The query results obtained using elasticsearchOperation differ from those obtained using curl

I am currently attempting to query certain indexes using spring-data-elasticsearch's elasticSearchOperations, employing the following query:

        val index = IndexCoordinates.of("test-*-en")

        val searchQuery = NativeSearchQueryBuilder()
            .withFilter(
                QueryBuilders.boolQuery()
                    .must(QueryBuilders.matchQuery("english","test query : 12345678"))
            ).withMaxResults(30)
            .build()

        return elasticsearchOperations.search(searchQuery, TestEsDocument::class.java, index)

However, I've observed a discrepancy in the query results obtained compared to those obtained via curl, despite employing the exact same query. I suspect this variance arises because the document with a higher score wasn't prioritized during the search when utilizing esSearchOperations.

I've included the curl query for reference:

POST test-*-en/_search?size=30
{
  "query": 
    {
  "bool" : {
    "must" : [
      {
        "match" : {
          "english" : {
            "query" : "test query : 12345678",
            "operator" : "OR",
            "prefix_length" : 0,
            "max_expansions" : 50,
            "fuzzy_transpositions" : true,
            "lenient" : false,
            "zero_terms_query" : "NONE",
            "auto_generate_synonyms_phrase_query" : true,
            "boost" : 1.0
          }
        }
      }
    ],
    "adjust_pure_negative" : true,
    "boost" : 1.0
  }
}
}

Though uncertain, I suspect that the issue may lie in elasticsearchOperation not accurately reading the scores from elasticSearch, as shown in the provided screenshots.

Unfortunately, I am constrained to using es version 7.10.1-oss and cannot upgrade so I am using spring-data-elasticsearch:4.2.12, internally setting elasticsearch.version to 7.10.1 using the following options:

extra["elasticsearch.version"] = "7.10.1"

Could you advise if this issue arises from the version mismatch between spring-data-elasticsearch and my Elasticsearch cluster, and whether it's possible to resolve this issue? Alternatively, should I consider altering my query request to utilize a custom-built WebClient Post request instead?

Many thanks in advance!

Spring Data Elasticsearch returns 30 of the top scored results (the max score is 1.0) and you get 30 of them, you did not specify a sort order. The ones returned by curl, do they have the same score?

Edit:

I just saw that in your java code you add the query as a filter. So you will first get the result of getting all data without any scoring and then pass them through the filter. Add the query as query and not as filter.