This is an automated email from the ASF dual-hosted git repository.
singhpk234 pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/iceberg.git
The following commit(s) were added to refs/heads/main by this push:
new da5ffce9a9 Fix for vectorized builder variant handling (#16087)
da5ffce9a9 is described below
commit da5ffce9a957130b562676680fefa2b8e2ac9d0a
Author: Neelesh Salian <[email protected]>
AuthorDate: Tue May 5 19:55:43 2026 -0700
Fix for vectorized builder variant handling (#16087)
* Fix for vectorized builder variant handling
* Simplify test query and add reg test
* PR comment: add describedAs for keys
* Add merge into test for spark 4.0
* PR comment: Add test for variant not in projection
---
.../arrow/vectorized/VectorizedReaderBuilder.java | 10 +++
.../vectorized/TestVectorizedReaderBuilder.java | 92 ++++++++++++++++++++++
.../iceberg/spark/sql/TestSparkVariantRead.java | 49 ++++++++++++
.../iceberg/spark/sql/TestSparkVariantRead.java | 49 ++++++++++++
4 files changed, 200 insertions(+)
diff --git
a/arrow/src/main/java/org/apache/iceberg/arrow/vectorized/VectorizedReaderBuilder.java
b/arrow/src/main/java/org/apache/iceberg/arrow/vectorized/VectorizedReaderBuilder.java
index 15b55fb48d..3fbd797c26 100644
---
a/arrow/src/main/java/org/apache/iceberg/arrow/vectorized/VectorizedReaderBuilder.java
+++
b/arrow/src/main/java/org/apache/iceberg/arrow/vectorized/VectorizedReaderBuilder.java
@@ -154,6 +154,16 @@ public class VectorizedReaderBuilder extends
TypeWithSchemaVisitor<VectorizedRea
return null;
}
+ @Override
+ public VectorizedReader<?> variant(
+ Types.VariantType iVariant, GroupType variant, VectorizedReader<?>
result) {
+ if (iVariant != null) {
+ throw new UnsupportedOperationException(
+ "Vectorized reads are not supported yet for variant fields");
+ }
+ return null;
+ }
+
@Override
public VectorizedReader<?> primitive(
org.apache.iceberg.types.Type.PrimitiveType expected, PrimitiveType
primitive) {
diff --git
a/arrow/src/test/java/org/apache/iceberg/arrow/vectorized/TestVectorizedReaderBuilder.java
b/arrow/src/test/java/org/apache/iceberg/arrow/vectorized/TestVectorizedReaderBuilder.java
new file mode 100644
index 0000000000..e3d76515bc
--- /dev/null
+++
b/arrow/src/test/java/org/apache/iceberg/arrow/vectorized/TestVectorizedReaderBuilder.java
@@ -0,0 +1,92 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied. See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.iceberg.arrow.vectorized;
+
+import static org.assertj.core.api.Assertions.assertThatNoException;
+import static org.assertj.core.api.Assertions.assertThatThrownBy;
+
+import org.apache.iceberg.Schema;
+import org.apache.iceberg.parquet.TypeWithSchemaVisitor;
+import org.apache.iceberg.relocated.com.google.common.collect.ImmutableMap;
+import org.apache.iceberg.types.Types.IntegerType;
+import org.apache.iceberg.types.Types.NestedField;
+import org.apache.iceberg.types.Types.VariantType;
+import org.apache.iceberg.variants.Variant;
+import org.apache.parquet.schema.LogicalTypeAnnotation;
+import org.apache.parquet.schema.MessageType;
+import org.apache.parquet.schema.PrimitiveType.PrimitiveTypeName;
+import org.apache.parquet.schema.Type;
+import org.apache.parquet.schema.Types;
+import org.junit.jupiter.api.Test;
+
+public class TestVectorizedReaderBuilder {
+
+ @Test
+ public void testVariantNotSupportedInVectorizedReads() {
