rahil-c commented on code in PR #19110: URL: https://github.com/apache/hudi/pull/19110#discussion_r3508653247
########## hudi-utilities/src/main/java/org/apache/hudi/utilities/transform/debezium/AbstractDebeziumTransformer.java: ########## @@ -0,0 +1,242 @@ +/* + * 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.hudi.utilities.transform.debezium; + +import org.apache.hudi.common.config.TypedProperties; +import org.apache.hudi.common.model.debezium.DebeziumConstants; +import org.apache.hudi.common.util.ConfigUtils; +import org.apache.hudi.common.util.Option; +import org.apache.hudi.utilities.config.DebeziumTransformerConfig; +import org.apache.hudi.utilities.transform.Transformer; + +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Column; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.functions; +import org.apache.spark.sql.types.StructField; +import org.apache.spark.sql.types.StructType; + +import java.util.ArrayList; +import java.util.Arrays; +import java.util.HashSet; +import java.util.List; +import java.util.Set; +import java.util.function.Function; +import java.util.stream.Collectors; + +import static org.apache.hudi.config.HoodieErrorTableConfig.ERROR_TABLE_ENABLED; +import static org.apache.hudi.utilities.streamer.BaseErrorTableWriter.ERROR_TABLE_CURRUPT_RECORD_COL_NAME; + +/** + * Base {@link Transformer} that flattens a Debezium change-event envelope into a Hudi row. + * + * <p>A Debezium change event is a nested record of the form + * {@code {op, ts_ms, before:{...}, after:{...}, source:{...}}}. This transformer: + * <ul> + * <li>selects the {@code before} image for deletes and the {@code after} image otherwise, + * and explodes it to the row's top level;</li> + * <li>surfaces the common Debezium metadata columns (operation type, processing/origin + * timestamps, shard) along with any database-specific metadata columns supplied by the + * subclass;</li> + * <li>optionally nests the metadata columns under a single {@code _debezium_metadata} struct + * (see {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS});</li> + * <li>optionally preserves the error-table corrupt-record column when the error table is + * enabled;</li> + * <li>applies an optional database-specific post-processing step (e.g. ordering/sequence + * columns, LSN defaulting);</li> + * <li>normalizes column nullability (see + * {@link DebeziumTransformerConfig#SCHEMA_AS_NULLABLE}).</li> + * </ul> + * + * <p>The flattened column names are defined in {@link DebeziumConstants}; the matching + * {@code DebeziumAvroPayload} implementations rely on these names for merge/ordering semantics. + * + * <p>Subclasses configure the database-specific behavior purely through the constructor; there is + * no abstract method to implement. + */ +public class AbstractDebeziumTransformer implements Transformer { + + public static final String DEBEZIUM_METADATA_FIELD = "_debezium_metadata"; + private static final String DATA_FIELD = "__data"; + + private static final List<Column> DEFAULT_ROOT_LEVEL_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_OP_FIELD).alias(DebeziumConstants.FLATTENED_OP_COL_NAME)); + + private static final List<Column> DEFAULT_NESTED_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_TS_MS_FIELD).alias(DebeziumConstants.UPSTREAM_PROCESSING_TS_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_NAME_FIELD).alias(DebeziumConstants.FLATTENED_SHARD_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_TS_MS_FIELD).alias(DebeziumConstants.FLATTENED_TS_COL_NAME)); + + private final List<Column> typeSpecificMetadataColumns; + private final Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption; + private final boolean nestedFieldsEnabledByDefault; + + protected AbstractDebeziumTransformer( + List<Column> typeSpecificMetadataColumns, + Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption) { + this(typeSpecificMetadataColumns, postProcessingOption, false); + } + + /** + * @param typeSpecificMetadataColumns database-specific metadata columns (already aliased to their + * flattened output names). + * @param postProcessingOption optional post-flatten transformation applied to the result. + * @param nestedFieldsEnabledByDefault default used for + * {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS} when + * the property is not set explicitly. Lets