vinishjail97 commented on code in PR #935: URL: https://github.com/apache/incubator-xtable/pull/935#discussion_r4225388138
########## xtable-core/src/main/java/org/apache/xtable/index/HudiBackedIcebergSecondaryIndex.java: ########## @@ -0,0 +1,237 @@ +/* + * 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.xtable.index; + +import static org.apache.hudi.metadata.HoodieTableMetadataUtil.PARTITION_NAME_SECONDARY_INDEX_PREFIX; +import static org.apache.spark.sql.functions.col; + +import java.util.ArrayList; +import java.util.Collections; +import java.util.List; +import java.util.Properties; + +import lombok.extern.log4j.Log4j2; + +import org.apache.hadoop.conf.Configuration; +import org.apache.spark.api.java.JavaRDD; +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Encoders; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.RowFactory; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.catalyst.CatalystTypeConverters; +import org.apache.spark.sql.catalyst.InternalRow; +import org.apache.spark.sql.types.DataTypes; +import org.apache.spark.sql.types.StructType; + +import org.apache.hudi.client.common.HoodieSparkEngineContext; +import org.apache.hudi.common.config.HoodieMetadataConfig; +import org.apache.hudi.common.data.HoodiePairData; +import org.apache.hudi.common.table.HoodieTableMetaClient; +import org.apache.hudi.common.table.HoodieTableVersion; +import org.apache.hudi.data.HoodieJavaPairRDD; +import org.apache.hudi.data.HoodieJavaRDD; +import org.apache.hudi.metadata.HoodieBackedTableMetadata; +import org.apache.hudi.metadata.HoodieTableMetadataUtil; +import org.apache.hudi.storage.HoodieStorage; +import org.apache.hudi.storage.hadoop.HoodieHadoopStorage; + +import org.apache.iceberg.PartitionSpec; +import org.apache.iceberg.Partitioning; +import org.apache.iceberg.Table; +import org.apache.iceberg.types.Types; + +import scala.Function1; +import scala.Tuple2; + +import org.apache.xtable.catalog.TableFormatUtils; +import org.apache.xtable.conversion.ConversionConfig; +import org.apache.xtable.conversion.ConversionController; +import org.apache.xtable.conversion.SourceTable; +import org.apache.xtable.conversion.TargetTable; +import org.apache.xtable.hudi.HudiTargetConfig; +import org.apache.xtable.iceberg.IcebergConversionSourceProvider; +import org.apache.xtable.model.storage.TableFormat; + +/** + * Secondary index for an Iceberg table backed by the Hudi metadata table. The index is built by + * syncing the Iceberg table to Hudi with the record index and a secondary index enabled; the Hudi + * metadata lives under {@code <data location>/.hoodie/metadata}. Because the Iceberg data files + * carry no Hudi record keys, the record key of a row is {@code <file path relative to the data + * location>_<row position>}, which a lookup resolves back to a file path and row position. + */ +@Log4j2 +public class HudiBackedIcebergSecondaryIndex implements Index<Table> { + private final String tableLocation; + private final String dataPath; + private final Properties targetTableProperties; + private final SparkSession sparkSession; + private final JavaSparkContext javaSparkContext; + private final HoodieSparkEngineContext engineContext; + + /** + * @param icebergTable the table to index + * @param sparkSession the Spark session used to build and query the index + * @param targetTableProperties additional {@link HudiTargetConfig} properties for the sync, for + * example the record index file group counts + */ + public HudiBackedIcebergSecondaryIndex( + Table icebergTable, SparkSession sparkSession, Properties targetTableProperties) { + this.tableLocation = icebergTable.location(); + this.dataPath = + TableFormatUtils.getTableDataLocation( + TableFormat.ICEBERG, tableLocation, icebergTable.properties()); + this.targetTableProperties = targetTableProperties; + this.sparkSession = sparkSession; + this.javaSparkContext = JavaSparkContext.fromSparkContext(sparkSession.sparkContext()); + this.engineContext = new HoodieSparkEngineContext(javaSparkContext); + } + + @Override + public boolean doesIndexExist(String columnName) { + return HoodieTableMetadataUtil.metadataPartitionExists( + dataPath, engineContext, PARTITION_NAME_SECONDARY_INDEX_PREFIX + columnName); + } + + @Override + public void syncIndex(Table icebergTable, String columnName) { + Configuration configuration = javaSparkContext.hadoopConfiguration(); + IcebergConversionSourceProvider sourceProvider = new IcebergConversionSourceProvider(); + sourceProvider.init(configuration); + new ConversionController(configuration) + .sync(getConversionConfig(icebergTable, columnName), sourceProvider); Review Comment: Fixed in dee82e40. `syncIndex` now throws when the Hudi sync result is missing or not `SUCCESS`, and the IT covers a failed initial sync and a failed update that keeps the existing index. ########## xtable-core/src/main/java/org/apache/xtable/index/IcebergPartitionConverter.java: ########## @@ -0,0 +1,159 @@ +/* + * 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.xtable.index; + +import java.math.BigDecimal; +import java.nio.ByteBuffer; +import java.util.List; + +import lombok.AccessLevel; +import lombok.NoArgsConstructor; + +import org.apache.avro.generic.GenericData; +import org.apache.avro.util.Utf8; +import org.apache.spark.sql.catalyst.InternalRow; +import org.apache.spark.sql.catalyst.expressions.GenericInternalRow; +import org.apache.spark.sql.types.DataType; +import org.apache.spark.sql.types.DataTypes; +import org.apache.spark.sql.types.Decimal; +import org.apache.spark.sql.types.StructField; +import org.apache.spark.unsafe.types.UTF8String; + +import org.apache.iceberg.DataFiles; +import org.apache.iceberg.PartitionSpec; +import org.apache.iceberg.StructLike; +import org.apache.iceberg.types.Type; +import org.apache.iceberg.types.Types; +import org.apache.iceberg.util.ByteBuffers; + +/** + * Converts Iceberg partition values to the Spark {@link InternalRow} representation Iceberg's own + * Spark reader exposes through the {@code _partition} metadata column. + */ +@NoArgsConstructor(access = AccessLevel.PRIVATE) +public final class IcebergPartitionConverter { + + /** + * Parses a partition path (for example {@code year=2024/month=01}) with the given spec and + * converts the values to an {@link InternalRow}. + * + * @return the partition row, or null when the table is unpartitioned + */ + public static InternalRow convertPartitionToInternalRow( + String partitionPath, Types.StructType partitionType, PartitionSpec spec) { + if (partitionPath == null || partitionPath.isEmpty() || partitionType.fields().isEmpty()) { + return null; + } + return convertStructLikeToInternalRow(DataFiles.data(spec, partitionPath), partitionType); Review Comment: Fixed in dee82e40 by removing `_partition` from the lookup result. The Hudi index only holds the directory text of each file, so the typed values cannot be recovered from it; a caller that needs `_partition` or `_spec_id` can read the returned files through Iceberg. ########## xtable-core/src/main/java/org/apache/xtable/index/HudiBackedIcebergSecondaryIndex.java: ########## @@ -0,0 +1,237 @@ +/* + * 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.xtable.index; + +import static org.apache.hudi.metadata.HoodieTableMetadataUtil.PARTITION_NAME_SECONDARY_INDEX_PREFIX; +import static org.apache.spark.sql.functions.col; + +import java.util.ArrayList; +import java.util.Collections; +import java.util.List; +import java.util.Properties; + +import lombok.extern.log4j.Log4j2; + +import org.apache.hadoop.conf.Configuration; +import org.apache.spark.api.java.JavaRDD; +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Encoders; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.RowFactory; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.catalyst.CatalystTypeConverters; +import org.apache.spark.sql.catalyst.InternalRow; +import org.apache.spark.sql.types.DataTypes; +import