Elasticsearch“滚动查询“（Scrolling）的机制的与Java使用ES Client 调用滚动查询

前言

ES在进行普通的查询时，默认只会查询出来10条数据。我们通过设置es中的size可以将最终的查询结果从10增加到10000。如果需要查询数据量大于es的翻页限制或者需要将es的数据进行导出又当如何？
Elasticsearch提供了一种称为"滚动查询"（Scrolling）的机制，用于处理大型数据集的分页查询。滚动查询允许在持续的时间段内保持一个活动的搜索上下文，然后使用滚动ID进行迭代检索结果。滚动查询和关系型数据库中的游标有点类似，因此也叫游标查询

1. 滚动查询的一般步骤

1.1 发起初始搜索请求,返回命中结果和滚动ID

scroll=5m表示每个滚动查询的有效时间为5分钟

POST /your_index/_search?scroll=5m
{
  "size": 100,       // 每次返回的结果数量
  "query": { ... }   // 查询条件
}

命中结果：

{
  "_scroll_id": "DXF1ZXJ5QW5kRmV0Y2gBAAAAAAAAAD4WYm9laVYtZndUQlNsdDcwakFMNjU1QQ==",
  "hits": {
    "total": {
      "value": 10000,
      "relation": "eq"
    },
    "hits": [ ... ]  // 检索到的文档
  }
}

示例：

1.2 使用滚动ID检索下一页结果

POST /_search/scroll
{
  "scroll": "5m",
  "scroll_id": "DXF1ZXJ5QW5kRmV0Y2gBAAAAAAAAAD4WYm9laVYtZndUQlNsdDcwakFMNjU1QQ=="
}

示例：

POST /_search/scroll
 {
   "scroll": "5m",
   "scroll_id": "FGluY2x1ZGVfY29udGV4dF91dWlkDXF1ZXJ5QW5kRmV0Y2gBFDJPRXc0WWdCY1BLWlo1MTk4MmR3AAAAAAAAAXYWcWgwSW5CQUtScEd2T2QtRGtYaWliQQ=="
 }

1.4 重复执行直到没有检索结果返回

Elasticsearch将返回下一页结果和一个新的滚动ID。可以根据需要重复这个步骤，直到没有更多结果为止

1.5 清除滚动上下文释放资源

滚动查询结束后，您可以通过发送一个清除滚动上下文的请求来释放资源：

DELETE /_search/scroll
{
  "scroll_id": [
    "DXF1ZXJ5QW5kRmV0Y2gBAAAAAAAAAD4WYm9laVYtZndUQlNsdDcwakFMNjU1QQ=="
  ]
}

以上为滚动查询进行分页检索的基本过程。在每个滚动请求中，都需要提供先前滚动请求返回的滚动ID。这样Elasticsearch才能够维护搜索上下文并返回正确的结果

2.Java Elasticsearch客户端执行滚动查询

public static void main(String[] args) {

    long start = System.currentTimeMillis();

    //构建es HttpHost对象
    HttpHost httpHost1 = new HttpHost("192.168.1.1", 9200, "http");

    // 滚动时间窗口
    long scrollTime = 1L;
    // 每次返回的文档数量
    int batchSize = 20000;
    //索引名
    String indexName = "你的索引名称";

    try (RestHighLevelClient client = new RestHighLevelClient(RestClient.builder(httpHost1))) {

        //构建查询请求
        SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder();
        searchSourceBuilder.query(QueryBuilders.boolQuery());
        searchSourceBuilder.size(batchSize);
        //设置查询返回字段
        String[] includes = {};
        searchSourceBuilder.fetchSource(includes, null);

        // 滚动查询请求
        SearchRequest searchRequest = new SearchRequest(indexName);
        searchRequest.source(searchSourceBuilder);
        //设置请求滚动时间窗口时间
        searchRequest.scroll(TimeValue.timeValueMinutes(scrollTime));