a subclass (e.g. + * Postgres) opt into nested metadata by default. + */ + protected AbstractDebeziumTransformer( + List<Column> typeSpecificMetadataColumns, + Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption, + boolean nestedFieldsEnabledByDefault) { + this.typeSpecificMetadataColumns = typeSpecificMetadataColumns; + this.postProcessingOption = postProcessingOption; + this.nestedFieldsEnabledByDefault = nestedFieldsEnabledByDefault; + } + + @Override + public Dataset<Row> apply(JavaSparkContext javaSparkContext, SparkSession sparkSession, Dataset<Row> rowDataset, TypedProperties props) { + if (rowDataset.columns().length == 0) { + return rowDataset; + } + // Pick selective debezium meta fields: pick the row values from before field for delete record + // and row values from after field for insert or update records. + rowDataset = rowDataset + .withColumn(DATA_FIELD, + functions.when(new Column(DebeziumConstants.INCOMING_OP_FIELD).equalTo(DebeziumConstants.DELETE_OP), + new Column(DebeziumConstants.INCOMING_BEFORE_FIELD)) + .otherwise(new Column(DebeziumConstants.INCOMING_AFTER_FIELD))) + .drop(DebeziumConstants.INCOMING_AFTER_FIELD, DebeziumConstants.INCOMING_BEFORE_FIELD); + + List<Column> allColumns = new ArrayList<>(); + boolean enableNestedFields = isNestedFieldsEnabled(props); + + // When nested fields are enabled, only _change_operation_type and root-level metadata columns should be at root level + if (enableNestedFields) { + Column lsnColumn = null; + List<Column> otherMetadata = new ArrayList<>(); + + // Extract LSN column to root level, keep other metadata nested + for (Column col : typeSpecificMetadataColumns) { + String colStr = col.toString(); + if (colStr.contains(DebeziumConstants.FLATTENED_LSN_COL_NAME)) { + lsnColumn = col; + } else { + otherMetadata.add(col); + } + } + + List<Column> nestedMetadataFields = new ArrayList<>(); + nestedMetadataFields.addAll(DEFAULT_NESTED_METADATA_COLUMNS); + nestedMetadataFields.addAll(otherMetadata); + + // Only add schema field if it exists in the source struct (not all databases have this field) + if (hasSchemaField(rowDataset)) { + nestedMetadataFields.add(new Column(DebeziumConstants.INCOMING_SOURCE_SCHEMA_FIELD).alias(DebeziumConstants.FLATTENED_SCHEMA_NAME)); + } + + rowDataset = rowDataset.withColumn(DEBEZIUM_METADATA_FIELD, + functions.struct(nestedMetadataFields.toArray(new Column[]{}))); + allColumns.add(new Column(DEBEZIUM_METADATA_FIELD)); + + allColumns.addAll(DEFAULT_ROOT_LEVEL_METADATA_COLUMNS); + if (lsnColumn != null) { + // Add LSN column to root level + allColumns.add(lsnColumn); + } + } else { + // When nested fields are disabled, all metadata fields are at root level + allColumns.addAll(DEFAULT_ROOT_LEVEL_METADATA_COLUMNS); + allColumns.addAll(DEFAULT_NESTED_METADATA_COLUMNS); + allColumns.addAll(typeSpecificMetadataColumns); Review Comment: Not applicable for this PR — this is intentional/pre-existing behavior (the schema column is DB-dependent and only surfaced when nesting is enabled) and out of scope for this change. ########## hudi-utilities/src/main/java/org/apache/hudi/utilities/transform/debezium/MysqlDebeziumTransformer.java: ########## @@ -0,0 +1,82 @@ +/* + * 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.hudi.utilities.transform.debezium; + +import org.apache.hudi.common.model.debezium.DebeziumConstants; +import org.apache.hudi.common.util.Option; + +import org.apache.spark.sql.Column; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.functions; + +import java.util.Arrays; +import java.util.List; + +/** + * {@link AbstractDebeziumTransformer} for MySQL Debezium change events. + * + * <p>Surfaces the MySQL binlog coordinates ({@code file}, {@code pos}, {@code row}) as the flattened + * {@code _event_bin_file}, {@code _event_pos} and {@code _event_row} columns, and derives the + * {@code _event_seq} ordering column as {@code "<binlog-file-suffix>.