Merge pull request #174 from ate47/diff_fix

Update JLargeArrays and use SequenceLog64Big instead of SequenceLog64
rdfhdt · Sep 28, 2022 · 43c374c · 43c374c
2 parents 7222f8a + 9d69a25
commit 43c374c
Show file tree

Hide file tree

Showing 12 changed files with 136 additions and 58 deletions.
diff --git a/hdt-java-core/pom.xml b/hdt-java-core/pom.xml
@@ -49,8 +49,9 @@
             <artifactId>jena-arq</artifactId>
         </dependency>
         <dependency>
-            <groupId>pl.edu.icm</groupId>
+            <groupId>org.visnow</groupId>
             <artifactId>JLargeArrays</artifactId>
+            <version>1.7-20220624.150242-7</version>
         </dependency>
         <dependency>
             <groupId>pl.pragmatists</groupId>
@@ -63,7 +64,12 @@
 		    <version>1.6</version>
 		</dependency>		        
     </dependencies>
-
+    <repositories>
+        <repository>
+            <id>gitlab-maven</id>
+            <url>https://gitlab.com/api/v4/projects/375779/packages/maven</url>
+        </repository>
+    </repositories>
     <build>
         <plugins>
             <plugin>

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/bitmap/Bitmap375Disk.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/bitmap/Bitmap375Disk.java
@@ -111,7 +111,7 @@ public boolean access(long bitIndex) {
         if (bitIndex < 0)
             throw new IndexOutOfBoundsException("bitIndex < 0: " + bitIndex);
 
-        int wordIndex = wordIndex(bitIndex);
+        long wordIndex = wordIndex(bitIndex);
         if(wordIndex>=words.length()) {
             return false;
         }
@@ -415,4 +415,4 @@ public void load(InputStream input, ProgressListener listener) throws IOExceptio
         super.load(input, listener);
         updateIndex();
     }
-}
+}
diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/bitmap/Bitmap64Disk.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/bitmap/Bitmap64Disk.java
@@ -21,6 +21,7 @@
 
 import org.rdfhdt.hdt.compact.integer.VByte;
 import org.rdfhdt.hdt.exceptions.NotImplementedException;
+import org.rdfhdt.hdt.hdt.HDTVocabulary;
 import org.rdfhdt.hdt.listener.ProgressListener;
 import org.rdfhdt.hdt.util.BitUtil;
 import org.rdfhdt.hdt.util.crc.CRC32;
@@ -37,7 +38,7 @@
 /**
  * Version of Bitmap64 which is backed up on disk
  */
-public class Bitmap64Disk implements Closeable {
+public class Bitmap64Disk implements Closeable, ModifiableBitmap {
 
     // Constants
     protected final static int LOGW = 6;
@@ -59,7 +60,7 @@ public Bitmap64Disk(String location, long nbits) {
     /**
      * Given a bit index, return word index containing it.
      */
-    protected static int wordIndex(long bitIndex) {
+    protected static long wordIndex(long bitIndex) {
         return (int) (bitIndex >>> LOGW);
     }
 
@@ -78,7 +79,7 @@ protected static int lastWordNumBits(long numbits) {
         return (int) ((numbits-1) % W)+1;	// +1 To have output in the range 1-64, -1 to compensate.
     }
 
-    protected final void ensureSize(int wordsRequired) {
+    protected final void ensureSize(long wordsRequired) {
         words.resize(Math.max(words.getSize()*2, wordsRequired));
     }
 
@@ -97,14 +98,73 @@ public boolean access(long bitIndex) {
         if (bitIndex < 0)
             throw new IndexOutOfBoundsException("bitIndex < 0: " + bitIndex);
 
