linliu-code commented on code in PR #18224: URL: https://github.com/apache/hudi/pull/18224#discussion_r2872771179
########## hudi-utilities/src/main/java/org/apache/hudi/utilities/sources/JsonKinesisSource.java: ########## @@ -0,0 +1,284 @@ +/* + * 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.sources; + +import org.apache.hudi.common.config.TypedProperties; +import org.apache.hudi.common.util.Option; +import org.apache.hudi.utilities.config.KinesisSourceConfig; +import org.apache.hudi.utilities.ingestion.HoodieIngestionMetrics; +import org.apache.hudi.utilities.schema.SchemaProvider; +import org.apache.hudi.utilities.sources.helpers.KinesisOffsetGen; +import org.apache.hudi.utilities.sources.helpers.KinesisReadConfig; +import org.apache.hudi.utilities.streamer.DefaultStreamContext; +import org.apache.hudi.utilities.streamer.StreamContext; + +import com.fasterxml.jackson.databind.ObjectMapper; +import com.fasterxml.jackson.databind.node.ObjectNode; +import lombok.AllArgsConstructor; +import lombok.Getter; +import lombok.extern.slf4j.Slf4j; +import org.apache.spark.api.java.JavaRDD; +import org.apache.spark.api.java.JavaSparkContext; +import org.apache.spark.sql.SparkSession; +import software.amazon.awssdk.services.kinesis.KinesisClient; +import software.amazon.awssdk.services.kinesis.model.Record; + +import java.io.Serializable; +import java.util.ArrayList; +import java.util.HashMap; +import java.util.HashSet; +import java.util.List; +import java.util.Map; +import java.util.Set; + +import static org.apache.hudi.common.util.ConfigUtils.getBooleanWithAltKeys; +import static org.apache.hudi.common.util.ConfigUtils.getIntWithAltKeys; +import static org.apache.hudi.common.util.ConfigUtils.getLongWithAltKeys; +import static org.apache.hudi.common.util.ConfigUtils.getStringWithAltKeys; +import static org.apache.hudi.utilities.sources.helpers.KinesisOffsetGen.LOCALSTACK_END_SEQ_SENTINEL; + +/** + * Source to read JSON data from AWS Kinesis Data Streams using Spark. + */ +@Slf4j +public class JsonKinesisSource extends KinesisSource<JavaRDD<String>> { + + private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper(); + + /** Result from reading a single shard in a partition. */ + @AllArgsConstructor + @Getter + private static class ShardFetchResult implements Serializable { + private final List<String> records; + private final String shardId; + private final Option<String> lastSequenceNumber; + private final boolean reachedEndOfShard; + } + + /** Metadata-only summary for checkpoint; avoids bringing records to driver. */ + @AllArgsConstructor + @Getter + private static class ShardFetchSummary implements Serializable { + private final String shardId; + private final Option<String> lastSequenceNumber; + private final int recordCount; + private final boolean reachedEndOfShard; + } + + /** Persisted fetch RDD - must be unpersisted in releaseResources to avoid memory leak. */ + private transient org.apache.spark.api.java.JavaRDD<ShardFetchResult> persistedFetchRdd; + /** Record count from fetch, avoids redundant batch.count() Spark job. */ + private long lastRecordCount; + /** Shard IDs where the executor observed nextShardIterator==null (end-of-shard reached). */ + protected Set<String> shardsReachedEnd; + + public JsonKinesisSource(TypedProperties properties, JavaSparkContext sparkContext, SparkSession sparkSession, + SchemaProvider schemaProvider, HoodieIngestionMetrics metrics) { + this(properties, sparkContext, sparkSession, metrics, + new DefaultStreamContext(schemaProvider, Option.empty())); + } + + public JsonKinesisSource(TypedProperties properties, JavaSparkContext sparkContext, SparkSession sparkSession, + HoodieIngestionMetrics metrics, StreamContext streamContext) { + super(properties, sparkContext, sparkSession, SourceType.JSON, metrics, + new DefaultStreamContext(streamContext.getSchemaProvider(), streamContext.getSourceProfileSupplier())); + this.offsetGen = new KinesisOffsetGen(props); + } + + @Override + protected JavaRDD<String> toBatch(KinesisOffsetGen.KinesisShardRange[] shardRanges) { + KinesisReadConfig readConfig = new