個(gè)人專題目錄](méi)(http://www.itdecent.cn/p/140e2a59db2c)

1. elasticsearch文檔過(guò)濾查詢結(jié)果

1.1 filter與query對(duì)比

filter，僅僅只是按照搜索條件過(guò)濾出需要的數(shù)據(jù)而已，不計(jì)算任何相關(guān)度分?jǐn)?shù)，對(duì)相關(guān)度沒(méi)有任何影響；
query，會(huì)去計(jì)算每個(gè)document相對(duì)于搜索條件的相關(guān)度，并按照相關(guān)度進(jìn)行排序；

一般來(lái)說(shuō)，如果你是在進(jìn)行搜索，需要將最匹配搜索條件的數(shù)據(jù)先返回，那么用query；如果你只是要根據(jù)一些條件篩選出一部分?jǐn)?shù)據(jù)，不關(guān)注其排序，那么用filter；
除非是你的這些搜索條件，你希望越符合這些搜索條件的document越排在前面返回，那么這些搜索條件要放在query中；如果你不希望一些搜索條件來(lái)影響你的document排序，那么就放在filter中即可；

1.2 filter與query性能對(duì)比

filter，不需要計(jì)算相關(guān)度分?jǐn)?shù)，不需要按照相關(guān)度分?jǐn)?shù)進(jìn)行排序，同時(shí)還有內(nèi)置的自動(dòng)cache最常使用filter的數(shù)據(jù)

query，相反，要計(jì)算相關(guān)度分?jǐn)?shù)，按照分?jǐn)?shù)進(jìn)行排序，而且無(wú)法cache結(jié)果.

ElasticSearch提供了一種特殊的緩存，即過(guò)濾器緩存（filter cache），用來(lái)存儲(chǔ)過(guò)濾器的結(jié)果，被緩存的過(guò)濾器并不需要消耗過(guò)多的內(nèi)存（因?yàn)樗鼈冎淮鎯?chǔ)了哪些文檔能與過(guò)濾器相匹配的相關(guān)信息），而且可供后續(xù)所有與之相關(guān)的查詢重復(fù)使用，從而極大地提高了查詢性能。

1.3 filter

POST /book-index/_search
{
  "query": {
    "bool": {
      "filter": [
        {
          "term": {
            "brandName": {
              "value": "飛利浦",
              "boost": 1
            }
          }
        },
        {
          "term": {
            "categoryName": {
              "value": "手機(jī)",
              "boost": 1
            }
          }
        }
      ],
      "adjust_pure_negative": true,
      "boost": 1
    }
  }
}

@Override
public void filterInBoolQuery(String indexName) throws Exception {
    SearchRequest searchRequest = new SearchRequest(indexName);
    SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
    BoolQueryBuilder queryBuilder = QueryBuilders.boolQuery();
    queryBuilder.filter(QueryBuilders.termQuery("brandName", "飛利浦"));
    queryBuilder.filter(QueryBuilders.termQuery("categoryName", "手機(jī)"));
    baseQuery.builder(indexName, queryBuilder);
}

1.4 range過(guò)濾器

range過(guò)濾器允許我們將搜索范圍限制在字段值給定的界限范圍內(nèi)。

POST /book-index/_search
{
  "from": 0,
  "size": 100,
  "query": {
    "bool": {
      "must": [
        {
          "match": {
            "title": {
              "query": "三星",
              "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
            }
          }
        }
      ],
      "filter": [
        {
          "range": {
            "price": {
              "from": 2000,
              "to": 3000,
              "include_lower": true,
              "include_upper": true,
              "boost": 1
            }
          }
        }
      ],
      "must_not": [
        {
          "term": {
            "brandName": {
              "value": "諾基亞",
              "boost": 1
            }
          }
        }
      ],
      "should": [
        {
          "term": {
            "categoryName": {
              "value": "手機(jī)",
              "boost": 1
            }
          }
        },
        {
          "term": {
            "categoryName": {
              "value": "平板電視",
              "boost": 1
            }
          }
        }
      ],
      "adjust_pure_negative": true,
      "boost": 1
    }
  }
}

@Override
public void rangeQuery(String indexName, String fieldName, int from, int to) throws Exception {
    BoolQueryBuilder queryBuilder = QueryBuilders.boolQuery();
    queryBuilder.should(QueryBuilders.termQuery("categoryName", "手機(jī)"));
    queryBuilder.should(QueryBuilders.termQuery("categoryName", "平板電視"));
    queryBuilder.must(QueryBuilders.matchQuery("title", "三星"));
    queryBuilder.mustNot(QueryBuilders.termQuery("brandName", "諾基亞"));
    queryBuilder.filter(QueryBuilders.rangeQuery(fieldName).from(from).to(to));
    baseQuery.builder(indexName, queryBuilder);
}

@Test
public void testRangeCardQuery() throws Exception {
    filterQuery.rangeQuery(indexName,"price",1,5);
}

1.5 exists 過(guò)濾器

exists 過(guò)濾器只選擇給定字段的文檔。如果某字段沒(méi)有值，就不選擇。

比如我們查詢響應(yīng)時(shí)間不為空的文檔：

POST /book-index/_search
{
  "query": {
    "exists": {
      "field": "title",
      "boost": 1
    }
  }
}

@Override
public void existQuery(String indexName, String fieldName) throws Exception {
    baseQuery.builder(indexName, QueryBuilders.existsQuery(fieldName));
}

@Test
public void testExistQuery() throws IOException {
    Goods goods = new Goods();
    goods.setId(1);
    //goods.setTitle("new2 - 阿爾卡特 (OT-927) 炭黑 聯(lián)通3G手機(jī) 雙卡雙待");
    goods.setPrice(22.32);
    goods.setStock(23232);
    goods.setSaleNum(32);
    goods.setCreateTime(new Date());
    goods.setCategoryName("平板電視");
    goods.setBrandName("三星");
    goods.setSpecStr("{\"電視屏幕尺寸\":\"46英寸\"}");
    docService.addDoc(Constants.INDEX_NAME, objectMapper.writeValueAsString(goods), "1");
    filterQuery.existQuery(Constants.INDEX_NAME, "title");
}