Merge pull request #10295 from neo-technology/snowgraph-deduplicate-node-id-projections

Deduplicate snowgraph node ids

Author: knutwalker

core/src/main/java/org/neo4j/gds/core/utils/paged/ShardedByteArrayLongMap.java

@@ -246,8 +246,12 @@ private Shard(AtomicLong nextId, long capacity) {
 
             long addNode(byte[] nodeId) {
                 this.assertIsUnderLock();
+                int index = mapping.indexOf(nodeId);
+                if (mapping.indexExists(index)) {
+                    return -mapping.indexGet(index) - 1;
+                }
                 long mappedId = nextId.getAndIncrement();
-                mapping.put(nodeId, mappedId);
+                mapping.indexInsert(index, nodeId, mappedId);
                 return mappedId;
             }
         }

core/src/test/java/org/neo4j/gds/core/utils/paged/ShardedByteArrayLongMapTest.java

@@ -100,14 +100,14 @@ void addNodesDifferentObject(@ForAll("nodes") @Size(100) byte[][] nodes) {
     }
 
     @Test
-    void doesNotActuallyDeduplicateNodes() {
+    void addExistingNode() {
         byte[] node1 = "foobar".getBytes(StandardCharsets.UTF_8);
         byte[] node2 = "foobar".getBytes(StandardCharsets.UTF_8);
         var builder = ShardedByteArrayLongMap.builder(new Concurrency(1));
         long mappedNode1 = builder.addNode(node1);
         assertThat(mappedNode1).isGreaterThanOrEqualTo(0);
         long mappedNode2 = builder.addNode(node2);
-        assertThat(mappedNode2).isGreaterThanOrEqualTo(0).isNotEqualTo(mappedNode1);
+        assertThat(mappedNode2).isEqualTo(-(mappedNode1 + 1));
     }
 
     @ParameterizedTest