cloud-fan commented on code in PR #53276: URL: https://github.com/apache/spark/pull/53276#discussion_r2601637199
########## sql/catalyst/src/main/java/org/apache/spark/sql/connector/read/VariantExtraction.java: ########## @@ -0,0 +1,62 @@ +/* + * 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.spark.sql.connector.read; + +import org.apache.spark.annotation.Experimental; +import org.apache.spark.sql.types.DataType; +import org.apache.spark.sql.types.Metadata; + +/** + * Variant extraction information that describes a single field extraction from a variant column. + * <p> + * This interface captures the information needed by data sources to optimize reading variant + * columns. Each instance represents one field extraction operation (e.g., from variant_get or + * try_variant_get). + * <p> + * For example, if a query contains `variant_get(v, '$.a', 'int')`, this would be represented + * as a VariantExtraction with columnName=["v"], path="$.a", and expectedDataType=IntegerType. + * + * @since 4.1.0 + */ +@Experimental +public interface VariantExtraction { + /** + * Returns the path to the variant column. For top-level variant columns, this is a single + * element array containing the column name. For nested variant columns within structs, + * this is an array representing the path (e.g., ["struct1", "struct2", "variantCol"]). Review Comment: ```suggestion * this is an array representing the path (e.g., ["structCol", "innerStruct", "variantCol"]). ``` ########## sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/v2/V2ScanRelationPushDown.scala: ########## @@ -318,6 +321,139 @@ object V2ScanRelationPushDown extends Rule[LogicalPlan] with PredicateHelper { case agg: Aggregate => rewriteAggregate(agg) } + def pushDownVariants(plan: LogicalPlan): LogicalPlan = plan.transformDown { + case p@PhysicalOperation(projectList, filters, sHolder @ ScanBuilderHolder(_, _, + builder: SupportsPushDownVariantExtractions)) + if conf.getConf(org.apache.spark.sql.internal.SQLConf.PUSH_VARIANT_INTO_SCAN) => + pushVariantExtractions(p, projectList, filters, sHolder, builder) + } + + /** + * Converts an ordinal path to a field name path. + * + * @param structType The top-level struct type + * @param ordinals The ordinal path (e.g., [1, 1] for nested.field) + * @return The field name path (e.g., ["nested", "field"]) + */ + private def getColumnName(structType: StructType, ordinals: Seq[Int]): Seq[String] = { + ordinals match { + case Seq() => + // Base case: no more ordinals + Seq.empty + case ordinal +: rest => + // Get the field at this ordinal + val field = structType.fields(ordinal) + if (rest.isEmpty) { + // Last ordinal in the path + Seq(field.name) + } else { + // Recurse into nested struct + field.dataType match { + case nestedStruct: StructType => + field.name +: getColumnName(nestedStruct, rest) + case _ => + throw new IllegalArgumentException( Review Comment: we should throw `SparkException.internal` ########## sql/catalyst/src/main/java/org/apache/spark/sql/connector/read/VariantExtractionImpl.java: ########## @@ -0,0 +1,95 @@ +/* + * 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.spark.sql.connector.read; + +import java.io.Serializable; +import java.util.Arrays; +import java.util.Objects; + +import org.apache.spark.annotation.Experimental; +import org.apache.spark.sql.types.DataType; +import org.apache.spark.sql.types.Metadata; + +/** + * Default implementation of {@link VariantExtraction}. + * + * @since 4.1.0 + */ +@Experimental +public final class VariantExtractionImpl implements VariantExtraction, Serializable { Review Comment: Does it need to be public API? ########## sql/catalyst/src/main/java/org/apache/spark/sql/connector/read/SupportsPushDownVariantExtractions.java: ########## @@ -0,0 +1,63 @@ +/* + * 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.spark.sql.connector.read; + +import org.apache.spark.annotation.Experimental; + +/** + * A mix-in interface for {@link ScanBuilder}. Data sources can implement this interface to + * support pushing down variant field extraction operations to the data source. + * <p> + * When variant columns are accessed with specific field extractions (e.g., variant_get, + * try_variant_get), the optimizer can push these extractions down to the data source. + * The data source can then read only the required fields from variant columns, reducing + * I/O and improving performance. + * <p> + * Each {@link VariantExtraction} in the input array represents one field extraction operation. + * Data sources should examine each extraction and determine which ones can be handled efficiently. + * The return value is a boolean array of the same length, where each element indicates whether + * the corresponding extraction was accepted. + * + * @since 4.1.0 + */ +@Experimental +public interface SupportsPushDownVariantExtractions extends ScanBuilder { + + /** + * Pushes down variant field extractions to the data source. + * <p> + * Each element in the input array represents one field extraction operation from a variant + * column. Data sources should examine each extraction and determine whether it can be Review Comment: Does Spark do any reconciliation? e.g. `variant_get(v, '$.a', 'int')` and `variant_get(v, '$.b', 'int')` will be two extractions, but how about `variant_get(v, '$.a', 'struct<b int>')` and `variant_get(v, '$.a.b', 'int')`? Are they two extractions as well? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