KinesisReadConfig( + offsetGen.getStreamName(), + offsetGen.getRegion(), + offsetGen.getEndpointUrl().orElse(null), + getStringWithAltKeys(props, KinesisSourceConfig.KINESIS_ACCESS_KEY, null), + getStringWithAltKeys(props, KinesisSourceConfig.KINESIS_SECRET_KEY, null), + offsetGen.getStartingPosition(), + shouldAddOffsets, + getBooleanWithAltKeys(props, KinesisSourceConfig.KINESIS_ENABLE_DEAGGREGATION), + getIntWithAltKeys(props, KinesisSourceConfig.KINESIS_GET_RECORDS_MAX_RECORDS), + getLongWithAltKeys(props, KinesisSourceConfig.KINESIS_GET_RECORDS_INTERVAL_MS), + // Evenly set the max events per shard. + shardRanges.length > 0 ? Math.max(1, getLongWithAltKeys(props, KinesisSourceConfig.MAX_EVENTS_FROM_KINESIS_SOURCE) / shardRanges.length) : Long.MAX_VALUE); + + JavaRDD<ShardFetchResult> fetchRdd = sparkContext.parallelize( + java.util.Arrays.asList(shardRanges), shardRanges.length) + .mapPartitions(shardRangeIt -> { + List<ShardFetchResult> results = new ArrayList<>(); + try (KinesisClient client = createKinesisClientFromConfig(readConfig)) { + while (shardRangeIt.hasNext()) { + KinesisOffsetGen.KinesisShardRange range = shardRangeIt.next(); + KinesisOffsetGen.ShardReadResult readResult = KinesisOffsetGen.readShardRecords( + client, readConfig.getStreamName(), range, readConfig.getStartingPosition(), + readConfig.getMaxRecordsPerRequest(), readConfig.getIntervalMs(), readConfig.getMaxRecordsPerShard(), + readConfig.isEnableDeaggregation()); + + List<String> recordStrings = new ArrayList<>(); + for (Record r : readResult.getRecords()) { + String json = recordToJsonStatic(r, range.getShardId(), readConfig.isShouldAddOffsets()); + if (json != null) { + recordStrings.add(json); + } + } + results.add(new ShardFetchResult(recordStrings, range.getShardId(), + readResult.getLastSequenceNumber(), readResult.isReachedEndOfShard())); + } + } + return results.iterator(); + }); + + // Cache so we can both get records and checkpoint from the same RDD + fetchRdd.persist(org.apache.spark.storage.StorageLevel.MEMORY_AND_DISK()); + persistedFetchRdd = fetchRdd; + + // RDD for the shard data. + JavaRDD<String> recordRdd = fetchRdd.flatMap(r -> r.getRecords().iterator()); + // Apply minimum partitions for downstream parallelism (similar to Kafka source) + long minPartitions = getLongWithAltKeys(props, KinesisSourceConfig.KINESIS_SOURCE_MIN_PARTITIONS); + if (minPartitions > 0 && minPartitions > shardRanges.length) { + int targetPartitions = (int) minPartitions; + log.info("Repartitioning from {} shards to {} partitions (minPartitions={})", + shardRanges.length, targetPartitions, minPartitions); + recordRdd = recordRdd.repartition(targetPartitions); + } + + // Information for checkpoint generation. + List<ShardFetchSummary> summaries = fetchRdd + .map(r -> new ShardFetchSummary(r.getShardId(), r.getLastSequenceNumber(), r.getRecords().size(), + r.isReachedEndOfShard())) + .collect(); + lastCheckpointData = buildCheckpointFromSummaries(summaries); + Set<String> reached = new HashSet<>(); + for (ShardFetchSummary s : summaries) { + if (s.isReachedEndOfShard()) { + reached.add(s.getShardId()); + } + } + shardsReachedEnd = reached; + lastRecordCount = summaries.stream().mapToLong(ShardFetchSummary::getRecordCount).sum(); + + return recordRdd; + } + + private static KinesisClient createKinesisClientFromConfig(KinesisReadConfig config) { + software.amazon.awssdk.services.kinesis.KinesisClientBuilder builder = + KinesisClient.builder().region(software.amazon.awssdk.regions.Region.of(config.getRegion())); + if (config.getEndpointUrl() != null && !config.getEndpointUrl().isEmpty()) { + builder = builder.endpointOverride(java.net.URI.create(config.getEndpointUrl())); + } + if (config.getAccessKey() != null && !config.getAccessKey().isEmpty() + && config.getSecretKey() != null && !config.getSecretKey().isEmpty()) { + builder = builder.credentialsProvider( + software.amazon.awssdk.auth.credentials.StaticCredentialsProvider.create( + software.amazon.awssdk.auth.credentials.AwsBasicCredentials.create( + config.getAccessKey(), config.getSecretKey()))); + } + return builder.build(); + } + + private static String recordToJsonStatic(Record record, String shardId, boolean shouldAddOffsets) { + String dataStr = record.data().asUtf8String(); + // Pure empty or null records in Kinesis is not meaningful. + if (dataStr == null || dataStr.trim().isEmpty()) { + return null; + } + if (shouldAddOffsets) { Review Comment: done. -- 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]