-        int wordIndex = wordIndex(bitIndex);
+        long wordIndex = wordIndex(bitIndex);
         if(wordIndex>=words.length()) {
             return false;
         }
 
         return (words.get(wordIndex) & (1L << bitIndex)) != 0;
     }
 
+    @Override
+    public long rank1(long pos) {
+        throw new NotImplementedException();
+    }
+
+    @Override
+    public long rank0(long pos) {
+        throw new NotImplementedException();
+    }
+
+    @Override
+    public long selectNext1(long fromIndex) {
+        if (fromIndex < 0)
+            throw new IndexOutOfBoundsException("fromIndex < 0: " + fromIndex);
+
+        long wordIndex = wordIndex(fromIndex);
+        if (wordIndex >= words.length())
+            return -1;
+
+        long word = words.get(wordIndex) & (~0L << fromIndex);
+
+        while (true) {
+            if (word != 0)
+                return ((long)wordIndex * W) + Long.numberOfTrailingZeros(word);
+            if (++wordIndex == words.length())
+                return -1;
+            word = words.get(wordIndex);
+        }
+    }
+
+    @Override
+    public long select0(long n) {
+        throw new NotImplementedException();
+    }
+
+    @Override
+    public long select1(long n) {
+        throw new NotImplementedException();
+    }
+
+    @Override
+    public long countOnes() {
+        if (words.length() == 0)
+            return 0;
+        long acc = 0;
+        long end = wordIndex(numbits);
+        if (end >= words.length()) {
+            end = words.length() - 1;
+        }
+        for (int i = 0; i <= end; i++)
+            acc += Long.bitCount(words.get(i));
+        return acc;
+    }
+
+    @Override
+    public long countZeros() {
+        return words.length() * 64L - countOnes();
+    }
+
     /* (non-Javadoc)
      * @see hdt.compact.bitmap.ModifiableBitmap#append(boolean)
      */
@@ -116,7 +176,7 @@ public void set(long bitIndex, boolean value) {
         if (bitIndex < 0)
             throw new IndexOutOfBoundsException("bitIndex < 0: " + bitIndex);
 
-        int wordIndex = wordIndex(bitIndex);
+        long wordIndex = wordIndex(bitIndex);
         ensureSize(wordIndex+1);
 
         if(value) {
@@ -128,27 +188,13 @@ public void set(long bitIndex, boolean value) {
         this.numbits = Math.max(this.numbits, bitIndex+1);
     }
 
-    public long selectPrev1(long start) {
-        throw new NotImplementedException();
+    @Override
+    public String getType() {
+        return HDTVocabulary.BITMAP_TYPE_PLAIN;
     }
 
-    public long selectNext1(long fromIndex) {
-        if (fromIndex < 0)
-            throw new IndexOutOfBoundsException("fromIndex < 0: " + fromIndex);
-
-        int wordIndex = wordIndex(fromIndex);
-        if (wordIndex >= words.length())
-            return -1;
-
-        long word = words.get(wordIndex) & (~0L << fromIndex);
-
-        while (true) {
-            if (word != 0)
-                return ((long)wordIndex * W) + Long.numberOfTrailingZeros(word);
-            if (++wordIndex == words.length())
-                return -1;
-            word = words.get(wordIndex);
-        }
+    public long selectPrev1(long start) {
+        throw new NotImplementedException();
     }
 
     public long getWord(int word) {
@@ -211,4 +257,4 @@ public long getRealSizeBytes() {
     public void close() throws IOException {
         words.close();
     }
-}
+}
diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceInt32.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceInt32.java
@@ -105,9 +105,13 @@ public void set(long position, long value) {
 	@Override
     public void append(long value) {
 		assert value>=0 && value<=Integer.MAX_VALUE;
-
-		if(data.length<numelements+1) {
-			resizeArray(data.length*2);
+
+		long neededSize = numelements+1L;
+		if (neededSize > Integer.MAX_VALUE - 5) {
+			throw new IllegalArgumentException("Needed size exceeds the maximum size of this data structure " + neededSize);
+		}
+		if(data.length < neededSize) {
+			resizeArray((int) Math.min(Integer.MAX_VALUE - 5L, data.length*2L));
 		}
 		data[numelements++] = (int) value;
 	}

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceInt64.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceInt64.java
@@ -94,19 +94,23 @@ public long get(long position) {
 	@Override
     public void set(long position, long value) {
 		assert position>=0 && position<=Integer.MAX_VALUE;
-		assert value>=0 && value<=Long.MAX_VALUE;
+		assert value>=0;
 
 		data[(int)position] = value;
 		numelements = (int) Math.max(numelements, position+1);
 	}
 