        //执行首次检索
        SearchResponse searchResponse = client.search(searchRequest, RequestOptions.DEFAULT);

        //首次检索返回scrollId，用于下一次的滚动查询
        String scrollId = searchResponse.getScrollId();
        //获取首次检索命中结果
        SearchHit[] searchHits = searchResponse.getHits().getHits();

        //计数
        int count = 0;
        // 处理第一批结果
        for (SearchHit hit : searchHits) {
            // 处理单个文档
            JSONObject dataJson = (JSONObject) JSON.parse(hit.getSourceAsString());
            System.out.println("====对首次请求的进行处理，当前计数：" + count++);
        }

        // 处理滚动结果
        while (searchHits != null && searchHits.length > 0) {
            SearchScrollRequest scrollRequest = new SearchScrollRequest(scrollId);
            scrollRequest.scroll(TimeValue.timeValueMinutes(scrollTime));

            searchResponse = client.scroll(scrollRequest, RequestOptions.DEFAULT);
            scrollId = searchResponse.getScrollId();
            searchHits = searchResponse.getHits().getHits();

            for (SearchHit hit : searchHits) {
                JSONObject dataJson = (JSONObject) JSON.parse(hit.getSourceAsString());
                System.out.println("====滚动查询，当前计数：" + count++);
            }
        }

        // 清理滚动上下文
        ClearScrollRequest clearScrollRequest = new ClearScrollRequest();
        clearScrollRequest.addScrollId(scrollId);

        ClearScrollResponse clearScrollResponse = client.clearScroll(clearScrollRequest, RequestOptions.DEFAULT);
        boolean succeeded = clearScrollResponse.isSucceeded();

        long end = System.currentTimeMillis();
        System.out.println("共执行时间：" + (end - start) / 1000 + " s");

    } catch (Exception e) {
        System.out.println("===error==" + e.getMessage());
        e.printStackTrace();
    }
}

3. SpringDataElasticsearch滚动查询

import org.elasticsearch.action.search .*;
import org.elasticsearch.client .*;
import org.elasticsearch.common.unit .*;
import org.elasticsearch.index.query .*;
import org.elasticsearch.search .*;
import org.elasticsearch.search.builder .*;
import org.springframework.beans.factory.annotation .*;
import org.springframework.data.elasticsearch.core .*;
import org.springframework.data.elasticsearch.core.query .*;

public class ScrollSearchExample {

    @Autowired
    private ElasticsearchOperations elasticsearchOperations;

    public void performScrollSearch() {
        String scrollTime = "1m";  // 滚动时间窗口
        int batchSize = 100;  // 每次返回的文档数量

        QueryBuilder queryBuilder = QueryBuilders.matchQuery("field", "value");
        NativeSearchQueryBuilder searchQuery = new NativeSearchQueryBuilder();
        searchQuery.withQuery(queryBuilder).withPageable(PageRequest.of(0, batchSize)).build();


        SearchResponse searchResponse = elasticsearchOperations.startScroll(
                scrollTime,
                searchQuery,
                YourEntityClass.class,
                IndexCoordinates.of("your_index")
        );

        String scrollId = searchResponse.getScrollId();
        SearchHits<YourEntityClass> searchHits = searchResponse.getSearchHits();

        // 处理第一批结果
        for (SearchHit<YourEntityClass> hit : searchHits) {
            YourEntityClass entity = hit.getContent();
            // 处理单个文档
        }

        // 处理滚动结果
        while (searchHits != null && searchHits.hasSearchHits()) {
            searchResponse = elasticsearchOperations.continueScroll(scrollId, scrollTime, YourEntityClass.class);
            scrollId = searchResponse.getScrollId();
            searchHits = searchResponse.getSearchHits();

            for (SearchHit<YourEntityClass> hit : searchHits) {
                YourEntityClass entity = hit.getContent();
                // 处理单个文档
            }
        }

        // 清理滚动上下文
        elasticsearchOperations.clearScroll(scrollId);
    }
}