Fix issue with not computing numberofpairs if should is false

IoannisPanagiotas · IoannisPanagiotas · commit 79c9c3af71f4 · 2026-03-16T17:28:34.000+01:00
diff --git a/algo-common/src/main/java/org/neo4j/gds/result/SimilarityStatistics.java b/algo-common/src/main/java/org/neo4j/gds/result/SimilarityStatistics.java
@@ -93,6 +93,6 @@ public record  Histogram(Optional<DoubleHistogram> histogram,  boolean success)
 
     public record  SimilarityStats(Optional<DoubleHistogram> histogram, long computeMilliseconds, boolean success) {}
 
-    public record  SimilarityDistributionResults(long numberOfEntries, Map<String,Object> distribution, long computeMilliseconds) {}
+    public record  SimilarityDistributionResults(long numberOfSimilarityPairs, Map<String,Object> distribution, long computeMilliseconds) {}
 
 }
diff --git a/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/FilteredKnnResultBuilderForStatsMode.java b/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/FilteredKnnResultBuilderForStatsMode.java
@@ -27,6 +27,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Optional;
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 class FilteredKnnResultBuilderForStatsMode implements StatsResultBuilder<FilteredKnnResult, Stream<KnnStatsResult>> {
@@ -64,7 +65,8 @@ public Stream<KnnStatsResult> build(
             configuration.concurrency(),
             filteredKnnResult.similarityResultStream(),
             shouldComputeSimilarityDistribution,
-            terminationFlag
+            terminationFlag,
+            OptionalLong.of(filteredKnnResult.numberOfSimilarityPairs())
         );
 
         return Stream.of(
@@ -73,7 +75,7 @@ public Stream<KnnStatsResult> build(
                 timings.computeMillis,
                 similarityStats.computeMilliseconds(),
                 filteredKnnResult.nodesCompared(),
-                filteredKnnResult.numberOfSimilarityPairs(),
+                similarityStats.numberOfSimilarityPairs(),
                 similarityStats.distribution(),
                 filteredKnnResult.didConverge(),
                 filteredKnnResult.ranIterations(),
diff --git a/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/GenericNodeSimilarityResultBuilderForStatsMode.java b/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/GenericNodeSimilarityResultBuilderForStatsMode.java
@@ -26,6 +26,7 @@
 
 import java.util.Map;
 import java.util.Optional;
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 class GenericNodeSimilarityResultBuilderForStatsMode {
@@ -51,7 +52,8 @@ Stream<SimilarityStatsResult> build(
             nodeSimilarityResult.isStream() ? concurrency : new Concurrency(1),
             nodeSimilarityResult.isStream()? nodeSimilarityResult.streamResult() : nodeSimilarityResult.topKMap().stream(),
             shouldComputeSimilarityDistribution,
-            terminationFlag
+            terminationFlag,
+            OptionalLong.empty()
         );
 
         return Stream.of(
@@ -60,7 +62,7 @@ Stream<SimilarityStatsResult> build(
                 timings.computeMillis,
                 similarityStats.computeMilliseconds(),
                 nodeSimilarityResult.comparedNodes(),
-                similarityStats.numberOfEntries(),
+                similarityStats.numberOfSimilarityPairs(),
                 similarityStats.distribution(),
                 configurationMap
             )
diff --git a/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/KnnResultBuilderForStatsMode.java b/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/KnnResultBuilderForStatsMode.java
@@ -27,6 +27,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Optional;
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 class KnnResultBuilderForStatsMode implements StatsResultBuilder<KnnResult, Stream<KnnStatsResult>> {
@@ -65,7 +66,8 @@ public Stream<KnnStatsResult> build(
             configuration.concurrency(),
             knnResult.streamSimilarityResult(),
             shouldComputeSimilarityDistribution,
-            terminationFlag
+            terminationFlag,
+            OptionalLong.of(knnResult.totalSimilarityPairs())
         );
 