<pos>"} (e.g. {@code "000001.100"} + * for a binlog file {@code "mysql-bin.000001"} at position {@code 100}). {@code _event_seq} is the + * precombine/ordering field consumed by {@code MySqlDebeziumAvroPayload}. + * + * <p>Metadata is flattened to the root level by default; set + * {@code hoodie.streamer.transformer.debezium.nested.fields.enable=true} to group it under a + * {@code _debezium_metadata} struct instead. + */ +public class MysqlDebeziumTransformer extends AbstractDebeziumTransformer { + + private static final List<Column> MYSQL_METADATA = Arrays.asList( + new Column(DebeziumConstants.INCOMING_SOURCE_FILE_FIELD).alias(DebeziumConstants.FLATTENED_FILE_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_POS_FIELD).alias(DebeziumConstants.FLATTENED_POS_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_ROW_FIELD).alias(DebeziumConstants.FLATTENED_ROW_COL_NAME)); + + public MysqlDebeziumTransformer() { + super(MYSQL_METADATA, Option.of(MysqlDebeziumTransformer::applySeqNo)); + } + + /** + * Builds the {@code _event_seq} ordering column from the binlog file and position. The file column + * holds a name like {@code "mysql-bin.000001"}; only the numeric suffix after the last dot is used, + * yielding a sequence such as {@code "000001.100"}. Handles both the flat and nested metadata + * layouts (reading {@code file}/{@code pos} from the {@code _debezium_metadata} struct when nested). + * + * @param dataset flattened MySQL Debezium dataset. + * @return dataset with the {@code _event_seq} column added. + */ + private static Dataset<Row> applySeqNo(Dataset<Row> dataset) { + boolean isNested = Arrays.asList(dataset.columns()).contains(DEBEZIUM_METADATA_FIELD); + + Column fileCol = isNested + ? dataset.col(DEBEZIUM_METADATA_FIELD + "." + DebeziumConstants.FLATTENED_FILE_COL_NAME) + : dataset.col(DebeziumConstants.FLATTENED_FILE_COL_NAME); + + Column posCol = isNested + ? dataset.col(DEBEZIUM_METADATA_FIELD + "." + DebeziumConstants.FLATTENED_POS_COL_NAME) + : dataset.col(DebeziumConstants.FLATTENED_POS_COL_NAME); + + return dataset.withColumn(DebeziumConstants.ADDED_SEQ_COL_NAME, functions.concat( Review Comment: Not applicable for this PR — this is pre-existing, intentional behavior and out of scope for this change. Open to adding an explicit guard as a follow-up if there's appetite for it. ########## hudi-utilities/src/main/java/org/apache/hudi/utilities/transform/debezium/AbstractDebeziumTransformer.java: ########## @@ -0,0 +1,242 @@ +/* + * 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.hudi.utilities.transform.debezium; + +import org.apache.hudi.common.config.TypedProperties; +import org.apache.hudi.common.model.debezium.DebeziumConstants; +import org.apache.hudi.common.util.ConfigUtils; +import org.apache.hudi.common.util.Option; +import org.apache.hudi.utilities.config.DebeziumTransformerConfig; +import org.apache.hudi.utilities.transform.Transformer; + +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Column; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.functions; +import org.apache.spark.sql.types.StructField; +import org.apache.spark.sql.types.StructType; + +import java.util.ArrayList; +import java.util.Arrays; +import java.util.HashSet; +import java.util.List; +import java.util.Set; +import java.util.function.Function; +import java.util.stream.Collectors; + +import static org.apache.hudi.config.HoodieErrorTableConfig.ERROR_TABLE_ENABLED; +import static org.apache.hudi.utilities.streamer.BaseErrorTableWriter.ERROR_TABLE_CURRUPT_RECORD_COL_NAME; + +/** + * Base {@link Transformer} that flattens a Debezium change-event envelope into a Hudi row. + * + * <p>A Debezium change event is a nested record of the form + * {@code {op, ts_ms, before:{...}, after:{...}, source:{...}}}. This transformer: + * <ul> + * <li>selects the {@code before} image for deletes and the {@code after} image otherwise, + * and explodes it to the row's top level;</li> + * <li>surfaces the common Debezium metadata columns (operation type, processing/origin + * timestamps, shard) along with any database-specific metadata columns supplied by the + * subclass;</li> + * <li>optionally nests the metadata columns under a single {@code _debezium_metadata} struct + * (see {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS});</li> + * <li>optionally preserves the error-table corrupt-record column when the error table is + * enabled;</li> + * <li>applies an optional database-specific post-processing step (e.g. ordering/sequence + * columns, LSN defaulting);</li> + * <li>normalizes column nullability (see + * {@link DebeziumTransformerConfig#SCHEMA_AS_NULLABLE}).