使用Elasticsearch Java Client时collapse的inner_hits解析失败问题排查

错误信息 #

Caused by: co.elastic.clients.elasticsearch._types.ElasticsearchException: [es/search] failed: [x_content_parse_exception] [1:500] [collapse] failed to parse field [inner_hits]

代码实现 #

FieldCollapse构建代码 #

FieldCollapse.Builder builder = new FieldCollapse.Builder() //
    .field(collapse.getField());

if (collapse.getInnerHits() != null) {
  List<InnerHits> innerHits = getInnerHitsBuilderList(collapse.getInnerHits());
  builder.innerHits(innerHits);
}
return Optional.of(builder.build());

InnerHits构建代码 #

InnerHits.Builder builder = new InnerHits.Builder().name(innerHit.getName());

NullOrEmpty.optional(innerHit.getFetchFields())
    .ifPresent(p -> builder.fields(Lists.mutable.ofAll(p)));
NullOrEmpty.optional(innerHit.getStoredFields())
    .ifPresent(p -> builder.storedField(Lists.mutable.ofAll(p)));

生成的查询结构（collapse部分） #

"collapse": {
    "field": "fielda.innerFieldb",
    "inner_hits": [
      {
        "name": "Topics",
        "size": 100,
        "collapse": {
          "field": "topic.topicId"
        },
        "fields": [
          "topic.shortName"
        ],
        "stored_field": [
          "topic.shortName"
        ]
      }
    ]
  }

官方文档示例格式 #

Elasticsearch官方语法中，inner_hits是单个对象而非数组：

"collapse": {
    "field": "user.id",
    "inner_hits": {
      "name": "most_recent",
      "size": 5,
      "sort": [ { "@timestamp": "desc" } ]
    },
    "max_concurrent_group_searches": 4
  }

注意：Java Client的FieldCollapse类中inner_hits被定义为List<InnerHits>类型，但服务端不接受数组格式，这是问题核心。

原因 #

Elasticsearch服务端的collapse语法仅支持单个inner_hits对象，但Java Client的FieldCollapse.Builder提供了接收列表的方法，导致生成的JSON中inner_hits是数组，与服务端要求的格式不匹配，触发解析错误。

解决方法 #

直接传入单个InnerHits实例，而非列表。修改后的FieldCollapse构建代码如下：

FieldCollapse.Builder builder = new FieldCollapse.Builder() //
    .field(collapse.getField());

if (collapse.getInnerHits() != null) {
  List<InnerHits> innerHitsList = getInnerHitsBuilderList(collapse.getInnerHits());
  if (!innerHitsList.isEmpty()) {
    builder.innerHits(innerHitsList.get(0)); // 传入单个InnerHits对象
  }
}
return Optional.of(builder.build());

修正后的查询结构 #

生成的collapse部分会符合服务端要求的对象格式：

"collapse": {
    "field": "fielda.innerFieldb",
    "inner_hits": {
      "name": "Topics",
      "size": 100,
      "collapse": {
        "field": "topic.topicId"
      },
      "fields": [
        "topic.shortName"
      ],
      "stored_field": [
        "topic.shortName"
      ]
    }
  }

内容的提问来源于stack exchange，提问作者John B