Spring Data ES跨多索引查询返回LinkedHashMap而非预期实体类问题求助

当你用Object.class作为目标类型查询多索引时，Spring Data Elasticsearch的默认映射器无法自动识别每个文档对应的具体实体/DTO类型：

• 单独查询单个索引时，你大概率是用对应实体类（比如Director.class）发起查询，映射器能明确目标类型完成转换；
• 跨多索引查询用Object.class时，映射器没有类型参考，对于结构无法匹配到已知实体类的文档，会直接反序列化为LinkedHashMap；
• 你存入时用JSONObject格式，没有在文档中嵌入类型标识（比如@TypeAlias对应的字段），映射器更无法自动推断类型。

方案1：根据文档来源索引手动转换 #

利用SearchHit的getIndex()方法获取文档所属索引，直接对应目标类型进行转换：

ObjectMapper objectMapper = new ObjectMapper();
List<Object> searchResult = filmSearchHits.stream().map(hit -> {
    Object content = hit.getContent();
    if (content instanceof LinkedHashMap<?, ?> map) {
        String indexName = hit.getIndex();
        return switch (indexName) {
            case "directors_index" -> objectMapper.convertValue(map, DirectorDTO.class);
            case "actors_index" -> objectMapper.convertValue(map, ActorDTO.class);
            case "native_film_index" -> objectMapper.convertValue(map, FilmDTO.class);
            case "serial_index" -> objectMapper.convertValue(map, SerialDTO.class);
            default -> content;
        };
    }
    return content;
}).toList();

方案2：给文档添加类型标识字段，按标识转换 #

1. 存入文档时，统一添加一个类型标识字段（比如doc_type），值对应索引类型：
   • directors_index的文档加"doc_type": "director"
   • actors_index的文档加"doc_type": "actor"
   • native_film_index的文档加"doc_type": "film"
   • serial_index的文档加"doc_type": "serial"
2. 查询时根据标识字段转换：

   ObjectMapper objectMapper = new ObjectMapper();
   List<Object> searchResult = filmSearchHits.stream().map(hit -> {
       Object content = hit.getContent();
       if (content instanceof LinkedHashMap<?, ?> map) {
           String docType = (String) map.get("doc_type");
           return switch (docType) {
               case "director" -> objectMapper.convertValue(map, DirectorDTO.class);
               case "actor" -> objectMapper.convertValue(map, ActorDTO.class);
               case "film" -> objectMapper.convertValue(map, FilmDTO.class);
               case "serial" -> objectMapper.convertValue(map, SerialDTO.class);
               default -> content;
           };
       }
       return content;
   }).toList();
   

方案3：自定义EntityMapper实现自动类型推断 #

自定义一个EntityMapper，在反序列化时根据文档来源索引选择目标类型，替换Spring Data ES的默认映射器：

1. 实现自定义EntityMapper：

   public class MultiIndexEntityMapper implements EntityMapper {
       private final ObjectMapper objectMapper;
   
       public MultiIndexEntityMapper(ObjectMapper objectMapper) {
           this.objectMapper = objectMapper;
       }
   
       @Override
       public String mapToString(Object object) throws IOException {
           return objectMapper.writeValueAsString(object);
       }
   
       @Override
       public <T> T mapToObject(String source, Class<T> clazz, IndexCoordinates indexCoordinates) throws IOException {
           if (Object.class.equals(clazz)) {
               // 根据索引名匹配目标类型
               Class<?> targetClass = switch (indexCoordinates.getIndexNames()[0]) {
                   case "directors_index" -> DirectorDTO.class;
                   case "actors_index" -> ActorDTO.class;
                   case "native_film_index" -> FilmDTO.class;
                   case "serial_index" -> SerialDTO.class;
                   default -> clazz;
               };
               return objectMapper.readValue(source, (Class<T>) targetClass);
           }
           return objectMapper.readValue(source, clazz);
       }
   }
   

2. 配置ElasticsearchOperations替换默认映射器：

   @Bean
   public ElasticsearchOperations elasticsearchOperations(Client client, ElasticsearchConverter converter) {
       SimpleElasticsearchOperations operations = new SimpleElasticsearchOperations(client, converter);
       operations.setEntityMapper(new MultiIndexEntityMapper(new ObjectMapper()));
       return operations;
   }
   

注意：此方案需结合Spring Data ES版本调整实现细节，部分版本的EntityMapper方法签名可能略有差异。

内容的提问来源于stack exchange，提问作者Dmitriy_Ze