</li> + * </ul> + * + * <p>The flattened column names are defined in {@link DebeziumConstants}; the matching + * {@code DebeziumAvroPayload} implementations rely on these names for merge/ordering semantics. + * + * <p>Subclasses configure the database-specific behavior purely through the constructor; there is + * no abstract method to implement. + */ +public class AbstractDebeziumTransformer implements Transformer { + + public static final String DEBEZIUM_METADATA_FIELD = "_debezium_metadata"; + private static final String DATA_FIELD = "__data"; + + private static final List<Column> DEFAULT_ROOT_LEVEL_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_OP_FIELD).alias(DebeziumConstants.FLATTENED_OP_COL_NAME)); + + private static final List<Column> DEFAULT_NESTED_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_TS_MS_FIELD).alias(DebeziumConstants.UPSTREAM_PROCESSING_TS_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_NAME_FIELD).alias(DebeziumConstants.FLATTENED_SHARD_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_TS_MS_FIELD).alias(DebeziumConstants.FLATTENED_TS_COL_NAME)); + + private final List<Column> typeSpecificMetadataColumns; + private final Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption; + private final boolean nestedFieldsEnabledByDefault; + + protected AbstractDebeziumTransformer( + List<Column> typeSpecificMetadataColumns, + Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption) { + this(typeSpecificMetadataColumns, postProcessingOption, false); + } + + /** + * @param typeSpecificMetadataColumns database-specific metadata columns (already aliased to their + * flattened output names). + * @param postProcessingOption optional post-flatten transformation applied to the result. + * @param nestedFieldsEnabledByDefault default used for + * {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS} when + * the property is not set explicitly. Lets a subclass (e.g. + * Postgres) opt into nested metadata by default. + */ + protected AbstractDebeziumTransformer( + List<Column> typeSpecificMetadataColumns, + Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption, + boolean nestedFieldsEnabledByDefault) { + this.typeSpecificMetadataColumns = typeSpecificMetadataColumns; + this.postProcessingOption = postProcessingOption; + this.nestedFieldsEnabledByDefault = nestedFieldsEnabledByDefault; + } + + @Override + public Dataset<Row> apply(JavaSparkContext javaSparkContext, SparkSession sparkSession, Dataset<Row> rowDataset, TypedProperties props) { + if (rowDataset.columns().length == 0) { + return rowDataset; + } + // Pick selective debezium meta fields: pick the row values from before field for delete record + // and row values from after field for insert or update records. + rowDataset = rowDataset + .withColumn(DATA_FIELD, + functions.when(new Column(DebeziumConstants.INCOMING_OP_FIELD).equalTo(DebeziumConstants.DELETE_OP), + new Column(DebeziumConstants.INCOMING_BEFORE_FIELD)) + .otherwise(new Column(DebeziumConstants.INCOMING_AFTER_FIELD))) + .drop(DebeziumConstants.INCOMING_AFTER_FIELD, DebeziumConstants.INCOMING_BEFORE_FIELD); + + List<Column> allColumns = new ArrayList<>(); + boolean enableNestedFields = isNestedFieldsEnabled(props); + + // When nested fields are enabled, only _change_operation_type and root-level metadata columns should be at root level + if (enableNestedFields) { + Column lsnColumn = null; + List<Column> otherMetadata = new ArrayList<>(); + + // Extract LSN column to root level, keep other metadata nested + for (Column col : typeSpecificMetadataColumns) { + String colStr = col.toString(); + if (colStr.contains(DebeziumConstants.FLATTENED_LSN_COL_NAME)) { Review Comment: Not applicable for this PR — this is pre-existing behavior and out of scope for this change. ########## hudi-utilities/src/main/java/org/apache/hudi/utilities/transform/debezium/AbstractDebeziumTransformer.java: ########## @@ -0,0 +1,242 @@ +/* + * 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.hudi.utilities.transform.debezium; + +import org.apache.hudi.common.config.TypedProperties; +import org.apache.hudi.common.model.debezium.DebeziumConstants; +import org.apache.hudi.common.util.ConfigUtils; +import org.apache.hudi.common.util.Option; +import org.apache.hudi.utilities.config.DebeziumTransformerConfig; +import org.apache.hudi.utilities.transform.Transformer; + +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Column; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.functions; +import