JingsongLi commented on a change in pull request #13724: URL: https://github.com/apache/flink/pull/13724#discussion_r517821825
########## File path: flink-formats/flink-orc/src/main/java/org/apache/flink/orc/OrcColumnarRowFileInputFormat.java ########## @@ -0,0 +1,171 @@ +/* + * 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.flink.orc; + +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.connector.file.src.FileSourceSplit; +import org.apache.flink.connector.file.src.util.Pool; +import org.apache.flink.orc.shim.OrcShim; +import org.apache.flink.orc.vector.ColumnBatchFactory; +import org.apache.flink.orc.vector.OrcVectorizedBatchWrapper; +import org.apache.flink.table.data.ColumnarRowData; +import org.apache.flink.table.data.RowData; +import org.apache.flink.table.data.vector.ColumnVector; +import org.apache.flink.table.data.vector.VectorizedColumnBatch; +import org.apache.flink.table.filesystem.ColumnarRowIterator; +import org.apache.flink.table.filesystem.PartitionFieldExtractor; +import org.apache.flink.table.runtime.typeutils.InternalTypeInfo; +import org.apache.flink.table.types.logical.LogicalType; +import org.apache.flink.table.types.logical.RowType; + +import org.apache.hadoop.conf.Configuration; +import org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch; +import org.apache.orc.TypeDescription; + +import java.util.Arrays; +import java.util.List; +import java.util.stream.Collectors; + +import static org.apache.flink.orc.OrcSplitReaderUtil.convertToOrcTypeWithPart; +import static org.apache.flink.orc.OrcSplitReaderUtil.getNonPartNames; +import static org.apache.flink.orc.OrcSplitReaderUtil.getSelectedOrcFields; +import static org.apache.flink.orc.vector.AbstractOrcColumnVector.createFlinkVector; +import static org.apache.flink.orc.vector.AbstractOrcColumnVector.createFlinkVectorFromConstant; + +/** + * An ORC reader that produces a stream of {@link ColumnarRowData} records. + * + * <p>This class can add extra fields through {@link ColumnBatchFactory}, for example, + * add partition fields, which can be extracted from path. Therefore, the {@link #getProducedType()} + * may be different and types of extra fields need to be added. + */ +public class OrcColumnarRowFileInputFormat<BatchT, SplitT extends FileSourceSplit> extends + AbstractOrcFileInputFormat<RowData, BatchT, SplitT> { + + private static final long serialVersionUID = 1L; + + private final ColumnBatchFactory<BatchT, SplitT> batchFactory; + private final RowType projectedOutputType; + + public OrcColumnarRowFileInputFormat( + final OrcShim<BatchT> shim, + final Configuration hadoopConfig, + final TypeDescription schema, + final int[] selectedFields, + final List<OrcFilters.Predicate> conjunctPredicates, + final int batchSize, + final ColumnBatchFactory<BatchT, SplitT> batchFactory, + final RowType projectedOutputType) { + super(shim, hadoopConfig, schema, selectedFields, conjunctPredicates, batchSize); + this.batchFactory = batchFactory; + this.projectedOutputType = projectedOutputType; + } + + @Override + public OrcReaderBatch<RowData, BatchT> createReaderBatch( + final SplitT split, + final OrcVectorizedBatchWrapper<BatchT> orcBatch, + final Pool.Recycler<OrcReaderBatch<RowData, BatchT>> recycler, + final int batchSize) { + + final VectorizedColumnBatch flinkColumnBatch = batchFactory.create(split, orcBatch.getBatch()); + return new VectorizedColumnReaderBatch<>(orcBatch, flinkColumnBatch, recycler); + } + + @Override + public TypeInformation<RowData> getProducedType() { + return InternalTypeInfo.of(projectedOutputType); + } + + // ------------------------------------------------------------------------ + + /** + * One batch of ORC columnar vectors and Flink column vectors. + */ + private static final class VectorizedColumnReaderBatch<BatchT> extends OrcReaderBatch<RowData, BatchT> { + + private final VectorizedColumnBatch flinkColumnBatch; + private final ColumnarRowIterator result; + + VectorizedColumnReaderBatch( + final OrcVectorizedBatchWrapper<BatchT> orcBatch, + final VectorizedColumnBatch flinkColumnBatch, + final Pool.Recycler<OrcReaderBatch<RowData, BatchT>> recycler) { + super(orcBatch, recycler); + this.flinkColumnBatch = flinkColumnBatch; + this.result = new ColumnarRowIterator(new ColumnarRowData(flinkColumnBatch), this::recycle); + } + + @Override + public RecordIterator<RowData> convertAndGetIterator( + final OrcVectorizedBatchWrapper<BatchT> orcBatch, + final long startingOffset) { + // no copying from the ORC column vectors to the Flink columns vectors necessary, + // because they point to the same data arrays internally design + int batchSize = orcBatch.size(); + flinkColumnBatch.setNumRows(batchSize); + result.set(batchSize, startingOffset, 0); + return result; + } + } + + /** + * Create a partitioned {@link OrcColumnarRowFileInputFormat}, the partition columns can be + * generated by split. + */ + public static <SplitT extends FileSourceSplit> OrcColumnarRowFileInputFormat<VectorizedRowBatch, SplitT> createPartitionedFormat( + OrcShim<VectorizedRowBatch> shim, + Configuration hadoopConfig, + RowType tableType, Review comment: We can not, because orc needs full schema to reader, but parquet does not need. ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected]