         return Stream.of(
@@ -74,7 +76,7 @@ public Stream<KnnStatsResult> build(
                 timings.computeMillis,
                 similarityStats.computeMilliseconds(),
                 knnResult.nodesCompared(),
-                knnResult.totalSimilarityPairs(),
+                similarityStats.numberOfSimilarityPairs(),
                 similarityStats.distribution(),
                 knnResult.didConverge(),
                 knnResult.ranIterations(),
diff --git a/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/SimilarityStatsProcessor.java b/procedures/algorithms-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/SimilarityStatsProcessor.java
@@ -29,6 +29,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Map;
+import java.util.OptionalLong;
 import java.util.concurrent.atomic.AtomicLong;
 import java.util.concurrent.atomic.LongAdder;
 import java.util.stream.Stream;
@@ -46,9 +47,16 @@ SimilarityStatistics.SimilarityDistributionResults computeSimilarityDistribution
         Concurrency concurrency,
         Stream<SimilarityResult> similarityResultStream,
         boolean shouldComputeSimilarityDistribution,
-        TerminationFlag terminationFlag
+        TerminationFlag terminationFlag,
+        OptionalLong numberOfEntries
     ) {
-        if (!shouldComputeSimilarityDistribution) return EMPTY;
+        if (!shouldComputeSimilarityDistribution && numberOfEntries.isPresent()){
+            return new SimilarityStatistics.SimilarityDistributionResults(
+                numberOfEntries.getAsLong(),
+                Map.of(),
+                0
+            );
+        }
 
         var statsMillis = new AtomicLong();
         Map<String,Object> distribution;
@@ -71,7 +79,7 @@ SimilarityStatistics.SimilarityDistributionResults computeSimilarityDistribution
         }
 
         return new SimilarityStatistics.SimilarityDistributionResults(
-            adder.longValue(),
+            numberOfEntries.orElse(adder.longValue()),
             distribution,
             statsMillis.get()
         );
diff --git a/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/FilteredKnnStatsResultTransformer.java b/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/FilteredKnnStatsResultTransformer.java
@@ -28,6 +28,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Map;
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 public class FilteredKnnStatsResultTransformer implements ResultTransformer<TimedAlgorithmResult<FilteredKnnResult>, Stream<KnnStatsResult>> {
@@ -68,7 +69,8 @@ public Stream<KnnStatsResult> apply(TimedAlgorithmResult<FilteredKnnResult> time
             concurrency,
             knnResult.similarityResultStream(),
             shouldComputeSimilarityDistribution,
-            terminationFlag
+            terminationFlag,
+            OptionalLong.of(knnResult.numberOfSimilarityPairs())
         );
 
 
@@ -78,7 +80,7 @@ public Stream<KnnStatsResult> apply(TimedAlgorithmResult<FilteredKnnResult> time
                 timedAlgorithmResult.computeMillis(),
                 similarityStats.computeMilliseconds(),
                 knnResult.nodesCompared(),
-                knnResult.numberOfSimilarityPairs(),
+                similarityStats.numberOfSimilarityPairs(),
                 similarityStats.distribution(),
                 knnResult.didConverge(),
                 knnResult.ranIterations(),
diff --git a/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/KnnStatsResultTransformer.java b/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/KnnStatsResultTransformer.java
@@ -28,6 +28,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Map;
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 public class KnnStatsResultTransformer implements ResultTransformer<TimedAlgorithmResult<KnnResult>, Stream<KnnStatsResult>> {
@@ -68,7 +69,8 @@ public Stream<KnnStatsResult> apply(TimedAlgorithmResult<KnnResult> timedAlgorit
             concurrency,
             knnResult.streamSimilarityResult(),
             shouldComputeSimilarityDistribution,
-            terminationFlag
+            terminationFlag,
+            OptionalLong.of(knnResult.totalSimilarityPairs())
         );
 
