fix array OOBE in blocked bloom filter when top 4 bits of hash are set (seed dependent behaviour)

Author: richardstartin

fastfilter/src/main/java/org/fastfilter/bloom/BlockedBloom.java

@@ -12,8 +12,8 @@
 public class BlockedBloom implements Filter {
 
     public static BlockedBloom construct(long[] keys, int bitsPerKey) {
-        long n = keys.length;
-        BlockedBloom f = new BlockedBloom((int) n, bitsPerKey);
+        int n = keys.length;
+        BlockedBloom f = new BlockedBloom(n, bitsPerKey);
         for(long x : keys) {
             f.add(x);
         }
@@ -34,7 +34,7 @@ public long getBitCount() {
         this.seed = Hash.randomSeed();
         long bits = (long) entryCount * bitsPerKey;
         this.buckets = (int) bits / 64;
-        data = new long[(int) (buckets + 16)];
+        data = new long[(buckets + 16)];
     }
 
     @Override
@@ -50,7 +50,7 @@ public void add(long key) {
         long m1 = (1L << hash) | (1L << (hash >> 6));
         long m2 = (1L << (hash >> 12)) | (1L << (hash >> 18));
         data[start] |= m1;
-        data[start + 1 + (int) (hash >>> 60)] |= m2;
+        data[start + (int) (hash >>> 60)] |= m2;
     }
 
     @Override
@@ -59,7 +59,7 @@ public boolean mayContain(long key) {
         int start = Hash.reduce((int) hash, buckets);
         hash = hash ^ Long.rotateLeft(hash, 32);
         long a = data[start];
-        long b = data[start + 1 + (int) (hash >>> 60)];
+        long b = data[start + (int) (hash >>> 60)];
         long m1 = (1L << hash) | (1L << (hash >> 6));
         long m2 = (1L << (hash >> 12)) | (1L << (hash >> 18));
         return ((m1 & a) == m1) && ((m2 & b) == m2);

fastfilter/src/test/java/org/fastfilter/bloom/BlockedBloomTest.java

@@ -0,0 +1,19 @@
+package org.fastfilter.bloom;
+
+import org.fastfilter.utils.Hash;
+import org.junit.Test;
+
+import static org.junit.Assert.assertTrue;
+
+public class BlockedBloomTest {
+
+    @Test
+    public void testCreateSmallBlockedBloomFilter() {
+        Hash.setSeed(872153271794238865L);
+        BlockedBloom filter = BlockedBloom.construct(new long[]{1, 2, 3}, 8);
+        assertTrue(filter.mayContain(1));
+        assertTrue(filter.mayContain(2));
+        assertTrue(filter.mayContain(3));
+    }
+
+}