cshuo commented on code in PR #18406: URL: https://github.com/apache/hudi/pull/18406#discussion_r3007191787
########## hudi-flink-datasource/hudi-flink/src/main/java/org/apache/hudi/source/reader/function/AbstractSplitReaderFunction.java: ########## @@ -0,0 +1,103 @@ +/* + * 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.source.reader.function; + +import org.apache.flink.configuration.Configuration; +import org.apache.flink.table.data.RowData; +import org.apache.hudi.common.util.collection.ClosableIterator; +import org.apache.hudi.config.HoodieWriteConfig; +import org.apache.hudi.configuration.HadoopConfigurations; +import org.apache.hudi.source.ExpressionPredicates; +import org.apache.hudi.table.format.InternalSchemaManager; +import org.apache.hudi.util.FlinkWriteClients; + +import java.util.List; + +/** + * Abstract implementation of SplitReaderFunction with shared limit push-down and + * Hadoop configuration support. + * + * <p>Subclasses call {@link #limitIterator(ClosableIterator, long)} to cap the number of + * records returned per split. Use {@link #NO_LIMIT} as the sentinel when no limit is set. + */ +public abstract class AbstractSplitReaderFunction implements SplitReaderFunction<RowData> { + + /** Sentinel value indicating that no row limit has been pushed down. */ + public static final long NO_LIMIT = -1L; + + protected final long limit; + protected final Configuration conf; + protected final InternalSchemaManager internalSchemaManager; + protected final List<ExpressionPredicates.Predicate> predicates; + protected final boolean emitDelete; + private transient HoodieWriteConfig writeConfig; + private transient org.apache.hadoop.conf.Configuration hadoopConf; + + public AbstractSplitReaderFunction( + Configuration conf, + List<ExpressionPredicates.Predicate> predicates, + InternalSchemaManager internalSchemaManager, + long limit, + boolean emitDelete) { + this.conf = conf; + this.predicates = predicates; + this.internalSchemaManager = internalSchemaManager; + this.limit = limit; + this.emitDelete = emitDelete; + } + + /** + * Wraps a {@link ClosableIterator} to stop after {@code limit} records. + */ + protected ClosableIterator<RowData> limitIterator(ClosableIterator<RowData> iterator, long limit) { + return new ClosableIterator<RowData>() { + private long currentReadCount = 0; + + @Override + public boolean hasNext() { + return currentReadCount < limit && iterator.hasNext(); + } + + @Override + public RowData next() { + currentReadCount++; + return iterator.next(); + } + + @Override + public void close() { + iterator.close(); + } + }; + } + + protected HoodieWriteConfig getWriteConfig() { + if (writeConfig == null) { + writeConfig = FlinkWriteClients.getHoodieClientConfig(conf); + } + return writeConfig; + } + + protected org.apache.hadoop.conf.Configuration getHadoopConf() { Review Comment: ditto. ########## hudi-flink-datasource/hudi-flink/src/main/java/org/apache/hudi/source/reader/function/HoodieSplitReaderFunction.java: ########## @@ -71,28 +62,41 @@ public HoodieSplitReaderFunction( String mergeType, List<ExpressionPredicates.Predicate> predicates, boolean emitDelete) { + this(configuration, tableSchema, requiredSchema, internalSchemaManager, mergeType, predicates, emitDelete, NO_LIMIT); + } + public HoodieSplitReaderFunction( + Configuration configuration, + HoodieSchema tableSchema, + HoodieSchema requiredSchema, + InternalSchemaManager internalSchemaManager, + String mergeType, + List<ExpressionPredicates.Predicate> predicates, + boolean emitDelete, + long limit) { + super(configuration, predicates, internalSchemaManager, limit, emitDelete); ValidationUtils.checkArgument(tableSchema != null, "tableSchema can't be null"); ValidationUtils.checkArgument(requiredSchema != null, "requiredSchema can't be null"); ValidationUtils.checkArgument(internalSchemaManager != null, "internalSchemaManager can't be null"); this.tableSchema = tableSchema; this.requiredSchema = requiredSchema; - this.internalSchemaManager = internalSchemaManager; - this.configuration = configuration; - this.writeConfig = FlinkWriteClients.getHoodieClientConfig(configuration); - this.predicates = predicates; this.mergeType = mergeType; - this.emitDelete = emitDelete; } @Override public RecordsWithSplitIds<HoodieRecordWithPosition<RowData>> read(HoodieSourceSplit split) { final String splitId = split.splitId(); - HoodieTableMetaClient metaClient = StreamerUtil.metaClientForReader(configuration, getHadoopConf()); + HoodieTableMetaClient metaClient = StreamerUtil.metaClientForReader(conf, getHadoopConf()); try { + if (fileGroupReader != null) { + fileGroupReader.close(); + } this.fileGroupReader = createFileGroupReader(split, metaClient); - final ClosableIterator<RowData> recordIterator = fileGroupReader.getClosableIterator(); + ClosableIterator<RowData> recordIterator = fileGroupReader.getClosableIterator(); + if (limit > 0) { + recordIterator = limitIterator(recordIterator, limit); Review Comment: The older MOR/COW input formats keep a cumulative counter at reader subtask level, while the new source applies the limit to each split iterator, so this maybe a regression espescially when there are many splits to read. Can we create something like record limiter in `HoodieSourceSplitReader`, and applied to the `SplitReaderFunction`, so that the limiting is applied at reader subtask level, and avoid reading all the splits. ########## hudi-flink-datasource/hudi-flink/src/main/java/org/apache/hudi/source/reader/function/AbstractSplitReaderFunction.java: ########## @@ -0,0 +1,103 @@ +/* + * 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.source.reader.function; + +import org.apache.flink.configuration.Configuration; +import org.apache.flink.table.data.RowData; +import org.apache.hudi.common.util.collection.ClosableIterator; +import org.apache.hudi.config.HoodieWriteConfig; +import org.apache.hudi.configuration.HadoopConfigurations; +import org.apache.hudi.source.ExpressionPredicates; +import org.apache.hudi.table.format.InternalSchemaManager; +import org.apache.hudi.util.FlinkWriteClients; + +import java.util.List; + +/** + * Abstract implementation of SplitReaderFunction with shared limit push-down and + * Hadoop configuration support. + * + * <p>Subclasses call {@link #limitIterator(ClosableIterator, long)} to cap the number of + * records returned per split. Use {@link #NO_LIMIT} as the sentinel when no limit is set. + */ +public abstract class AbstractSplitReaderFunction implements SplitReaderFunction<RowData> { + + /** Sentinel value indicating that no row limit has been pushed down. */ + public static final long NO_LIMIT = -1L; + + protected final long limit; + protected final Configuration conf; + protected final InternalSchemaManager internalSchemaManager; + protected final List<ExpressionPredicates.Predicate> predicates; + protected final boolean emitDelete; + private transient HoodieWriteConfig writeConfig; + private transient org.apache.hadoop.conf.Configuration hadoopConf; + + public AbstractSplitReaderFunction( + Configuration conf, + List<ExpressionPredicates.Predicate> predicates, + InternalSchemaManager internalSchemaManager, + long limit, + boolean emitDelete) { + this.conf = conf; + this.predicates = predicates; + this.internalSchemaManager = internalSchemaManager; + this.limit = limit; + this.emitDelete = emitDelete; + } + + /** + * Wraps a {@link ClosableIterator} to stop after {@code limit} records. + */ + protected ClosableIterator<RowData> limitIterator(ClosableIterator<RowData> iterator, long limit) { + return new ClosableIterator<RowData>() { + private long currentReadCount = 0; + + @Override + public boolean hasNext() { + return currentReadCount < limit && iterator.hasNext(); + } + + @Override + public RowData next() { + currentReadCount++; + return iterator.next(); + } + + @Override + public void close() { + iterator.close(); + } + }; + } + + protected HoodieWriteConfig getWriteConfig() { Review Comment: nit: we can use `Lazy.lazily(() -> FlinkWriteClients.getHoodieClientConfig(conf))`. -- 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]