org.apache.spark.sql.types.StructType; + +import org.apache.hudi.client.common.HoodieSparkEngineContext; +import org.apache.hudi.common.config.HoodieMetadataConfig; +import org.apache.hudi.common.data.HoodiePairData; +import org.apache.hudi.common.table.HoodieTableMetaClient; +import org.apache.hudi.common.table.HoodieTableVersion; +import org.apache.hudi.data.HoodieJavaPairRDD; +import org.apache.hudi.data.HoodieJavaRDD; +import org.apache.hudi.metadata.HoodieBackedTableMetadata; +import org.apache.hudi.metadata.HoodieTableMetadataUtil; +import org.apache.hudi.storage.HoodieStorage; +import org.apache.hudi.storage.hadoop.HoodieHadoopStorage; + +import org.apache.iceberg.PartitionSpec; +import org.apache.iceberg.Partitioning; +import org.apache.iceberg.Table; +import org.apache.iceberg.types.Types; + +import scala.Function1; +import scala.Tuple2; + +import org.apache.xtable.catalog.TableFormatUtils; +import org.apache.xtable.conversion.ConversionConfig; +import org.apache.xtable.conversion.ConversionController; +import org.apache.xtable.conversion.SourceTable; +import org.apache.xtable.conversion.TargetTable; +import org.apache.xtable.hudi.HudiTargetConfig; +import org.apache.xtable.iceberg.IcebergConversionSourceProvider; +import org.apache.xtable.model.storage.TableFormat; + +/** + * Secondary index for an Iceberg table backed by the Hudi metadata table. The index is built by + * syncing the Iceberg table to Hudi with the record index and a secondary index enabled; the Hudi + * metadata lives under {@code <data location>/.hoodie/metadata}. Because the Iceberg data files + * carry no Hudi record keys, the record key of a row is {@code <file path relative to the data + * location>_<row position>}, which a lookup resolves back to a file path and row position. + */ +@Log4j2 +public class HudiBackedIcebergSecondaryIndex implements Index<Table> { + private final String tableLocation; + private final String dataPath; + private final Properties targetTableProperties; + private final SparkSession sparkSession; + private final JavaSparkContext javaSparkContext; + private final HoodieSparkEngineContext engineContext; + + /** + * @param icebergTable the table to index + * @param sparkSession the Spark session used to build and query the index + * @param targetTableProperties additional {@link HudiTargetConfig} properties for the sync, for + * example the record index file group counts + */ + public HudiBackedIcebergSecondaryIndex( + Table icebergTable, SparkSession sparkSession, Properties targetTableProperties) { + this.tableLocation = icebergTable.location(); + this.dataPath = + TableFormatUtils.getTableDataLocation( + TableFormat.ICEBERG, tableLocation, icebergTable.properties()); + this.targetTableProperties = targetTableProperties; + this.sparkSession = sparkSession; + this.javaSparkContext = JavaSparkContext.fromSparkContext(sparkSession.sparkContext()); + this.engineContext = new HoodieSparkEngineContext(javaSparkContext); + } + + @Override + public boolean doesIndexExist(String columnName) { + return HoodieTableMetadataUtil.metadataPartitionExists( + dataPath, engineContext, PARTITION_NAME_SECONDARY_INDEX_PREFIX + columnName); + } + + @Override + public void syncIndex(Table icebergTable, String columnName) { + Configuration configuration = javaSparkContext.hadoopConfiguration(); + IcebergConversionSourceProvider sourceProvider = new IcebergConversionSourceProvider(); + sourceProvider.init(configuration); + new ConversionController(configuration) + .sync(getConversionConfig(icebergTable, columnName), sourceProvider); + log.info("Synced secondary index for column {} of table {}", columnName, tableLocation); + } + + private ConversionConfig getConversionConfig(Table icebergTable, String columnName) { + SourceTable sourceTable = + SourceTable.builder() + .name(icebergTable.name()) + .basePath(tableLocation) + .dataPath(dataPath) + .formatName(TableFormat.ICEBERG) + .build(); + Properties mergedProperties = new