 
@@ -78,7 +80,7 @@ public Stream<KnnStatsResult> apply(TimedAlgorithmResult<KnnResult> timedAlgorit
                 timedAlgorithmResult.computeMillis(),
                 similarityStats.computeMilliseconds(),
                 knnResult.nodesCompared(),
-                knnResult.totalSimilarityPairs(),
+                similarityStats.numberOfSimilarityPairs(),
                 similarityStats.distribution(),
                 knnResult.didConverge(),
                 knnResult.ranIterations(),
diff --git a/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/NodeSimilarityStatsResultTransformer.java b/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/NodeSimilarityStatsResultTransformer.java
@@ -27,6 +27,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Map;
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 public class NodeSimilarityStatsResultTransformer implements ResultTransformer<TimedAlgorithmResult<NodeSimilarityResult>, Stream<SimilarityStatsResult>> {
@@ -65,7 +66,8 @@ public Stream<SimilarityStatsResult> apply(TimedAlgorithmResult<NodeSimilarityRe
             result.isStream() ? concurrency : new Concurrency(1),
             result.isStream()? result.streamResult() : result.topKMap().stream(),
             shouldComputeSimilarityDistribution,
-            terminationFlag
+            terminationFlag,
+            OptionalLong.empty()
         );
 
         return Stream.of(
@@ -74,7 +76,7 @@ public Stream<SimilarityStatsResult> apply(TimedAlgorithmResult<NodeSimilarityRe
                 timedAlgorithmResult.computeMillis(),
                 similarityStats.computeMilliseconds(),
                 result.comparedNodes(),
-                similarityStats.numberOfEntries(),
+                similarityStats.numberOfSimilarityPairs(),
                 similarityStats.distribution(),
                 configuration
             )
diff --git a/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/SimilarityStatsTools.java b/procedures/pushback-procedures-facade/src/main/java/org/neo4j/gds/procedures/algorithms/similarity/stats/SimilarityStatsTools.java
@@ -19,8 +19,8 @@
  */
 package org.neo4j.gds.procedures.algorithms.similarity.stats;
 
-import org.neo4j.gds.algorithms.similarity.ActualSimilaritySummaryBuilder;
 import org.neo4j.gds.algorithms.similarity.SimilarityResultStreamDelegate;
+import org.neo4j.gds.algorithms.similarity.SimilaritySummaryBuilderFactory;
 import org.neo4j.gds.api.properties.relationships.RelationshipWithPropertyConsumer;
 import org.neo4j.gds.core.concurrency.Concurrency;
 import org.neo4j.gds.core.utils.ProgressTimer;
@@ -29,6 +29,7 @@
 import org.neo4j.gds.termination.TerminationFlag;
 
 import java.util.Map;
+import java.util.OptionalLong;
 import java.util.concurrent.atomic.AtomicLong;
 import java.util.concurrent.atomic.LongAdder;
 import java.util.stream.Stream;
@@ -47,20 +48,28 @@ static SimilarityStatistics.SimilarityDistributionResults computeSimilarityDistr
         Concurrency concurrency,
         Stream<SimilarityResult> similarityResultStream,
         boolean shouldComputeSimilarityDistribution,
-        TerminationFlag terminationFlag
+        TerminationFlag terminationFlag,
+        OptionalLong numberOfEntries
     ) {
-        if (!shouldComputeSimilarityDistribution) return EMPTY;
 
+        if (!shouldComputeSimilarityDistribution && numberOfEntries.isPresent()){
+            return new SimilarityStatistics.SimilarityDistributionResults(
+                numberOfEntries.getAsLong(),
+                Map.of(),
+                0
+            );
+        }
         var statsMillis = new AtomicLong();
         Map<String,Object> distribution;
-        var similaritySummaryBuilder  =  ActualSimilaritySummaryBuilder.create(concurrency);
+        var similaritySummaryBuilder  =  SimilaritySummaryBuilderFactory.create(concurrency,shouldComputeSimilarityDistribution);
         LongAdder adder = new LongAdder();
 
