使用es 自定义function score做排序

如何初始化es客户端连接可以参考这个文章： 

【es笔记三】Java Low Level REST Client_西瓜游侠的博客-CSDN博客_restclientbuilder

 ElasticSearch6.5.0 【Java客户端之REST Client】_weixin_33704234的博客-CSDN博客

 通过从文件中读取算分脚本以及参数，创建script对象

Map params = new HashMap<>();
params.put("factor1", factor1Value);        
script script = new script(scriptType.INLINE, "painless", script, params);
scriptScoreFunctionBuilder builder = new scriptScoreFunctionBuilder(script);
FunctionScoreQueryBuilder scoreBuilder =
            QueryBuilders.functionScoreQuery(getQueryBuilder(esReq), builder).boostMode(CombineFunction.REPLACE);

BoolQueryBuilder boolQuery = QueryBuilders.boolQuery();
boolQuery.must(queryBuilder);
boolQuery.filter(queryBuilder);
sourceBuilder.query(boolQuery);
searchRequest.source(sourceBuilder);

通过 @Scheduled(cron = "${refresh.script.cron}")

定时从文件重新读取排序脚本。

比如要实现置顶和沉底功能，在doc中添加topsink字段

curl -H 'Content-Type: application/json' -X PUT -d '{"properties":{"sink":{"type": "byte"}}}' "http://localhost:9200/docName/_mapping"

打分脚本添加判断

if (doc['top'].size() !=0 && doc['top'].value == 1) {
    return 100000
}
if (doc['sinkLevel'].size() !=0 && doc['sink'].value == 1) {
    return 0
}

type为text的和keyword的区别

关于ES字符串类型（Text vs keyword）的选择_HiterCoder-CSDN博客_es keyword text

如果需要分词才应该使用text类型，每个字段可以单独定义。比如

      "sensitiveWord": {
        "type": "text",
        "analyzer": "comma",
        "search_analyzer": "comma"
      }