+ Schema icebergSchema =
+ new Schema(
+ NestedField.required(1, "id", IntegerType.get()),
+ NestedField.optional(2, "data", VariantType.get()));
+
+ MessageType parquetSchema = parquetSchemaWithVariant();
+
+ VectorizedReaderBuilder builder =
+ new VectorizedReaderBuilder(
+ icebergSchema, parquetSchema, false, ImmutableMap.of(), readers ->
null);
+
+ assertThatThrownBy(
+ () -> TypeWithSchemaVisitor.visit(icebergSchema.asStruct(),
parquetSchema, builder))
+ .isInstanceOf(UnsupportedOperationException.class)
+ .hasMessageContaining("Vectorized reads are not supported yet for
variant fields");
+ }
+
+ @Test
+ public void testVariantSkippedWhenNotInProjection() {
+ Schema icebergSchema = new Schema(NestedField.required(1, "id",
IntegerType.get()));
+
+ MessageType parquetSchema = parquetSchemaWithVariant();
+
+ VectorizedReaderBuilder builder =
+ new VectorizedReaderBuilder(
+ icebergSchema, parquetSchema, false, ImmutableMap.of(), readers ->
null);
+
+ assertThatNoException()
+ .describedAs("Variant not in projection should not throw")
+ .isThrownBy(
+ () -> TypeWithSchemaVisitor.visit(icebergSchema.asStruct(),
parquetSchema, builder));
+ }
+
+ private static MessageType parquetSchemaWithVariant() {
+ return Types.buildMessage()
+ .addField(
+ Types.primitive(PrimitiveTypeName.INT32,
Type.Repetition.REQUIRED).id(1).named("id"))
+ .addField(
+ Types.buildGroup(Type.Repetition.OPTIONAL)
+
.as(LogicalTypeAnnotation.variantType(Variant.VARIANT_SPEC_VERSION))
+ .addField(
+ Types.primitive(PrimitiveTypeName.BINARY,
Type.Repetition.REQUIRED)
+ .named("metadata"))
+ .addField(
+ Types.primitive(PrimitiveTypeName.BINARY,
Type.Repetition.REQUIRED)
+ .named("value"))
+ .id(2)
+ .named("data"))
+ .named("table");
+ }
+}
diff --git
a/spark/v4.0/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
b/spark/v4.0/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
index 599bf591e9..2d6e919a91 100644
---
a/spark/v4.0/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
+++
b/spark/v4.0/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
@@ -302,6 +302,55 @@ public class TestSparkVariantRead extends TestBase {
sql("DROP TABLE IF EXISTS %s", mapTable);
}
+ @ParameterizedTest
+ @ValueSource(booleans = {false, true})
+ public void testMergeIntoWithVariant(boolean vectorized) {
+ // Variant columns are not vectorized yet, but MERGE INTO should not crash
regardless of the
+ // vectorization setting. The reader falls back to non-vectorized for
variant columns.
+ String mergeTable = CATALOG + ".default.var_merge";
+ sql("DROP TABLE IF EXISTS %s", mergeTable);
+ sql(
+ "CREATE TABLE %s (id BIGINT, data VARIANT) USING iceberg "
+ + "TBLPROPERTIES ('format-version'='3')",
+ mergeTable);
+ setVectorization(mergeTable, vectorized);
+
+ sql(
+ "INSERT INTO %s VALUES "
+ + "(1, parse_json('{\"name\":\"alice\",\"age\":30}')), "
+ + "(2, parse_json('{\"name\":\"bob\",\"age\":25}'))",
+ mergeTable);
+
+ sql(
+ "MERGE INTO %s AS target "
+ + "USING (SELECT 1 AS id,
parse_json('{\"name\":\"alice\",\"age\":31}') AS data) AS source "
+ + "ON target.id = source.id "
+ + "WHEN MATCHED THEN UPDATE SET target.data = source.data "
+ + "WHEN NOT MATCHED THEN INSERT *",
+ mergeTable);
+
+ List<Row> rows = spark.table(mergeTable).select("id",
"data").orderBy("id").collectAsList();
+
+ assertThat(rows).hasSize(2);
+ assertThat(rows.get(0).getLong(0)).isEqualTo(1L);
