Add hashCode() and equals() to the value class of ExprJavaType

core/src/main/java/org/opensearch/sql/calcite/type/ExprJavaType.java

@@ -14,7 +14,9 @@
 
 /**
  * The JavaType for ExprUDT. The UDT which needs to use self-implemented java class should extend
- * this.
+ * this. Its javaClazz should override equals() and hashCode() methods. For example, {@link
+ * org.opensearch.sql.data.model.ExprIpValue} (javaClazz of {@link ExprIPType}) overrides the
+ * equals() and hashCode().
  */
 public class ExprJavaType extends AbstractExprRelDataType<JavaType> {
   public ExprJavaType(OpenSearchTypeFactory typeFactory, ExprUDT exprUDT, Class<?> javaClazz) {

core/src/main/java/org/opensearch/sql/data/model/ExprIpValue.java

@@ -6,11 +6,13 @@
 package org.opensearch.sql.data.model;
 
 import inet.ipaddr.IPAddress;
+import lombok.EqualsAndHashCode;
 import org.opensearch.sql.data.type.ExprCoreType;
 import org.opensearch.sql.data.type.ExprType;
 import org.opensearch.sql.utils.IPUtils;
 
 /** Expression IP Address Value. */
+@EqualsAndHashCode(callSuper = false)
 public class ExprIpValue extends AbstractExprValue {
   private final IPAddress value;
 

integ-test/src/yamlRestTest/resources/rest-api-spec/test/issues/4726.yml

@@ -0,0 +1,83 @@
+setup:
+  - do:
+      query.settings:
+        body:
+          transient:
+            plugins.calcite.enabled : true
+  - do:
+      indices.create:
+        index: test1
+        body:
+          mappings:
+            properties:
+              "timestamp":
+                type: date
+              "status":
+                type: integer
+              "client_ip":
+                type: ip
+  - do:
+      indices.create:
+        index: test2
+        body:
+          mappings:
+            properties:
+              "client_ip":
+                type: ip
+              "city":
+                type: keyword
+              "country":
+                type: keyword
+  - do:
+      bulk:
+        index: test1
+        refresh: true
+        body:
+          - '{"index":{}}'
+          - '{"datetime":"2025-01-15T00:30:00Z","status":200,"client_ip":"10.0.0.1"}'
+          - '{"index":{}}'
+          - '{"datetime":"2025-01-15T02:15:00Z","status":200,"client_ip":"10.0.0.2"}'
+          - '{"index":{}}'
+          - '{"datetime":"2025-01-15T10:50:00Z","status":200,"client_ip":"10.0.0.11"}'
+          - '{"index":{}}'
+          - '{"datetime":"2025-01-15T23:45:00Z","status":200,"client_ip":"10.0.0.24"}'
+  - do:
+      bulk:
+        index: test2
+        refresh: true
+        body:
+          - '{"index":{}}'
+          - '{"client_ip": "10.0.0.1","country": "Canada","city": "Toronto"}'
+          - '{"index":{}}'
+          - '{"client_ip": "10.0.0.24","country": "UK","city": "London"}'
+          - '{"index":{}}'
+          - '{"client_ip": "10.0.1.1","country": "USA","city": "New York"}'
+          - '{"index":{}}'
+          - '{"client_ip": "10.0.1.2","country": "USA","city": "Seattle"}'
+
+---
+teardown:
+  - do:
+      query.settings:
+        body:
+          transient:
+            plugins.calcite.enabled : false
+
+---
+"hash join on IP type should work":
+  - skip:
+      features:
+        - headers
+        - allowed_warnings
+  - do:
+      allowed_warnings:
+        - 'Loading the fielddata on the _id field is deprecated and will be removed in future versions. If you require sorting or aggregating on this field you should also include the id in the body of your documents, and map this field as a keyword field that has [doc_values] enabled'
+      headers:
+        Content-Type: 'application/json'
+      ppl:
+        body:
+          query: source=test1 | stats count() as cnt by client_ip | join type=inner client_ip test2 | fields client_ip, cnt
+
+
+  - match: { total: 2 }
+  - match: {"datarows": [["10.0.0.1", 1], ["10.0.0.24", 1]]}