shouldLog = len > 1024 * 1024 * 2;
+		if (shouldLog) {
+			logger.info("Recalculating estimated location array 0%");
+		}
+
+		for (int i = 0; i < len; i++) {
+			long val = get(i);
+			if (val == 0) {
+				continue;
+			}
+
+			int index = (int) (val / estimatedLocationBucketSize + 1);
+			estimatedLocationMax[index] = Math.max(estimatedLocationMax[index], i);
+			if (estimatedLocationMin[index] == 0) {
+				estimatedLocationMin[index] = i;
+			} else {
+				estimatedLocationMin[index] = Math.min(estimatedLocationMin[index], i);
+			}
+			estimatedLocation[index] = (estimatedLocationMax[index] + estimatedLocationMin[index]) / 2;
+
+			if (shouldLog && i % (1024 * 1024) == 0) {
+				logger.info("Recalculating estimated location array {}%", (int) Math.floor(100.0 / len * i));
+			}
+		}
+
+		if (shouldLog) {
+			logger.info("Recalculating estimated location array 100%");
+		}
+	}
+
+	@Override
+	public final void set(long index, long value) {
+		maxValue = Math.max(maxValue, value);
+		innerSet(index, value);
+	}
+
+	abstract protected void innerSet(long index, long value);
+
+}

qendpoint-core/src/main/java/com/the_qa_company/qendpoint/core/util/disk/LargeLongArray.java

@@ -10,7 +10,7 @@
  *
  * @author Antoine Willerval
  */
-public class LargeLongArray implements LongArray {
+public class LargeLongArray extends AbstractLongArray {
 	private UnsafeLongArray array;
 
 	/**
@@ -26,7 +26,7 @@ public long get(long index) {
 	}
 
 	@Override
-	public void set(long index, long value) {
+	protected void innerSet(long index, long value) {
 		array.set(index, value);
 	}
 
@@ -55,4 +55,5 @@ public void resize(long newSize) throws IOException {
 	public void clear() {
 		array.clear();
 	}
+
 }

qendpoint-core/src/main/java/com/the_qa_company/qendpoint/core/util/disk/LongArray.java

@@ -1,6 +1,8 @@
 package com.the_qa_company.qendpoint.core.util.disk;
 
 import com.the_qa_company.qendpoint.core.util.io.IOUtil;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 import java.io.IOException;
 import java.util.Iterator;
@@ -10,6 +12,10 @@
  * Describe a large array of longs
  */
 public interface LongArray extends Iterable<Long> {
+
+	Logger logger = LoggerFactory.getLogger(LongArray.class);
+	long[] EMPTY_ARRAY = new long[0];
+
 	/**
 	 * create an in memory long array
 	 *
@@ -208,4 +214,84 @@ public Long next() {
 			}
 		};
 	}
+
+	/**
+	 * @return the estimated location array that contains the highest location
+	 *         for a given value
+	 */
+	default long[] getEstimatedLocationArrayMax() {
+		return getEstimatedLocationArray();
+	}
+
+	/**
+	 * @return the estimated location array that contains the lowest location
+	 *         for a given value
+	 */
+	default long[] getEstimatedLocationArrayMin() {
+		return getEstimatedLocationArray();
+	}
+
+	/**
+	 * @return the estimated location array
+	 */
+	default long[] getEstimatedLocationArray() {
+		return EMPTY_ARRAY;
+	}
+
+	default int getEstimatedLocationArrayBucketSize() {
+		return 65536;
+	}
+
+	default long getEstimatedLocationLowerBound(long val) {
+		int index = (int) (val / getEstimatedLocationArrayBucketSize() + 1);
+		if (index - 1 >= 0) {
+			long t = getEstimatedLocationArrayMax()[index - 1];
+			if (t > 0) {
+				return t;
+			}
+		}
+		return 0;
+	}
+
+	default long getEstimatedLocationUpperBound(long val) {
+		int index = (int) (val / getEstimatedLocationArrayBucketSize() + 1);
+		long[] estimatedLocationMin = getEstimatedLocationArrayMin();
+		if (index + 1 < estimatedLocationMin.length) {
+			long t = estimatedLocationMin[index + 1];
+			if (t > 0) {
+				return Math.min(length(), t);
+			}
+		}
+
+		return length();
+	}
+
+	default long getEstimatedLocation(long val, long min, long max) {
+		int index = (int) (val / getEstimatedLocationArrayBucketSize() + 1);
+		var estimatedLocation = getEstimatedLocationArray();
+
+		if (index >= estimatedLocation.length) {
+			return (min + max) / 2;
+		}
+		long t = estimatedLocation[index];
+		if (t > min && t < max) {
+			return t;
+		} else {
+			return (min + max) / 2;
+		}
+	}
+
+	default void recalculateEstimatedValueLocation() {
+		logger.info("Class {} does not support recalculateEstimatedValueLocation()",
+				this.getClass().getCanonicalName());
+	}
+
+	default void updateEstimatedValueLocation(long val, long min) {
+		int index = (int) (val / getEstimatedLocationArrayBucketSize() + 1);
+		long[] estimatedLocation = getEstimatedLocationArray();
+		if (index >= estimatedLocation.length) {
+			return;
+		}
+		estimatedLocation[index] = min;
+	}
 }

qendpoint-core/src/main/java/com/the_qa_company/qendpoint/core/util/disk/LongArrayDisk.java

@@ -33,7 +33,7 @@
 
 //Implementing an array of longs that is backed up on disk. Following this: http://vanillajava.blogspot.fr/2011/12/using-memory-mapped-file-for-huge.html
 
-public class LongArrayDisk implements Closeable, LongArray {
+public class LongArrayDisk extends AbstractLongArray implements Closeable {
 	private static final long MAPPING_SIZE = 1 << 30;
 	private final boolean closeChannel;
 	private final FileChannel channel;
@@ -137,7 +137,7 @@ public long get(long x) {
 	}
 
 	@Override
-	public void set(long index, long value) {
+	protected void innerSet(long index, long value) {
 		if (index >= size || index < 0) {
 			throw new IndexOutOfBoundsException();
 		}