org.apache.spark.sql.types.StructField; +import org.apache.spark.sql.types.StructType; + +import java.util.ArrayList; +import java.util.Arrays; +import java.util.HashSet; +import java.util.List; +import java.util.Set; +import java.util.function.Function; +import java.util.stream.Collectors; + +import static org.apache.hudi.config.HoodieErrorTableConfig.ERROR_TABLE_ENABLED; +import static org.apache.hudi.utilities.streamer.BaseErrorTableWriter.ERROR_TABLE_CURRUPT_RECORD_COL_NAME; + +/** + * Base {@link Transformer} that flattens a Debezium change-event envelope into a Hudi row. + * + * <p>A Debezium change event is a nested record of the form + * {@code {op, ts_ms, before:{...}, after:{...}, source:{...}}}. This transformer: + * <ul> + * <li>selects the {@code before} image for deletes and the {@code after} image otherwise, + * and explodes it to the row's top level;</li> + * <li>surfaces the common Debezium metadata columns (operation type, processing/origin + * timestamps, shard) along with any database-specific metadata columns supplied by the + * subclass;</li> + * <li>optionally nests the metadata columns under a single {@code _debezium_metadata} struct + * (see {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS});</li> + * <li>optionally preserves the error-table corrupt-record column when the error table is + * enabled;</li> + * <li>applies an optional database-specific post-processing step (e.g. ordering/sequence + * columns, LSN defaulting);</li> + * <li>normalizes column nullability (see + * {@link DebeziumTransformerConfig#SCHEMA_AS_NULLABLE}).</li> + * </ul> + * + * <p>The flattened column names are defined in {@link DebeziumConstants}; the matching + * {@code DebeziumAvroPayload} implementations rely on these names for merge/ordering semantics. + * + * <p>Subclasses configure the database-specific behavior purely through the constructor; there is + * no abstract method to implement. + */ +public class AbstractDebeziumTransformer implements Transformer { + + public static final String DEBEZIUM_METADATA_FIELD = "_debezium_metadata"; + private static final String DATA_FIELD = "__data"; + + private static final List<Column> DEFAULT_ROOT_LEVEL_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_OP_FIELD).alias(DebeziumConstants.FLATTENED_OP_COL_NAME)); + + private static final List<Column> DEFAULT_NESTED_METADATA_COLUMNS = Arrays.asList( Review Comment: Not applicable for this PR — the naming is pre-existing and out of scope for this change. ########## hudi-utilities/src/main/java/org/apache/hudi/utilities/transform/debezium/AbstractDebeziumTransformer.java: ########## @@ -0,0 +1,242 @@ +/* + * 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.hudi.utilities.transform.debezium; + +import org.apache.hudi.common.config.TypedProperties; +import org.apache.hudi.common.model.debezium.DebeziumConstants; +import org.apache.hudi.common.util.ConfigUtils; +import org.apache.hudi.common.util.Option; +import org.apache.hudi.utilities.config.DebeziumTransformerConfig; +import org.apache.hudi.utilities.transform.Transformer; + +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.Column; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.functions; +import org.apache.spark.sql.types.StructField; +import org.apache.spark.sql.types.StructType; + +import java.util.ArrayList; +import java.util.Arrays; +import java.util.HashSet; +import java.util.List; +import java.util.Set; +import java.util.function.Function; +import java.util.stream.Collectors; + +import static org.apache.hudi.config.HoodieErrorTableConfig.ERROR_TABLE_ENABLED; +import static org.apache.hudi.utilities.streamer.BaseErrorTableWriter.ERROR_TABLE_CURRUPT_RECORD_COL_NAME; + +/** + * Base {@link Transformer} that flattens a Debezium change-event envelope into a Hudi row. + * + * <p>A Debezium change event is a nested record of the form + * {@code {op, ts_ms, before:{...}, after:{...}, source:{...