 	@Override
     public void append(long value) {
-		assert value>=0 && value<=Long.MAX_VALUE;
+		assert value>=0;
 		assert numelements<Long.MAX_VALUE;
-
-		if(data.length<numelements+1) {
-			resizeArray(data.length*2);
+
+		long neededSize = numelements+1L;
+		if (neededSize > Integer.MAX_VALUE - 5) {
+			throw new IllegalArgumentException("Needed size exceeds the maximum size of this data structure " + neededSize);
+		}
+		if(data.length < neededSize) {
+			resizeArray((int) Math.min(Integer.MAX_VALUE - 5L, data.length*2L));
 		}
 		data[(int)numelements++] = value;
 	}

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceLog64.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceLog64.java
@@ -241,8 +241,11 @@ public void append(long value) {
 		}
 
 		long neededSize = numWordsFor(numbits, numentries+1);
-		if(data.length<neededSize) {
-			resizeArray(data.length*2);
+		if (neededSize > Integer.MAX_VALUE - 5) {
+			throw new IllegalArgumentException("Needed size exceeds the maximum size of this data structure " + neededSize);
+		}
+		if(data.length < neededSize) {
+			resizeArray((int) Math.min(Integer.MAX_VALUE - 5L, data.length*2L));
 		}
 
 		this.set((int)numentries, value);

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceLog64Big.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/compact/sequence/SequenceLog64Big.java
@@ -44,9 +44,8 @@
 import org.rdfhdt.hdt.util.crc.CRCInputStream;
 import org.rdfhdt.hdt.util.crc.CRCOutputStream;
 import org.rdfhdt.hdt.util.io.IOUtil;
-
-import pl.edu.icm.jlargearrays.LongLargeArray;
-import pl.edu.icm.jlargearrays.LargeArrayUtils;
+import org.visnow.jlargearrays.LargeArrayUtils;
+import org.visnow.jlargearrays.LongLargeArray;
 
 /**
  * @author mario.arias,Lyudmila Balakireva
@@ -55,8 +54,8 @@
 public class SequenceLog64Big implements DynamicSequence { 
 	private static final byte W = 64;
 	private static final int INDEX = 1073741824;
-	
-    LongLargeArray data;
+
+	LongLargeArray data;
 	private int numbits;
 	private long numentries=0;
 	private long maxvalue;
@@ -77,7 +76,7 @@ public SequenceLog64Big(int numbits, long capacity) {
 		long size = numWordsFor(numbits, capacity);
 		LongLargeArray.setMaxSizeOf32bitArray(SequenceLog64Big.INDEX);
 
-		data = new LongLargeArray(Math.max((int)size,1)); 
+		data = new LongLargeArray(Math.max(size,1));
 	}
 
 	public SequenceLog64Big(int numbits, long capacity, boolean initialize) {
@@ -158,11 +157,7 @@ private void resizeArray(long size) {
 		//data = Arrays.copyOf(data, size);
 		if(size > 0) {
 			LongLargeArray a = new LongLargeArray(size);
-			if (size < data.length()) {
-				LargeArrayUtils.arraycopy(data, 0, a, 0, size);
-			} else {
-				LargeArrayUtils.arraycopy(data, 0, a, 0, data.length());
-			}
+			LargeArrayUtils.arraycopy(data, 0, a, 0, Math.min(size, data.length()));
 			data = a;
 		}else{
 			this.numentries = 0;
@@ -182,7 +177,7 @@ public void add(Iterator<Long> elements) {
 		// Count and calculate number of bits needed per element.
 		while(elements.hasNext()) {
 			long val = elements.next();
-			max = val>max ? val : max;
+			max = Math.max(val, max);
 			numentries++;
 		}
 
@@ -208,7 +203,7 @@ public void addIntegers(ArrayList<Integer> elements) {
 		// Count and calculate number of bits needed per element.
 		for (int i=0;i<elements.size();i++){
 			long val = elements.get(i).longValue();
-			max = val>max ? val : max;
+			max = Math.max(val, max);
 			numentries++;
 		}
 

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/HDTImpl.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/hdt/impl/HDTImpl.java
@@ -786,7 +786,7 @@ public void diffBit(String location, HDT hdt, Bitmap deleteBitmap, ProgressListe
 			log.debug("Generating Triples...");
 			il.notifyProgress(40, "Generating Triples...");
 			// map the triples based on the new dictionary
-			BitmapTriplesIteratorMapDiff mapIter = new BitmapTriplesIteratorMapDiff(hdt, deleteBitmap, diff, iter.getCount() + 1);
+			BitmapTriplesIteratorMapDiff mapIter = new BitmapTriplesIteratorMapDiff(hdt, deleteBitmap, diff);
 
