Elasticsearch篩選器文檔按字段分組

Question

我有一些文件：

{"name": "John", "district": 1},
{"name": "Mary", "district": 2},
{"name": "Nick", "district": 1},
{"name": "Bob", "district": 3},
{"name": "Kenny", "district": 1}

如何按地區過濾/選擇不同的文檔？

{"name": "John", "district": 1},
{"name": "Mary", "district": 2},
{"name": "Bob", "district": 3}

在SQL中，我可以使用GROUP BY。 我嘗試了條件聚合，但返回的計數卻不同。

"aggs": {
  "distinct": {
    "terms": {
      "field": "district",
      "size": 0
    }
  }
}

感謝您的幫助！ :-)

Answer 1

如果您的ElasticSearch版本為1.3或更高版本，則可以使用top_hits類型的子聚合 ，默認情況下，它將為您提供按查詢得分排序的前三個匹配文檔（此處為1，因為您使用match_all查詢）。

您可以將size參數設置為大於3。

以下數據集和查詢：

POST /test/districts/
{"name": "John", "district": 1}

POST /test/districts/
{"name": "Mary", "district": 2}

POST /test/districts/
{"name": "Nick", "district": 1}

POST /test/districts/
{"name": "Bob", "district": 3}

POST test/districts/_search
{
  "size": 0, 
  "aggs":{
    "by_district":{
      "terms": {
        "field": "district",
        "size": 0
      },
      "aggs": {
        "tops": {
          "top_hits": {
            "size": 10
          }
        }
      }
    }
  }
}

將以您想要的方式輸出文檔：

{
   "took": 5,
   "timed_out": false,
   "_shards": {
      "total": 5,
      "successful": 5,
      "failed": 0
   },
   "hits": {
      "total": 4,
      "max_score": 0,
      "hits": []
   },
   "aggregations": {
      "by_district": {
         "buckets": [
            {
               "key": 1,
               "key_as_string": "1",
               "doc_count": 2,
               "tops": {
                  "hits": {
                     "total": 2,
                     "max_score": 1,
                     "hits": [
                        {
                           "_index": "test",
                           "_type": "districts",
                           "_id": "XYHu4I-JQcOfLm3iWjTiOg",
                           "_score": 1,
                           "_source": {
                              "name": "John",
                              "district": 1
                           }
                        },
                        {
                           "_index": "test",
                           "_type": "districts",
                           "_id": "5dul2XMTRC2IpV_tKRRltA",
                           "_score": 1,
                           "_source": {
                              "name": "Nick",
                              "district": 1
                           }
                        }
                     ]
                  }
               }
            },
            {
               "key": 2,
               "key_as_string": "2",
               "doc_count": 1,
               "tops": {
                  "hits": {
                     "total": 1,
                     "max_score": 1,
                     "hits": [
                        {
                           "_index": "test",
                           "_type": "districts",
                           "_id": "I-9Gd4OYSRuexhP1dCdQ-g",
                           "_score": 1,
                           "_source": {
                              "name": "Mary",
                              "district": 2
                           }
                        }
                     ]
                  }
               }
            },
            {
               "key": 3,
               "key_as_string": "3",
               "doc_count": 1,
               "tops": {
                  "hits": {
                     "total": 1,
                     "max_score": 1,
                     "hits": [
                        {
                           "_index": "test",
                           "_type": "districts",
                           "_id": "bti2y-OUT3q2mBNhhI3xeA",
                           "_score": 1,
                           "_source": {
                              "name": "Bob",
                              "district": 3
                           }
                        }
                     ]
                  }
               }
            }
         ]
      }
   }
}

Answer 2

彈性搜索不會提供有關價值的獨特文檔或按唯一價值分組的文檔。 但是有解決方法，如果您使用的是Java客戶端，則可以執行此操作，也可以將其轉換為適合的語言

SearchResponse response = client.prepareSearch().execute().actionGet();
SearchHits hits = response.getHits();

Iterator<SearchHit> iterator = hits.iterator();
Map<String, SearchHit> distinctObjects = new HashMap<String,SearchHit>();
while (iterator.hasNext()) {
    SearchHit searchHit = (SearchHit) iterator.next();
    Map<String, Object> source = searchHit.getSource();
    if(source.get("district") != null){
        distinctObjects.put(source.get("district").toString(),source);
    }

}