Properties(); + mergedProperties.putAll(targetTableProperties); + mergedProperties.setProperty(HudiTargetConfig.SECONDARY_INDEX_COLUMN, columnName); + mergedProperties.setProperty( + HudiTargetConfig.EXECUTION_ENGINE, HudiTargetConfig.EXECUTION_ENGINE_SPARK); + // the secondary index is only available from Hudi table version 8 onwards + mergedProperties.setProperty( + HudiTargetConfig.HUDI_TABLE_VERSION, String.valueOf(HoodieTableVersion.NINE.versionCode())); + TargetTable targetTable = + TargetTable.builder() + .name(icebergTable.name()) + .basePath(dataPath) + .formatName(TableFormat.HUDI) + .additionalProperties(mergedProperties) + .build(); + return ConversionConfig.builder() + .sourceTable(sourceTable) + .targetTables(Collections.singletonList(targetTable)) + .build(); + } + + @Override + public Dataset<Row> lookup(Table icebergTable, Dataset<Row> keys, String columnName) { + Types.StructType partitionType = Partitioning.partitionType(icebergTable); + PartitionSpec spec = icebergTable.spec(); + StructType partitionSparkType = IcebergPartitionConverter.toSparkType(partitionType); + StructType resultSchema = + new StructType() + .add(columnName, DataTypes.StringType, false) + .add(FILE_COLUMN, DataTypes.StringType, false) + .add(POSITION_COLUMN, DataTypes.LongType, false) + .add(PARTITION_COLUMN, partitionSparkType, true); + // the secondary index stores the values of the indexed column as strings + JavaRDD<String> secondaryKeys = + keys.where(col(columnName).isNotNull()) + .select(col(columnName).cast(DataTypes.StringType)) + .as(Encoders.STRING()) + .toJavaRDD(); + HoodieStorage storage = + new HoodieHadoopStorage(dataPath, javaSparkContext.hadoopConfiguration()); + HoodieTableMetaClient metaClient = + HoodieTableMetaClient.builder().setStorage(storage).setBasePath(dataPath).build(); + HoodieMetadataConfig metadataConfig = + HoodieMetadataConfig.newBuilder().enable(true).withSecondaryIndexEnabled(true).build(); + try (HoodieBackedTableMetadata tableMetadata = + (HoodieBackedTableMetadata) + metaClient + .getTableFormat() + .getMetadataFactory() + .create(engineContext, storage, metadataConfig, dataPath)) { + HoodiePairData<String, String> recordKeysBySecondaryKey = + tableMetadata.readSecondaryIndexDataTableRecordKeysWithKeys( + HoodieJavaRDD.of(secondaryKeys), PARTITION_NAME_SECONDARY_INDEX_PREFIX + columnName); + // capture the field so the closure does not serialize the index instance + String dataPath = this.dataPath; + JavaRDD<Row> lookupResults = + HoodieJavaPairRDD.getJavaPairRDD(recordKeysBySecondaryKey) Review Comment: Fixed in dee82e40. A list backed `HoodiePairData` is now collected and parallelized, and the IT covers empty and null-only key DataFrames. ########## xtable-core/src/main/java/org/apache/xtable/index/HudiBackedIcebergSecondaryIndex.java: ########## @@ -0,0 +1,237 @@ +/* + * 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.xtable.index; + +import static org.apache.hudi.metadata.HoodieTableMetadataUtil.PARTITION_NAME_SECONDARY_INDEX_PREFIX; +import static org.apache.spark.sql.functions.col; + +import java.util.ArrayList; +import java.util.Collections; +import java.util.List; +import java.util.Properties; + +import lombok.extern.log4j.Log4j2; + +import org.apache.hadoop.conf.Configuration; +import org.apache.spark.api.java.JavaRDD; +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Encoders; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.RowFactory; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.catalyst.CatalystTypeConverters; +import org.apache.spark.sql.catalyst.InternalRow; +import org.apache.spark.sql.types.DataTypes; +import org.apache.spark.sql.types.StructType; + +import org.apache.hudi.client.common.HoodieSparkEngineContext; +import org.apache.hudi.common.config.HoodieMetadataConfig; +import org.apache.hudi.common.data.HoodiePairData; +import org.apache.hudi.common.table.HoodieTableMetaClient; +import org.apache.hudi.common.table.HoodieTableVersion; +import