问题描述

更新 （5.5及更高版本）

Confluent版本5.5通过PrimitiveAvroSerde（cf. https://github.com/confluentinc/schema-registry/blob/5.5.x/avro-serde/src/main/java/io/confluent/kafka/streams添加了对原始Avro类型的本地支持/serdes/avro/PrimitiveAvroSerde.java）

原始答案 （版本5.4及更高版本）：

这是一个已知问题。原始Avro类型无法与Confluent的AvroSerdes配合使用，因为Serdes只能与GenericAvroRecord和一起SpecificAvroRecord使用。

比较https://github.com/confluentinc/schema-registry/tree/master/avro-serde/src/main/java/io/confluent/kafka/streams/serdes/avro。

因此，基于KafkaAvroSerializer和构建您自己的SerdeKafkaAvroDeserializer是正确的方法。为了能够将其作为默认Serde传递到配置中，您不能使用，Serdes.serdeFrom因为由于泛型类型擦除而导致类型信息丢失。

但是，您可以实现自己的扩展Serde接口的类，并将自定义类传递给config：

public class MySerde extends Serde<Integer> {
    // use KafkaAvroSerializer and KafkaAvroDeserializer and cast `Object` to `Integer`
}

config.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG, MySerde.class);

感谢@Matthias J. Sax的提示，我想发布解决方案周围的工作。请免费进行改进。

import java.util.Collections;
import java.util.Map;

import org.apache.kafka.common.serialization.Deserializer;
import org.apache.kafka.common.serialization.Serde;
import org.apache.kafka.common.serialization.Serdes;
import org.apache.kafka.common.serialization.Serializer;

import io.confluent.kafka.schemaregistry.client.SchemaRegistryClient;
import io.confluent.kafka.serializers.KafkaAvroDeserializer;
import io.confluent.kafka.serializers.KafkaAvroSerializer;

public class GenericPrimitiveAvroSerDe<T> implements Serde<T> {

    private final Serde<Object> inner;

    /**
     * Constructor used by Kafka Streams.
     */
    public GenericPrimitiveAvroSerDe() {
        inner = Serdes.serdeFrom(new KafkaAvroSerializer(), new KafkaAvroDeserializer());
    }

    public GenericPrimitiveAvroSerDe(SchemaRegistryClient client) {
        this(client, Collections.emptyMap());
    }

    public GenericPrimitiveAvroSerDe(SchemaRegistryClient client, Map<String, ?> props) {
        inner = Serdes.serdeFrom(new KafkaAvroSerializer(client), new KafkaAvroDeserializer(client, props));
    }

    @Override
    public void configure(final Map<String, ?> serdeConfig, final boolean isSerdeForRecordKeys) {
        inner.serializer().configure(serdeConfig, isSerdeForRecordKeys);
        inner.deserializer().configure(serdeConfig, isSerdeForRecordKeys);
    }

    @Override
    public void close() {
        // TODO Auto-generated method stub
        inner.serializer().close();
        inner.deserializer().close();

    }

    @SuppressWarnings("unchecked")
    @Override
    public Serializer<T> serializer() {
        // TODO Auto-generated method stub
        Object obj = inner.serializer();
        return (Serializer<T>) obj;

    }

    @SuppressWarnings("unchecked")
    @Override
    public Deserializer<T> deserializer() {
        // TODO Auto-generated method stub
        Object obj = inner.deserializer();
        return (Deserializer<T>) obj;

    }

}

用作默认流配置：

props.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG,GenericPrimitiveAvroSerDe.class);
            props.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG,GenericPrimitiveAvroSerDe.class);

覆盖默认值：

final Map<String, String> serdeConfig = Collections.singletonMap("schema.registry.url",
                                                                        "http://localhost:8081");
       final GenericPrimitiveAvroSerDe<String> keyGenericAvroSerde = new GenericPrimitiveAvroSerDe<String>();
       keyGenericAvroSerde.configure(serdeConfig, true); // `true` for record keys
       final GenericPrimitiveAvroSerDe<Long> valueGenericAvroSerde = new GenericPrimitiveAvroSerDe<Long>();
       valueGenericAvroSerde.configure(serdeConfig, false); // `false` for record values

解决方法

key-schema: "int"
value-schema:{
"type": "record","name": "User","fields": [
    {"name": "firstname","type": "string"},{"name": "lastname","type": "string"}
]}

Map<String,String> serdeConfig = Collections.singletonMap("schema.registry.url",schemaRegistryUrl);

Serde<User> userSerde = new SpecificAvroSerde<>();
userSerde.configure(serdeConfig,false);

KafkaAvroDeserializer keyDeserializer = new KafkaAvroDeserializer()
KafkaAvroSerializer keySerializer = new KafkaAvroSerializer();
keyDeserializer.configure(serdeConfig,true); 
keySerializer.configure(serdeConfig,true);
Serde<Integer> keySerde = (Serde<Integer>)(Serde)Serdes.serdeFrom(keySerializer,keyDeserializer);

streamsConfiguration.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG,SpecificAvroSerde.class);