+ Variant v1 =
+ new Variant(
+ ((VariantVal) rows.get(0).get(1)).getValue(),
+ ((VariantVal) rows.get(0).get(1)).getMetadata());
+
assertThat(v1.getFieldByKey("name").getString()).describedAs("v1.name").isEqualTo("alice");
+
assertThat(v1.getFieldByKey("age").getLong()).describedAs("v1.age").isEqualTo(31L);
+
+ assertThat(rows.get(1).getLong(0)).isEqualTo(2L);
+ Variant v2 =
+ new Variant(
+ ((VariantVal) rows.get(1).get(1)).getValue(),
+ ((VariantVal) rows.get(1).get(1)).getMetadata());
+
assertThat(v2.getFieldByKey("name").getString()).describedAs("v2.name").isEqualTo("bob");
+
assertThat(v2.getFieldByKey("age").getLong()).describedAs("v2.age").isEqualTo(25L);
+
+ sql("DROP TABLE IF EXISTS %s", mergeTable);
+ }
+
private void setVectorization(boolean on) {
sql(
"ALTER TABLE %s SET TBLPROPERTIES
('read.parquet.vectorization.enabled'='%s')",
diff --git
a/spark/v4.1/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
b/spark/v4.1/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
index 599bf591e9..2d6e919a91 100644
---
a/spark/v4.1/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
+++
b/spark/v4.1/spark/src/test/java/org/apache/iceberg/spark/sql/TestSparkVariantRead.java
@@ -302,6 +302,55 @@ public class TestSparkVariantRead extends TestBase {
sql("DROP TABLE IF EXISTS %s", mapTable);
}
+ @ParameterizedTest
+ @ValueSource(booleans = {false, true})
+ public void testMergeIntoWithVariant(boolean vectorized) {
+ // Variant columns are not vectorized yet, but MERGE INTO should not crash
regardless of the
+ // vectorization setting. The reader falls back to non-vectorized for
variant columns.
+ String mergeTable = CATALOG + ".default.var_merge";
+ sql("DROP TABLE IF EXISTS %s", mergeTable);
+ sql(
+ "CREATE TABLE %s (id BIGINT, data VARIANT) USING iceberg "
+ + "TBLPROPERTIES ('format-version'='3')",
+ mergeTable);
+ setVectorization(mergeTable, vectorized);
+
+ sql(
+ "INSERT INTO %s VALUES "
+ + "(1, parse_json('{\"name\":\"alice\",\"age\":30}')), "
+ + "(2, parse_json('{\"name\":\"bob\",\"age\":25}'))",
+ mergeTable);
+
+ sql(
+ "MERGE INTO %s AS target "
+ + "USING (SELECT 1 AS id,
parse_json('{\"name\":\"alice\",\"age\":31}') AS data) AS source "
+ + "ON target.id = source.id "
+ + "WHEN MATCHED THEN UPDATE SET target.data = source.data "
+ + "WHEN NOT MATCHED THEN INSERT *",
+ mergeTable);
+
+ List<Row> rows = spark.table(mergeTable).select("id",
"data").orderBy("id").collectAsList();
+
+ assertThat(rows).hasSize(2);
+ assertThat(rows.get(0).getLong(0)).isEqualTo(1L);
+ Variant v1 =
+ new Variant(
+ ((VariantVal) rows.get(0).get(1)).getValue(),
+ ((VariantVal) rows.get(0).get(1)).getMetadata());
+
assertThat(v1.getFieldByKey("name").getString()).describedAs("v1.name").isEqualTo("alice");
+
assertThat(v1.getFieldByKey("age").getLong()).describedAs("v1.age").isEqualTo(31L);
+
+ assertThat(rows.get(1).getLong(0)).isEqualTo(2L);
+ Variant v2 =
+ new Variant(
+ ((VariantVal) rows.get(1).get(1)).getValue(),
+ ((VariantVal) rows.get(1).get(1)).getMetadata());
+
assertThat(v2.getFieldByKey("name").getString()).describedAs("v2.name").isEqualTo("bob");
+
assertThat(v2.getFieldByKey("age").getLong()).describedAs("v2.age").isEqualTo(25L);
+
+ sql("DROP TABLE IF EXISTS %s", mergeTable);
+ }
+
private void setVectorization(boolean on) {
sql(
"ALTER TABLE %s SET TBLPROPERTIES
('read.parquet.vectorization.enabled'='%s')",
