Spring Data Elasticsearch升级后Set属性报MappingInstantiationException如何解决

问题根因 #

你遇到的报错是Spring Data Elasticsearch 4.x（对应Spring Boot 2.5.x版本）的实体映射模块反序列化限制导致的：当实体类中集合类型属性声明为接口（比如Set、List）时，默认的实体实例化器无法直接实例化接口类型，就会抛出找不到构造函数的异常。

最优解决方案（无需修改任何实体代码，不影响JPA逻辑） #

全局自定义Spring Data Elasticsearch的实体实例化策略，指定集合接口的默认实现类，配置代码如下：

首先新增自定义实例化器实现：

import org.springframework.data.mapping.InstanceCreatorMetadata;
import org.springframework.data.mapping.Parameter;
import org.springframework.data.mapping.PersistentEntity;
import org.springframework.data.mapping.model.EntityInstantiator;
import org.springframework.data.mapping.model.ParameterValueProvider;
import org.springframework.util.ReflectionUtils;

import java.util.*;
import java.util.concurrent.ConcurrentHashMap;

public class CustomCollectionInstantiator implements EntityInstantiator {

    private final Map<Class<?>, Class<?>> COLLECTION_INTERFACE_MAPPINGS = new ConcurrentHashMap<>();

    public CustomCollectionInstantiator() {
        // 配置集合接口对应的默认实现类
        COLLECTION_INTERFACE_MAPPINGS.put(Set.class, HashSet.class);
        COLLECTION_INTERFACE_MAPPINGS.put(List.class, ArrayList.class);
        COLLECTION_INTERFACE_MAPPINGS.put(Map.class, HashMap.class);
        COLLECTION_INTERFACE_MAPPINGS.put(Collection.class, ArrayList.class);
        // 可按需添加其他集合接口映射
    }

    @Override
    public <T> T createInstance(PersistentEntity<?, T> entity, ParameterValueProvider provider) {
        // 先处理带参构造函数场景
        InstanceCreatorMetadata<? extends Parameter<?, ?>> creator = entity.getInstanceCreatorMetadata();
        if (creator != null && !creator.getParameters().isEmpty()) {
            Object[] params = creator.getParameters().stream()
                    .map(provider::getParameterValue)
                    .toArray();
            Class<?>[] paramTypes = creator.getParameters().stream()
                    .map(Parameter::getType)
                    .toArray(Class[]::new);
            try {
                return entity.getType().getConstructor(paramTypes).newInstance(params);
            } catch (Exception e) {
                throw new RuntimeException(e);
            }
        }
        // 无参构造函数实例化
        try {
            T instance = entity.getType().getDeclaredConstructor().newInstance();
            // 初始化未赋值的集合类型字段
            entity.doWithProperties(property -> {
                if (property.getType().isInterface() && COLLECTION_INTERFACE_MAPPINGS.containsKey(property.getType())) {
                    ReflectionUtils.makeAccessible(property.getField());
                    try {
                        // 字段已有初始值则跳过赋值
                        if (ReflectionUtils.getField(property.getField(), instance) == null) {
                            Object collectionImpl = COLLECTION_INTERFACE_MAPPINGS.get(property.getType()).getDeclaredConstructor().newInstance();
                            ReflectionUtils.setField(property.getField(), instance, collectionImpl);
                        }
                    } catch (Exception e) {
                        throw new RuntimeException(e);
                    }
                }
            });
            return instance;
        } catch (Exception e) {
            throw new RuntimeException(e);
        }
    }
}

然后将这个实例化器注册到Spring Data Elasticsearch的配置中：

import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.data.elasticsearch.config.ElasticsearchConfigurationSupport;
import org.springframework.data.elasticsearch.core.convert.MappingElasticsearchConverter;
import org.springframework.data.elasticsearch.core.mapping.ElasticsearchPersistentEntity;
import org.springframework.data.mapping.context.MappingContext;
import org.springframework.data.mapping.model.EntityInstantiators;

@Configuration
public class ElasticsearchConfig extends ElasticsearchConfigurationSupport {

    @Bean
    @Override
    public MappingElasticsearchConverter elasticsearchEntityMapper(MappingContext<? extends ElasticsearchPersistentEntity<?>, ?> mappingContext) {
        MappingElasticsearchConverter converter = super.elasticsearchEntityMapper(mappingContext);
        // 注册自定义实例化器
        converter.setInstantiators(new EntityInstantiators((entity, provider) -> new CustomCollectionInstantiator().createInstance(entity, provider)));
        return converter;
    }
}

该配置为Spring Data Elasticsearch专属，和JPA配置完全隔离，不会影响JPA Metamodel的正常运行。

可选轻量方案（仅需新增实体注解，无需修改字段声明） #

如果不想调整全局配置，可以给所有实体类增加全参数+无参构造函数，配合Lombok注解实现即可：

import lombok.AllArgsConstructor;
import lombok.Data;
import lombok.NoArgsConstructor;
import org.springframework.data.elasticsearch.annotations.Document;

import java.util.Set;

@Data
@NoArgsConstructor
@AllArgsConstructor
@Document(indexName = "customer")
public class Customer {
    private String id;
    private String name;
    private Set<Address> addresses;
}

Spring Data Elasticsearch检测到全参数构造函数时，会直接用查询到的参数值注入构造函数，不需要单独实例化集合接口，也不会和JPA逻辑冲突。

内容的提问来源于stack exchange，提问作者Ma Va