jest 客户端 实现 Sliced+Scroll并行查询

首先感谢 Bboss作者兼 elasticsearch交流 群主(一个非常热心的大佬)

这里直接看代码：

public Map<Integer, List<JestResult>> searchSlicedScrolls(MyYangBao yangBao, QueryBuilder queryBuilders, Set<String> includePatterns) {
    final int slicesMax = 5;
    ExecutorService singleThreadPool = UtilsJava.returnSingleThreadPool(slicesMax);
    Map<Integer, List<JestResult>> stringListMap = Collections.synchronizedMap(new HashMap<>(slicesMax));
    //分成5个切片去执行
    for (int id = 0; id < slicesMax; id++) {
        int finalId = id;
        ArrayList<JestResult> jestResults = new ArrayList<>();
        singleThreadPool.submit(() -> {
            SliceBuilder sliceBuilder = new SliceBuilder(finalId, slicesMax);
            SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder()
                    .query(queryBuilders)
                    // 设置 不排序 按照文档存入顺序返回
                    .sort("_doc")
                    .slice(sliceBuilder);
            Search.Builder builder = new Search.Builder(searchSourceBuilder.toString())
                    .addIndex(yangBao.getIndexName())
                    .addType(yangBao.getTypeName())
                    .setParameter(Parameters.SIZE, GlobalConstants.HITS_SIZE)
                    .setParameter(Parameters.SCROLL, "5m")
                    //可以屏蔽元数据
                    .setParameter("filter_path", "hits.hits,hits.hits._source,hits.hits._id,_scroll_id");
            for (String includePattern : includePatterns) {
                builder.addSourceIncludePattern(includePattern);
            }
            Search search = builder.build();
            JestResult jestResult = null;
            try {
                jestResult = yangBao.getJestClient().execute(search);
            } catch (IOException e) {
                e.printStackTrace();
            }
            assert jestResult != null;
            if (jestResult.getJsonObject().getAsJsonObject(GlobalConstants.JSON_HITS).getAsJsonArray(GlobalConstants.JSON_HITS).size() > 0) {
                String   scrollId = jestResult.getJsonObject().get(GlobalConstants.JSON_SCROLL_ID).getAsString();
                jestResults.add(jestResult);
                do {
                    SearchScroll scroll = new SearchScroll.Builder(scrollId, "5m")
                            .setParameter("filter_path", "hits.hits,hits.hits._source,hits.hits._id,_scroll_id")
                            .build();
                    try {
                        jestResult = yangBao.getJestClient().execute(scroll);
                    } catch (IOException e) {
                        e.printStackTrace();
                    }
                    if (jestResult.getJsonObject().getAsJsonObject(GlobalConstants.JSON_HITS).getAsJsonArray(GlobalConstants.JSON_HITS).size() > 0) {
                        jestResults.add(jestResult);
                    } else {
                        break;
                    }
                } while (true);

            }
        });
        stringListMap.put(finalId, jestResults);
    }
    singleThreadPool.shutdown();
    try {
        singleThreadPool.awaitTermination(100, TimeUnit.SECONDS);
    } catch (InterruptedException e) {
        e.printStackTrace();
    }

    return stringListMap;
}

参考：https://my.oschina.net/bboss/blog/1788729