         RelationshipWithPropertyConsumer relationshipWithPropertyConsumer= (s,t,w)->{
             adder.increment();
             similaritySummaryBuilder.accept(s,t,w);
             return true;
         };
+
         var similarityResultStreamDelegate = new SimilarityResultStreamDelegate();
         try (var ignored = ProgressTimer.start(statsMillis::set)) {
             similarityResultStreamDelegate.consumeStream(
@@ -74,7 +83,7 @@ static SimilarityStatistics.SimilarityDistributionResults computeSimilarityDistr
         }
 
         return new SimilarityStatistics.SimilarityDistributionResults(
-            adder.longValue(),
+            numberOfEntries.orElse(adder.longValue()),
             distribution,
             statsMillis.get()
         );
diff --git a/procedures/pushback-procedures-facade/src/test/java/org/neo4j/gds/procedures/algorithms/similarity/stats/SimilarityStatsToolsTest.java b/procedures/pushback-procedures-facade/src/test/java/org/neo4j/gds/procedures/algorithms/similarity/stats/SimilarityStatsToolsTest.java
@@ -25,6 +25,7 @@
 import org.neo4j.gds.similarity.SimilarityResult;
 import org.neo4j.gds.termination.TerminationFlag;
 
+import java.util.OptionalLong;
 import java.util.stream.Stream;
 
 import static org.assertj.core.api.Assertions.assertThat;
@@ -39,9 +40,25 @@ void shouldReturnEmptyDistributionIfNotSpecified() {
             new Concurrency(1),
             Stream.of(new SimilarityResult(0, 1, 10)),
             false,
-            TerminationFlag.RUNNING_TRUE
+            TerminationFlag.RUNNING_TRUE,
+            OptionalLong.empty()
         );
         assertThat(stats.distribution()).isEmpty();
+        assertThat(stats.numberOfSimilarityPairs()).isEqualTo(1);
+    }
+
+    @Test
+    void shouldReturnEmptyDistributionIfNotSpecifiedButPairsProvided() {
+
+        var stats = SimilarityStatsTools.computeSimilarityDistribution(
+            new Concurrency(1),
+            Stream.of(new SimilarityResult(0, 1, 10)),
+            false,
+            TerminationFlag.RUNNING_TRUE,
+            OptionalLong.of(100)
+        );
+        assertThat(stats.distribution()).isEmpty();
+        assertThat(stats.numberOfSimilarityPairs()).isEqualTo(100);
     }
 
     @Test
@@ -55,11 +72,32 @@ void shouldReturnValidDistributionIfTrue(){
                 new SimilarityResult(1,3,9)
             ),
             true,
-            TerminationFlag.RUNNING_TRUE
+            TerminationFlag.RUNNING_TRUE,
+            OptionalLong.empty()
+        );
+        assertThat(stats.distribution().get("mean")).asInstanceOf(DOUBLE).isCloseTo(8.0, Offset.offset(1e-3));
+        assertThat(stats.computeMilliseconds()).isGreaterThanOrEqualTo(0L);
+        assertThat(stats.numberOfSimilarityPairs()).isEqualTo(3L);
+
+    }
+
+    @Test
+    void shouldReturnValueBasedOnProvided(){
+
+        var stats = SimilarityStatsTools.computeSimilarityDistribution(
+            new Concurrency(1),
+            Stream.of(
+                new SimilarityResult(0, 1, 10),
+                new SimilarityResult(1,2,5),
+                new SimilarityResult(1,3,9)
+            ),
+            true,
+            TerminationFlag.RUNNING_TRUE,
+            OptionalLong.of(100L)
         );
         assertThat(stats.distribution().get("mean")).asInstanceOf(DOUBLE).isCloseTo(8.0, Offset.offset(1e-3));
         assertThat(stats.computeMilliseconds()).isGreaterThanOrEqualTo(0L);
-        assertThat(stats.numberOfEntries()).isEqualTo(3L);
+        assertThat(stats.numberOfSimilarityPairs()).isEqualTo(100L);
 
     }
 

Original file line number	Diff line number	Diff line change
`@@ -93,6 +93,6 @@ public record Histogram(Optional<DoubleHistogram> histogram, boolean success)`
`93`	`93`
`94`	`94`	`public record SimilarityStats(Optional<DoubleHistogram> histogram, long computeMilliseconds, boolean success) {}`
`95`	`95`
`96`		`- public record SimilarityDistributionResults(long numberOfEntries, Map<String,Object> distribution, long computeMilliseconds) {}`
	`96`	`+ public record SimilarityDistributionResults(long numberOfSimilarityPairs, Map<String,Object> distribution, long computeMilliseconds) {}`
`97`	`97`
`98`	`98`	`}`