 			BitmapTriples triples = new BitmapTriples(spec);
 			triples.load(mapIter, listener);

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/triples/impl/BitmapTriples.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/triples/impl/BitmapTriples.java
@@ -129,8 +129,8 @@ public void load(IteratorTripleID it, ProgressListener listener) {
 
 		long number = it.estimatedNumResults();
 
-		SequenceLog64 vectorY = new SequenceLog64(BitUtil.log2(number), number);
-		SequenceLog64 vectorZ = new SequenceLog64(BitUtil.log2(number), number);
+		DynamicSequence vectorY = new SequenceLog64Big(BitUtil.log2(number), number);
+		DynamicSequence vectorZ = new SequenceLog64Big(BitUtil.log2(number), number);
 
 		ModifiableBitmap bitY = new Bitmap375(number);
 		ModifiableBitmap bitZ = new Bitmap375(number);

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/triples/impl/BitmapTriplesIteratorMapDiff.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/triples/impl/BitmapTriplesIteratorMapDiff.java
@@ -32,13 +32,13 @@ public class BitmapTriplesIteratorMapDiff implements IteratorTripleID {
     TripleIDComparator tripleIDComparator = new TripleIDComparator(TripleComponentOrder.SPO);
     Bitmap bitArrayDisk;
 
-    public BitmapTriplesIteratorMapDiff(HDT hdtOriginal, Bitmap deleteBitmap, DictionaryDiff dictionaryDiff, long countTriples) {
+    public BitmapTriplesIteratorMapDiff(HDT hdtOriginal, Bitmap deleteBitmap, DictionaryDiff dictionaryDiff) {
         this.subjMapping = dictionaryDiff.getAllMappings().get("subject");
         this.objMapping = dictionaryDiff.getAllMappings().get("object");
         this.predMapping = dictionaryDiff.getAllMappings().get("predicate");
         this.sharedMapping = dictionaryDiff.getAllMappings().get("shared");
         this.dictionaryDiff = dictionaryDiff;
-        this.countTriples = countTriples;
+        this.countTriples = Math.max(0, hdtOriginal.getTriples().getNumberOfElements() - deleteBitmap.countOnes());
         this.triples = hdtOriginal.getTriples();
         this.bitArrayDisk = deleteBitmap;
         list = getTripleID(0).listIterator();

diff --git a/hdt-java-core/src/main/java/org/rdfhdt/hdt/util/io/IOUtil.java b/hdt-java-core/src/main/java/org/rdfhdt/hdt/util/io/IOUtil.java
@@ -30,7 +30,7 @@
 import org.apache.commons.compress.compressors.xz.XZCompressorInputStream;
 import org.rdfhdt.hdt.listener.ProgressListener;
 import org.rdfhdt.hdt.util.string.ByteStringUtil;
-import pl.edu.icm.jlargearrays.LargeArrayUtils;
+import org.visnow.jlargearrays.LargeArrayUtils;
 
 import java.io.*;
 import java.net.URL;

diff --git a/hdt-java-core/src/test/java/org/rdfhdt/hdt/compact/sequence/LargeArrayTest.java b/hdt-java-core/src/test/java/org/rdfhdt/hdt/compact/sequence/LargeArrayTest.java
@@ -0,0 +1,20 @@
+package org.rdfhdt.hdt.compact.sequence;
+
+import org.junit.Test;
+import org.visnow.jlargearrays.LargeArray;
+import org.visnow.jlargearrays.LongLargeArray;
+
+public class LargeArrayTest {
+
+	@Test
+	public void allocationTest() {
+		int old = LargeArray.getMaxSizeOf32bitArray();
+		try {
+			LargeArray.setMaxSizeOf32bitArray(100);
+			long size = LargeArray.getMaxSizeOf32bitArray() + 2L;
+			new LongLargeArray(size);
+		} finally {
+			LargeArray.setMaxSizeOf32bitArray(old);
+		}
+	}
+}