org.apache.hudi.data.HoodieJavaPairRDD; +import org.apache.hudi.data.HoodieJavaRDD; +import org.apache.hudi.metadata.HoodieBackedTableMetadata; +import org.apache.hudi.metadata.HoodieTableMetadataUtil; +import org.apache.hudi.storage.HoodieStorage; +import org.apache.hudi.storage.hadoop.HoodieHadoopStorage; + +import org.apache.iceberg.PartitionSpec; +import org.apache.iceberg.Partitioning; +import org.apache.iceberg.Table; +import org.apache.iceberg.types.Types; + +import scala.Function1; +import scala.Tuple2; + +import org.apache.xtable.catalog.TableFormatUtils; +import org.apache.xtable.conversion.ConversionConfig; +import org.apache.xtable.conversion.ConversionController; +import org.apache.xtable.conversion.SourceTable; +import org.apache.xtable.conversion.TargetTable; +import org.apache.xtable.hudi.HudiTargetConfig; +import org.apache.xtable.iceberg.IcebergConversionSourceProvider; +import org.apache.xtable.model.storage.TableFormat; + +/** + * Secondary index for an Iceberg table backed by the Hudi metadata table. The index is built by + * syncing the Iceberg table to Hudi with the record index and a secondary index enabled; the Hudi + * metadata lives under {@code <data location>/.hoodie/metadata}. Because the Iceberg data files + * carry no Hudi record keys, the record key of a row is {@code <file path relative to the data + * location>_<row position>}, which a lookup resolves back to a file path and row position. + */ +@Log4j2 +public class HudiBackedIcebergSecondaryIndex implements Index<Table> { + private final String tableLocation; + private final String dataPath; + private final Properties targetTableProperties; + private final SparkSession sparkSession; + private final JavaSparkContext javaSparkContext; + private final HoodieSparkEngineContext engineContext; + + /** + * @param icebergTable the table to index + * @param sparkSession the Spark session used to build and query the index + * @param targetTableProperties additional {@link HudiTargetConfig} properties for the sync, for + * example the record index file group counts + */ + public HudiBackedIcebergSecondaryIndex( + Table icebergTable, SparkSession sparkSession, Properties targetTableProperties) { + this.tableLocation = icebergTable.location(); + this.dataPath = + TableFormatUtils.getTableDataLocation( + TableFormat.ICEBERG, tableLocation, icebergTable.properties()); + this.targetTableProperties = targetTableProperties; + this.sparkSession = sparkSession; + this.javaSparkContext = JavaSparkContext.fromSparkContext(sparkSession.sparkContext()); + this.engineContext = new HoodieSparkEngineContext(javaSparkContext); + } + + @Override + public boolean doesIndexExist(String columnName) { + return HoodieTableMetadataUtil.metadataPartitionExists( + dataPath, engineContext, PARTITION_NAME_SECONDARY_INDEX_PREFIX + columnName); + } + + @Override + public void syncIndex(Table icebergTable, String columnName) { + Configuration configuration = javaSparkContext.hadoopConfiguration(); + IcebergConversionSourceProvider sourceProvider = new IcebergConversionSourceProvider(); + sourceProvider.init(configuration); + new ConversionController(configuration) + .sync(getConversionConfig(icebergTable, columnName), sourceProvider); + log.info("Synced secondary index for column {} of table {}", columnName, tableLocation); + } + + private ConversionConfig getConversionConfig(Table icebergTable, String columnName) { + SourceTable sourceTable = + SourceTable.builder() + .name(icebergTable.name()) + .basePath(tableLocation) + .dataPath(dataPath) + .formatName(TableFormat.ICEBERG) + .build(); Review Comment: Fixed in dee82e40. The index constructor now takes a `SourceTable`, so a catalog table is loaded through its catalog config instead of `HadoopTables`. The IT adds a test catalog that keeps no version hint, and the sync fails with `NoSuchTableException` without the catalog config. -- 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]
