(paimon) branch master updated: [common] Read VECTOR fields from NestedRow (#9864)

[lzljs3620320]

This is an automated email from the ASF dual-hosted git repository.

JingsongLi pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/paimon.git


The following commit(s) were added to refs/heads/master by this push:
     new 599c728f8c [common] Read VECTOR fields from NestedRow (#9864)
599c728f8c is described below

commit 599c728f8cad9bea925d8474f22ebeac5cc1d6dc
Author: Eunbin Son <58901024+thswls...@users.noreply.github.com>
AuthorDate: Thu Sep 17 15:33:23 2026 +0900

    [common] Read VECTOR fields from NestedRow (#9864)
---
 .../java/org/apache/paimon/data/NestedRow.java     |  3 +-
 .../java/org/apache/paimon/data/NestedRowTest.java | 38 ++++++++++++++++++++++
 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/paimon-common/src/main/java/org/apache/paimon/data/NestedRow.java 
b/paimon-common/src/main/java/org/apache/paimon/data/NestedRow.java
index 9f1be1e7b6..4944cf20ad 100644
--- a/paimon-common/src/main/java/org/apache/paimon/data/NestedRow.java
+++ b/paimon-common/src/main/java/org/apache/paimon/data/NestedRow.java
@@ -308,7 +308,8 @@ public final class NestedRow extends BinarySection 
implements InternalRow, DataS
 
     @Override
     public InternalVector getVector(int pos) {
-        throw new IllegalArgumentException("Unsupported type: VectorType");
+        assertIndexIsValid(pos);
+        return MemorySegmentUtils.readVectorData(segments, offset, 
getLong(pos));
     }
 
     @Override
diff --git 
a/paimon-common/src/test/java/org/apache/paimon/data/NestedRowTest.java 
b/paimon-common/src/test/java/org/apache/paimon/data/NestedRowTest.java
index 0b4903760d..ea5cddfc26 100644
--- a/paimon-common/src/test/java/org/apache/paimon/data/NestedRowTest.java
+++ b/paimon-common/src/test/java/org/apache/paimon/data/NestedRowTest.java
@@ -21,6 +21,7 @@ package org.apache.paimon.data;
 import org.apache.paimon.data.serializer.InternalRowSerializer;
 import org.apache.paimon.memory.MemorySegment;
 import org.apache.paimon.types.DataTypes;
+import org.apache.paimon.types.VectorType;
 
 import org.junit.jupiter.api.Test;
 
@@ -30,6 +31,8 @@ import static org.assertj.core.api.Assertions.assertThat;
 /** Test for {@link NestedRow}s. */
 public class NestedRowTest {
 
+    private static final VectorType VECTOR_TYPE = DataTypes.VECTOR(3, 
DataTypes.FLOAT());
+
     @Test
     public void testNestedRowWithOneSegment() {
         BinaryRow row = getBinaryRow();
@@ -101,6 +104,41 @@ public class NestedRowTest {
         assertThat(nestedRow.isNullAt(3)).isTrue();
     }
 
+    @Test
+    public void testNestedRowWithVector() {
+        float[] values = new float[] {1.5f, -2f, 3.25f};
+        BinaryRow row = 
getBinaryRowWithVector(BinaryVector.fromPrimitiveArray(values));
+
+        // round-trip: the nested row reads the vector slot 
BinaryWriter.writeVector wrote
+        InternalRow nestedRow = row.getRow(1, 2);
+        assertThat(nestedRow.getInt(0)).isEqualTo(7);
+        assertThat(nestedRow.isNullAt(1)).isFalse();
+        assertThat(nestedRow.getVector(1).toFloatArray()).isEqualTo(values);
+
+        // the generic field getter used by format writers dispatches VECTOR 
to getVector
+        InternalRow.FieldGetter getter = 
InternalRow.createFieldGetter(VECTOR_TYPE, 1);
+        InternalVector viaGetter = (InternalVector) 
getter.getFieldOrNull(nestedRow);
+        assertThat(viaGetter).isNotNull();
+        assertThat(viaGetter.toFloatArray()).isEqualTo(values);
+
+        // the nested row may cross a segment boundary
+        MemorySegment[] segments = 
splitBytes(row.getSegments()[0].getHeapMemory(), 3);
+        row.pointTo(segments, 3, row.getSizeInBytes());
+        assertThat(row.getRow(1, 
2).getVector(1).toFloatArray()).isEqualTo(values);
+
+        // a null vector field stays null
+        BinaryRow nullRow = getBinaryRowWithVector(null);
+        assertThat(nullRow.getRow(1, 2).isNullAt(1)).isTrue();
+    }
+
+    private BinaryRow getBinaryRowWithVector(InternalVector vector) {
+        InternalRowSerializer serializer =
+                new InternalRowSerializer(
+                        DataTypes.INT(), DataTypes.ROW(DataTypes.INT(), 
VECTOR_TYPE));
+        // copy: toBinaryRow returns the serializer's reused row
+        return serializer.toBinaryRow(GenericRow.of(1, GenericRow.of(7, 
vector))).copy();
+    }
+
     private BinaryRow getBinaryRow() {
         BinaryRow row = new BinaryRow(1);
         BinaryRowWriter writer = new BinaryRowWriter(row);