AHeise commented on a change in pull request #17501: URL: https://github.com/apache/flink/pull/17501#discussion_r769957779
########## File path: flink-formats/flink-parquet/src/main/java/org/apache/flink/formats/parquet/avro/AvroParquetRecordFormat.java ########## @@ -0,0 +1,236 @@ +/* + * 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.formats.parquet.avro; + +import org.apache.flink.annotation.VisibleForTesting; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.configuration.Configuration; +import org.apache.flink.connector.file.src.reader.StreamFormat; +import org.apache.flink.connector.file.src.util.CheckpointedPosition; +import org.apache.flink.core.fs.FSDataInputStream; + +import org.apache.avro.generic.GenericData; +import org.apache.avro.generic.GenericRecord; +import org.apache.avro.reflect.ReflectData; +import org.apache.avro.specific.SpecificData; +import org.apache.parquet.avro.AvroParquetReader; +import org.apache.parquet.hadoop.ParquetReader; +import org.apache.parquet.io.DelegatingSeekableInputStream; +import org.apache.parquet.io.InputFile; +import org.apache.parquet.io.SeekableInputStream; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import javax.annotation.Nullable; + +import java.io.IOException; + +import static org.apache.flink.util.Preconditions.checkArgument; + +/** */ +public class AvroParquetRecordFormat<E> implements StreamFormat<E> { + + private static final long serialVersionUID = 1L; + + static final Logger LOG = LoggerFactory.getLogger(AvroParquetRecordFormat.class); + + private final TypeInformation<E> type; + + AvroParquetRecordFormat(TypeInformation<E> type) { + this.type = type; + } + + /** + * Creates a new reader to read avro {@link GenericRecord} from Parquet input stream. + * + * <p>Several wrapper classes haven be created to Flink abstraction become compatible with the + * parquet abstraction. Please refer to the inner classes {@link AvroParquetRecordReader}, + * {@link ParquetInputFile}, {@link FSDataInputStreamAdapter} for details. + */ + @Override + public Reader<E> createReader( + Configuration config, FSDataInputStream stream, long fileLen, long splitEnd) + throws IOException { + + // current version does not support splitting. + checkNotSplit(fileLen, splitEnd); + + return new AvroParquetRecordReader<E>( + AvroParquetReader.<E>builder(new ParquetInputFile(stream, fileLen)) + .withDataModel(getDataModel()) + .build()); + } + + /** + * Restores the reader from a checkpointed position. It is in fact identical since only {@link + * CheckpointedPosition#NO_OFFSET} as the {@code restoredOffset} is support. + */ + @Override + public Reader<E> restoreReader( + Configuration config, + FSDataInputStream stream, + long restoredOffset, + long fileLen, + long splitEnd) + throws IOException { + + // current version does not support splitting. + checkNotSplit(fileLen, splitEnd); + + checkArgument( + restoredOffset == CheckpointedPosition.NO_OFFSET, + "The restoredOffset should always be NO_OFFSET"); + + return createReader(config, stream, fileLen, splitEnd); + } + + @VisibleForTesting + GenericData getDataModel() { + Class<E> typeClass = getProducedType().getTypeClass(); + if (org.apache.avro.specific.SpecificRecordBase.class.isAssignableFrom(typeClass)) { + return SpecificData.get(); + } else if (org.apache.avro.generic.GenericRecord.class.isAssignableFrom(typeClass)) { + return GenericData.get(); + } else { + return ReflectData.get(); + } + } Review comment: Er, tbh I forgot to check whether `GenericData` is serializable. While `readObject` is the most canonical solution, it's also the least developer-friendly version (there are surprisingly many devs that don't know how that works correctly). So from your options, I'd prefer the enum. I'd add two more options: 3) revert to the state before my comment (duplicated logic). 4) Use `SerializableSupplier` (I'm turning it into `PublicEvolving` into another PR so that this will also work if we outsource formats). So you have ``` AvroParquetRecordFormat(TypeInformation<E> type, SerializableSupplier<GenericData> modelSupplier) { this.modelSupplier = modelSupplier; ... } ``` and ``` return new AvroParquetRecordFormat<>(new AvroTypeInfo<>(typeClass), () -> GenericData.get()); ``` -- 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]