}}}. This transformer: + * <ul> + * <li>selects the {@code before} image for deletes and the {@code after} image otherwise, + * and explodes it to the row's top level;</li> + * <li>surfaces the common Debezium metadata columns (operation type, processing/origin + * timestamps, shard) along with any database-specific metadata columns supplied by the + * subclass;</li> + * <li>optionally nests the metadata columns under a single {@code _debezium_metadata} struct + * (see {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS});</li> + * <li>optionally preserves the error-table corrupt-record column when the error table is + * enabled;</li> + * <li>applies an optional database-specific post-processing step (e.g. ordering/sequence + * columns, LSN defaulting);</li> + * <li>normalizes column nullability (see + * {@link DebeziumTransformerConfig#SCHEMA_AS_NULLABLE}).</li> + * </ul> + * + * <p>The flattened column names are defined in {@link DebeziumConstants}; the matching + * {@code DebeziumAvroPayload} implementations rely on these names for merge/ordering semantics. + * + * <p>Subclasses configure the database-specific behavior purely through the constructor; there is + * no abstract method to implement. + */ +public class AbstractDebeziumTransformer implements Transformer { + + public static final String DEBEZIUM_METADATA_FIELD = "_debezium_metadata"; + private static final String DATA_FIELD = "__data"; + + private static final List<Column> DEFAULT_ROOT_LEVEL_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_OP_FIELD).alias(DebeziumConstants.FLATTENED_OP_COL_NAME)); + + private static final List<Column> DEFAULT_NESTED_METADATA_COLUMNS = Arrays.asList( + new Column(DebeziumConstants.INCOMING_TS_MS_FIELD).alias(DebeziumConstants.UPSTREAM_PROCESSING_TS_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_NAME_FIELD).alias(DebeziumConstants.FLATTENED_SHARD_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_TS_MS_FIELD).alias(DebeziumConstants.FLATTENED_TS_COL_NAME)); + + private final List<Column> typeSpecificMetadataColumns; + private final Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption; + private final boolean nestedFieldsEnabledByDefault; + + protected AbstractDebeziumTransformer( + List<Column> typeSpecificMetadataColumns, + Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption) { + this(typeSpecificMetadataColumns, postProcessingOption, false); + } + + /** + * @param typeSpecificMetadataColumns database-specific metadata columns (already aliased to their + * flattened output names). + * @param postProcessingOption optional post-flatten transformation applied to the result. + * @param nestedFieldsEnabledByDefault default used for + * {@link DebeziumTransformerConfig#ENABLE_NESTED_FIELDS} when + * the property is not set explicitly. Lets a subclass (e.g. + * Postgres) opt into nested metadata by default. + */ + protected AbstractDebeziumTransformer( + List<Column> typeSpecificMetadataColumns, + Option<Function<Dataset<Row>, Dataset<Row>>> postProcessingOption, + boolean nestedFieldsEnabledByDefault) { + this.typeSpecificMetadataColumns = typeSpecificMetadataColumns; + this.postProcessingOption = postProcessingOption; + this.nestedFieldsEnabledByDefault = nestedFieldsEnabledByDefault; + } + + @Override + public Dataset<Row> apply(JavaSparkContext javaSparkContext, SparkSession sparkSession, Dataset<Row> rowDataset, TypedProperties props) { + if (rowDataset.columns().length == 0) { + return rowDataset; + } + // Pick selective debezium meta fields: pick the row values from before field for delete record + // and row values from after field for insert or update records. + rowDataset = rowDataset + .withColumn(DATA_FIELD, + functions.when(new Column(DebeziumConstants.INCOMING_OP_FIELD).equalTo(DebeziumConstants.DELETE_OP), + new Column(DebeziumConstants.INCOMING_BEFORE_FIELD)) + .otherwise(new Column(DebeziumConstants.INCOMING_AFTER_FIELD))) + .drop(DebeziumConstants.INCOMING_AFTER_FIELD, DebeziumConstants.INCOMING_BEFORE_FIELD); + + List<Column> allColumns = new ArrayList<>(); + boolean enableNestedFields = isNestedFieldsEnabled(props); + + // When nested fields are enabled, only _change_operation_type and root-level metadata columns should be at root level + if (enableNestedFields) { + Column lsnColumn = null; + List<Column> otherMetadata = new ArrayList<>(); + + // Extract LSN column to root level, keep other metadata nested + for (Column col : typeSpecificMetadataColumns) { + String colStr = col.toString(); + if (colStr.contains(DebeziumConstants.FLATTENED_LSN_COL_NAME)) { + lsnColumn = col; + } else { + otherMetadata.add(col); + } + } + + List<Column> nestedMetadataFields = new ArrayList<>(); + nestedMetadataFields.addAll(DEFAULT_NESTED_METADATA_COLUMNS); + nestedMetadataFields.addAll(otherMetadata); + + // Only add schema field if it exists in the source struct (not all databases have this field) + if (hasSchemaField(rowDataset)) { + nestedMetadataFields.add(new Column(DebeziumConstants.INCOMING_SOURCE_SCHEMA_FIELD).alias(DebeziumConstants.FLATTENED_SCHEMA_NAME)); + } + + rowDataset = rowDataset.withColumn(DEBEZIUM_METADATA_FIELD, + functions.struct(nestedMetadataFields.toArray(new Column[]{}))); + allColumns.add(new Column(DEBEZIUM_METADATA_FIELD)); + + allColumns.addAll(DEFAULT_ROOT_LEVEL_METADATA_COLUMNS); + if (lsnColumn != null) { + // Add LSN column to root level + allColumns.add(lsnColumn); + } + } else { + // When nested fields are disabled, all metadata fields are at root level + allColumns.addAll(DEFAULT_ROOT_LEVEL_METADATA_COLUMNS); + allColumns.addAll(DEFAULT_NESTED_METADATA_COLUMNS); + allColumns.addAll(typeSpecificMetadataColumns); + } + + allColumns.add(new Column(String.format("%s.*", DATA_FIELD))); + + if (ConfigUtils.getBooleanWithAltKeys(props, ERROR_TABLE_ENABLED)) { Review Comment: Not applicable for this PR — this is pre-existing style and out of scope for this change. ########## hudi-utilities/src/main/java/org/apache/hudi/utilities/transform/debezium/PostgresDebeziumTransformer.java: ########## @@ -0,0 +1,82 @@ +/* + * 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.hudi.utilities.transform.debezium; + +import org.apache.hudi.common.model.debezium.DebeziumConstants; +import org.apache.hudi.common.util.Option; + +import org.apache.spark.sql.Column; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; + +import java.util.Arrays; +import java.util.List; + +import static org.apache.spark.sql.functions.expr; +import static org.apache.spark.sql.functions.when; + +/** + * {@link AbstractDebeziumTransformer} for Postgres Debezium change events. + * + * <p>Surfaces the Postgres-specific source metadata ({@code txId}, {@code lsn}, {@code xmin}) as the + * flattened {@code _event_tx_id}, {@code _event_lsn} and {@code _event_xmin} columns. The + * {@code _event_lsn} column is the ordering field used by {@code PostgresDebeziumAvroPayload}. + * + * <p>Post-processing defaults a null {@code _event_lsn} to {@code 0} for snapshot records, since the + * LSN is not populated for rows produced by Debezium incremental snapshots + * (see <a href="https://debezium.io/blog/2021/10/07/incremental-snapshots/";>incremental snapshots</a>). + * + * <p>Nested metadata is enabled by default for Postgres (the {@code _event_lsn} and operation-type + * columns stay at the root level so payload ordering keeps working); set + * {@code hoodie.streamer.transformer.debezium.nested.fields.enable=false} to flatten all metadata. + */ +public class PostgresDebeziumTransformer extends AbstractDebeziumTransformer { + + // Snapshot operation type emitted by Debezium ("read"). + private static final String SNAPSHOT_OP = "r"; + + private static final List<Column> POSTGRES_METADATA = Arrays.asList( + new Column(DebeziumConstants.INCOMING_SOURCE_TXID_FIELD).alias(DebeziumConstants.FLATTENED_TX_ID_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_LSN_FIELD).alias(DebeziumConstants.FLATTENED_LSN_COL_NAME), + new Column(DebeziumConstants.INCOMING_SOURCE_XMIN_FIELD).alias(DebeziumConstants.FLATTENED_XMIN_COL_NAME)); + + public PostgresDebeziumTransformer() { + super(POSTGRES_METADATA, Option.of(PostgresDebeziumTransformer::useDefaultValuesForLsnIfNull), true); + } + + /** + * Defaults a null {@code _event_lsn} to {@code 0} for snapshot records. The LSN is null when a + * table is added via a Debezium incremental snapshot; leaving it null would break LSN-based + * ordering in the payload. + * + * @param dataset flattened Postgres Debezium dataset. + * @return dataset where null {@code _event_lsn} values on snapshot rows are replaced with 0. + */ + private static Dataset<Row> useDefaultValuesForLsnIfNull(Dataset<Row> dataset) { + if (!Arrays.asList(dataset.columns()).contains(DebeziumConstants.FLATTENED_LSN_COL_NAME)) { + return dataset; Review Comment: Not applicable for this PR — pre-existing naming and out of scope for this change. -- 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]
