Merge pull request #324 from dynatrace-oss/distinct-count-util

make utility function for the deduplication for hash tokens public
dynatrace-oss · Feb 4, 2025 · a902c9e · a902c9e
2 parents 13e58de + b809887
commit a902c9e
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 20 deletions.
diff --git a/src/main/java/com/dynatrace/hash4j/distinctcount/DistinctCountUtil.java b/src/main/java/com/dynatrace/hash4j/distinctcount/DistinctCountUtil.java
@@ -19,7 +19,8 @@
 
 import java.util.Arrays;
 
-class DistinctCountUtil {
+/** A utility functions for distinct counting. */
+public final class DistinctCountUtil {
 
   private DistinctCountUtil() {}
 
@@ -191,13 +192,13 @@ static double solveMaximumLikelihoodEquation(
     return x;
   }
 
-  static int computeToken1(long hashValue) {
+  static int computeToken(long hashValue) {
     int idx = (int) (hashValue >>> 38);
     int nlz = Long.numberOfLeadingZeros(~(~hashValue << 26));
     return (idx << 6) | nlz;
   }
 
-  static long reconstructHash1(int token) {
+  static long reconstructHash(int token) {
     long idx = token & 0xFFFFFFC0L;
     return (0x3FFFFFFFFFL >>> token) | (idx << 32);
   }

diff --git a/src/main/java/com/dynatrace/hash4j/distinctcount/HyperLogLog.java b/src/main/java/com/dynatrace/hash4j/distinctcount/HyperLogLog.java
@@ -286,7 +286,7 @@ public HyperLogLog add(long hashValue) {
    */
   @Override
   public HyperLogLog addToken(int token) {
-    return add(DistinctCountUtil.reconstructHash1(token));
+    return add(DistinctCountUtil.reconstructHash(token));
   }
 
   /**
@@ -309,7 +309,7 @@ public HyperLogLog addToken(int token) {
    * @return the 32-bit token
    */
   public static int computeToken(long hashValue) {
-    return DistinctCountUtil.computeToken1(hashValue);
+    return DistinctCountUtil.computeToken(hashValue);
   }
 
   /**
@@ -353,7 +353,7 @@ public HyperLogLog add(long hashValue, StateChangeObserver stateChangeObserver)
    */
   @Override
   public HyperLogLog addToken(int token, StateChangeObserver stateChangeObserver) {
-    return add(DistinctCountUtil.reconstructHash1(token), stateChangeObserver);
+    return add(DistinctCountUtil.reconstructHash(token), stateChangeObserver);
   }
 
   // returns register change probability scaled by 2^64

diff --git a/src/main/java/com/dynatrace/hash4j/distinctcount/UltraLogLog.java b/src/main/java/com/dynatrace/hash4j/distinctcount/UltraLogLog.java
@@ -203,7 +203,7 @@ public UltraLogLog add(long hashValue) {
    */
   @Override
   public UltraLogLog addToken(int token) {
-    return add(DistinctCountUtil.reconstructHash1(token));
+    return add(DistinctCountUtil.reconstructHash(token));
   }
 
   /**
@@ -226,7 +226,7 @@ public UltraLogLog addToken(int token) {
    * @return the 32-bit token
    */
   public static int computeToken(long hashValue) {
-    return DistinctCountUtil.computeToken1(hashValue);
+    return DistinctCountUtil.computeToken(hashValue);
   }
 
   /**
@@ -275,7 +275,7 @@ public UltraLogLog add(long hashValue, StateChangeObserver stateChangeObserver)
    */
   @Override
   public UltraLogLog addToken(int token, StateChangeObserver stateChangeObserver) {
-    return add(DistinctCountUtil.reconstructHash1(token), stateChangeObserver);
+    return add(DistinctCountUtil.reconstructHash(token), stateChangeObserver);
   }
 
   /**

diff --git a/src/test/java/com/dynatrace/hash4j/distinctcount/DistinctCountUtilTest.java b/src/test/java/com/dynatrace/hash4j/distinctcount/DistinctCountUtilTest.java
@@ -162,7 +162,7 @@ void testSolveMaximumLikelihoodEquation() {
   }
 
   @Test
-  void testComputeToken1() {
+  void testComputeToken() {
     SplittableRandom random = new SplittableRandom(0xbafc97ad730480acL);
 
     int numCycles = 100;
@@ -175,20 +175,18 @@ void testComputeToken1() {
         long mask = 0xFFFFFFC000000000L | (0x0000003FFFFFFFFFL >>> nlz);
         long hash = (r | (0x0000002000000000L >>> nlz)) & mask;
 
-        int token = DistinctCountUtil.computeToken1(hash);
-        long reconstructedHash = DistinctCountUtil.reconstructHash1(token);
-        int tokenFromReconstructedHash = DistinctCountUtil.computeToken1(reconstructedHash);
+        int token = DistinctCountUtil.computeToken(hash);
+        long reconstructedHash = DistinctCountUtil.reconstructHash(token);
+        int tokenFromReconstructedHash = DistinctCountUtil.computeToken(reconstructedHash);
         assertThat(reconstructedHash).isEqualTo(hash | (0x0000001FFFFFFFFFL >>> nlz));
         assertThat(tokenFromReconstructedHash).isEqualTo(token);
       }
     }
   }
 
   private static TokenIterable fromSortedArray(int[] tokens) {
-    return new TokenIterable() {
-      @Override
-      public TokenIterator iterator() {
-        return new TokenIterator() {
+    return () ->
+        new TokenIterator() {
           private int idx = 0;
 
           @Override
@@ -201,8 +199,6 @@ public int nextToken() {
             return tokens[idx++];
           }
         };
-      }
-    };
   }
 
   private static void testEstimationFromTokens(int distinctCount) {
@@ -215,7 +211,7 @@ private static void testEstimationFromTokens(int distinctCount) {
 
     for (int i = 0; i < numIterations; ++i) {
       for (int c = 0; c < distinctCount; ++c) {
-        tokens[c] = DistinctCountUtil.computeToken1(prg.nextLong());
+        tokens[c] = DistinctCountUtil.computeToken(prg.nextLong());
       }
       Arrays.sort(tokens);