Elasticsearch Group Bucket结果

Question

我目前正在尝试根据elasticsearch中的索引生成排行榜。由于我的初始分组依赖于索引上的2个字段，我有：

{
    "size" : 0,
    "query" : {
        "bool": {
            // conditions and filters
        }
    },
    "aggs" : {
        "rank" : {
            "terms" : {
                "script" : {
                    "source": "doc['fieldA'].value + '-' + doc['fieldB'].value" // i needed 2 fields for grouping
                },
                "size" : 10000,
                "min_doc_count" : 10 // requirement for minimum records
            }
        }
    }
}

结果是：

{
    "took": 6,
    "timed_out": false,
    "_shards": {
        "total": 1,
        "successful": 1,
        "skipped": 0,
        "failed": 0
    },
    "hits": {
        "total": 1134,
        "max_score": 0,
        "hits": []
    },
    "aggregations": {
        "rank": {
            "doc_count_error_upper_bound": 0,
            "sum_other_doc_count": 0,
            "buckets": [
                {
                    "key": "groupA-user1",
                    "doc_count": 758
                },
                {
                    "key": "groupA-user2",
                    "doc_count": 289
                },
                {
                    "key": "groupB-user3",
                    "doc_count": 51
                },
                {
                    "key": "groupA-user4",
                    "doc_count": 30
                },
                {
                    "key": "groupA-user1",
                    "doc_count": 30
                },
                {
                    "key": "groupB-user10",
                    "doc_count": 10
                }
            ]
        }
    }
}

由于它按降序doc_count排序，我可以从中生成某种排行榜groupA-user1是第一位，groupA-user2是第二位，因为我提取了客户端...但是在groupA-user4和groupA-user1的情况下，两者都具有相同的doc_count，并且应该排在第4位。有没有办法在弹性搜索中单独执行此操作？比如通过doc_count对生成的桶进行分组？或者可能在第一个脚本中立即分组？顺便说一句，我使用elasticsearch 5.6