86_熟练掌握ES Java API_基于scroll实现月度销

86_熟练掌握ES Java API_基于scroll实现月度销售数据批量下载

比如说，现在要下载大批量的数据，从es，放到excel中，我们说，月度，或者年度，销售记录，很多，比如几千条，几万条，几十万条

其实就要用到我们之前讲解的es scroll api，对大量数据批量的获取和处理

PUT /car_shop/sales/4
{
    "brand": "宝马",
    "name": "宝马320",
    "price": 320000,
    "produce_date": "2017-01-01",
    "sale_price": 280000,
    "sale_date": "2017-01-25"
}

就是要看宝马的销售记录

2条数据，做一个演示，每个批次下载一条宝马的销售记录，分2个批次给它下载完

SearchResponse scrollResp = client.prepareSearch("car_shop")
        .addTypes("sales")
        .setScroll(new TimeValue(60000))
        .setQuery(termQuery("brand.raw", "宝马"))
        .setSize(1)
        .get(); 
do {
    for (SearchHit hit : scrollResp.getHits().getHits()) {
        
    }
    
    scrollResp = client.prepareSearchScroll(scrollResp.getScrollId())
            .setScroll(new TimeValue(60000))
            .execute()
            .actionGet();
} while(scrollResp.getHits().getHits().length != 0);

源代码如下:

public class ScollDownloadSalesDataApp {
    
    @SuppressWarnings({ "resource", "unchecked" })
    public static void main(String[] args) throws Exception {
        Settings settings = Settings.builder()
                .put("cluster.name", "elasticsearch")
                .build();
        
        TransportClient client = new PreBuiltTransportClient(settings)
                .addTransportAddress(new InetSocketTransportAddress(InetAddress.getByName("localhost"), 9300)); 
    
        SearchResponse searchResponse = client.prepareSearch("car_shop") 
                .setTypes("sales")
                .setQuery(QueryBuilders.termQuery("brand.keyword", "宝马"))
                .setScroll(new TimeValue(60000))
                .setSize(1)
                .get();
        
        int batchCount = 0;
        
        do {
            for(SearchHit searchHit : searchResponse.getHits().getHits()) {
                System.out.println("batch: " + ++batchCount); 
                System.out.println(searchHit.getSourceAsString());  
                
                // 每次查询一批数据，比如1000行，然后写入本地的一个excel文件中
                
                // 如果说你一下子查询几十万条数据，不现实，jvm内存可能都会爆掉
            }
            
            searchResponse = client.prepareSearchScroll(searchResponse.getScrollId())
                    .setScroll(new TimeValue(60000))
                    .execute()
                    .actionGet();
        } while(searchResponse.getHits().getHits().length != 0);
        
        client.close();
    }
    
}