Is it possible to deserialize Avro message(consuming message from Kafka) without giving Reader schema in ConfluentRegistryAvroDeserializationSchema

2
votes

I am using Kafka Connector in Apache Flink for access to streams served by Confluent Kafka.

Apart from schema registry url ConfluentRegistryAvroDeserializationSchema.forGeneric(...) expecting 'reader' schema. Instead of providing read schema I want to use same writer's schema(lookup in registry) for reading the message too because Consumer will not have latest schema.

FlinkKafkaConsumer010<GenericRecord> myConsumer =
        new FlinkKafkaConsumer010<>("topic-name", ConfluentRegistryAvroDeserializationSchema.forGeneric(<reader schema goes here>, "http://host:port"), properties);
myConsumer.setStartFromLatest();

https://ci.apache.org/projects/flink/flink-docs-release-1.9/dev/connectors/kafka.html "Using these deserialization schema record will be read with the schema that was retrieved from Schema Registry and transformed to a statically provided"

Since I do not want to keep schema definition at consumer side how do I deserialize Avro message from Kafka using writer's schema?

Appreciate your help!

1
apache-kafkaapache-flinkconfluent-schema-registry
The default behavior is to parse incoming messages using the writer schemas and then converting to the reader schema. Isn't it what you want? – 0x26res
I do not want it to convert, just use writer schema for deserializing the message. – Sparkle8
I think you want to wrap/adapt a KafkaAvroDeserializer into a KeyedDeserializationSchema and make sure you set specific.avro.reader to false, and this should give you the raw GenericRecord as they were emitted by the producer. – 0x26res

1 Answers

3
votes

I don't think it is possible to use directly ConfluentRegistryAvroDeserializationSchema.forGeneric. It is intended to be used with a reader schema and they have preconditions checking for this.

You have to implement your own. Two import things:

  • Set specific.avro.reader to false (other wise you'll get specific records)
  • The KafkaAvroDeserializer has to be lazily initialized (because it isn't serializable it self, as it holds a reference to the schema registry client)
import io.confluent.kafka.schemaregistry.client.CachedSchemaRegistryClient;
import io.confluent.kafka.schemaregistry.client.SchemaRegistryClient;
import io.confluent.kafka.serializers.AbstractKafkaAvroSerDeConfig;
import io.confluent.kafka.serializers.KafkaAvroDeserializer;
import io.confluent.kafka.serializers.KafkaAvroDeserializerConfig;
import java.util.HashMap;
import java.util.Map;
import org.apache.avro.generic.GenericRecord;
import org.apache.flink.api.common.typeinfo.TypeInformation;
import org.apache.flink.api.java.typeutils.TypeExtractor;
import org.apache.flink.streaming.util.serialization.KeyedDeserializationSchema;

public class KafkaGenericAvroDeserializationSchema
    implements KeyedDeserializationSchema<GenericRecord> {

  private final String registryUrl;
  private transient KafkaAvroDeserializer inner;

  public KafkaGenericAvroDeserializationSchema(String registryUrl) {
    this.registryUrl = registryUrl;
  }

  @Override
  public GenericRecord deserialize(
      byte[] messageKey, byte[] message, String topic, int partition, long offset) {
    checkInitialized();
    return (GenericRecord) inner.deserialize(topic, message);
  }

  @Override
  public boolean isEndOfStream(GenericRecord nextElement) {
    return false;
  }

  @Override
  public TypeInformation<GenericRecord> getProducedType() {
    return TypeExtractor.getForClass(GenericRecord.class);
  }

  private void checkInitialized() {
    if (inner == null) {
      Map<String, Object> props = new HashMap<>();
      props.put(AbstractKafkaAvroSerDeConfig.SCHEMA_REGISTRY_URL_CONFIG, registryUrl);
      props.put(KafkaAvroDeserializerConfig.SPECIFIC_AVRO_READER_CONFIG, false);
      SchemaRegistryClient client =
          new CachedSchemaRegistryClient(
              registryUrl, AbstractKafkaAvroSerDeConfig.MAX_SCHEMAS_PER_SUBJECT_DEFAULT);
      inner = new KafkaAvroDeserializer(client, props);
    }
  }
}


env.addSource(
  new FlinkKafkaConsumer<>(
    topic, 
    new KafkaGenericAvroDeserializationSchema(schemaReigstryUrl), 
    kafkaProperties));