Merge pull request #9892 from IoannisPanagiotas/node-sim-and-wcc-pt

Make WCC's progress trackable during node sim

Author: IoannisPanagiotas

algo/src/main/java/org/neo4j/gds/similarity/nodesim/NodeSimilarity.java

@@ -38,8 +38,8 @@
 import org.neo4j.gds.similarity.SimilarityResult;
 import org.neo4j.gds.similarity.filtering.NodeFilter;
 import org.neo4j.gds.termination.TerminationFlag;
-import org.neo4j.gds.wcc.WccStub;
 import org.neo4j.gds.wcc.WccParameters;
+import org.neo4j.gds.wcc.WccStub;
 
 import java.util.Objects;
 import java.util.Optional;
@@ -173,21 +173,32 @@ private void prepare() {
         progressTracker.beginSubTask();
 
         components = initComponents();
-        if (parameters.runWCC()) {
-            progressTracker.beginSubTask();
+
+        if (parameters.runWCC()){
+            prepareAsChildTask();
+        }else{
+            prepareAsLeaf();
         }
-        initNodeSpecificFields();
 
-        sourceNodesStream = initSourceNodesStream();
+        progressTracker.endSubTask();
+    }
 
-        targetNodesStream = initTargetNodesStream();
+    private  void prepareAsLeaf(){
+        setUpNodesAndFilters();
+    }
 
-        if (parameters.runWCC()) {
-            progressTracker.endSubTask();
-        }
+    private  void prepareAsChildTask(){
+        progressTracker.beginSubTask();
+        setUpNodesAndFilters();
         progressTracker.endSubTask();
     }
 
+    private void setUpNodesAndFilters(){
+        initNodeSpecificFields();
+        sourceNodesStream = initSourceNodesStream();
+        targetNodesStream = initTargetNodesStream();
+    }
+
     private Stream<SimilarityResult> computeSimilarityResultStream() {
         if (parameters.hasTopK()) {
             var topKMap = computeTopKMap();
@@ -219,10 +230,8 @@ private LongUnaryOperator initComponents() {
         }
 
         // run WCC to determine components
-        progressTracker.beginSubTask();
         var wccParameters = new WccParameters(0D, concurrency);
-        var disjointSets = wccStub.wcc(graph, wccParameters, ProgressTracker.NULL_TRACKER, false);
-        progressTracker.endSubTask();
+        var disjointSets = wccStub.wcc(graph, wccParameters, progressTracker, false);
         return disjointSets::setIdOf;
     }
 

algo/src/test/java/org/neo4j/gds/similarity/nodesim/NodeSimilarityTest.java

@@ -902,6 +902,14 @@ void shouldLogProgressForWccOptimization() {
                 "Node Similarity :: compare node pairs :: Start",
                 "Node Similarity :: compare node pairs :: Finished"
             );
+
+       assertThat(log.getMessages(INFO)
+                .stream()
+                .filter( v -> v.contains("WCC"))
+                .count())
+           .isGreaterThan(3);
+
+
     }
 
     @Test