嵌套聚合求百分比——es记一次关于 bucket_script 根据需求不断改进的DSL及jestAPI
- 需求：求每个科室下不同病历级别的百分比。
- 思路：按每个科室分组，再按级别分组，再求占比。
- 开动。
第一次尝试：
POST /medicalrecord/my_type/_search?size=0
{
"size": 0,
"aggs": {
"department_group": {
"terms": {
"field": "department_secondary",
"size": 1000000
},
"aggs": {
"total_record": {
"value_count": {
"field": "emr_id"
}
},
"levels": {
"filter": {
"terms": {
"has_bug": [
"1",
"2",
"3"
]
}
},
"aggs": {
"level_group": {
"terms": {
"field": "has_bug",
"size": 3
},
"aggs": {
"every_level_count": {
"value_count": {
"field": "emr_id"
}
},
"level_percentage": {
"bucket_script": {
"buckets_path": {
"every_level_count": "every_level_count",
"department_total_record": "total_record"
},
"script": "params.every_level_count / params.department_total_record"
}
}
}
}
}
}
}
}
}
}报错：
{
"error": {
"root_cause": [
{
"type": "illegal_argument_exception",
"reason": "No aggregation found for path [total_record]"
}
],
"type": "search_phase_execution_exception",
"reason": "all shards failed",
"phase": "query",
"grouped": true,
"failed_shards": [
{
"shard": 0,
"index": "medicalrecord",
"node": "p9ZUXdr9R3mxulJKxgNtGA",
"reason": {
"type": "illegal_argument_exception",
"reason": "No aggregation found for path [total_record]"
}
}
]
},
"status": 400
}第二次尝试：
POST /medicalrecord/my_type/_search?size=0
{
"size": 0,
"aggs": {
"department_group": {
"terms": {
"field": "department_secondary",
"size": 1000000
},
"aggs": {
"total_record": {
"value_count": {
"field": "emr_id"
}
},
"levels": {
"filter": {
"terms": {
"has_bug": [
"1",
"2",
"3"
]
}
},
"aggs": {
"level_group": {
"terms": {
"field": "has_bug",
"size": 3
},
"aggs": {
"every_level_count": {
"value_count": {
"field": "emr_id"
}
}
}
}
}
},
"level_percentage": {
"bucket_script": {
"buckets_path": {
"every_level_count": "levels > level_group> every_level_count",
"department_total_record": "total_record"
},
"script": "params.every_level_count / params.department_total_record"
}
}
}
}
}
}报错：
{
"error": {
"root_cause": [],
"type": "search_phase_execution_exception",
"reason": "",
"phase": "fetch",
"grouped": true,
"failed_shards": [],
"caused_by": {
"type": "aggregation_execution_exception",
"reason": "buckets_path must reference either a number value or a single value numeric metric aggregation, got: java.lang.Object[]"
}
},
"status": 503
}错误总结：buckets_path 不能向上 取值，因为他说是相对路径，返回值必须是个数字类型。
哎，好失望啊。最终能完成需求的版本，has_bug 其实就是级别。
#求所有科室病历某个或者某几个级别的占比
POST /medicalrecord/my_type/_search?size=0
{
"size": 0,
"aggs": {
"department_group": {
"terms": {
"field": "department_secondary",
"size": 1000000
},
"aggs": {
"levels": {
"filter": {
"terms": {
"has_bug": [
"1",
"2"
]
}
}
},
"level_percentage": {
"bucket_script": {
"buckets_path": {
"level_count": "levels > _count",
"department_total_record": "_count"
},
"script": "params.level_count / params.department_total_record"
}
}
}
}
}
}过了一天，我突然想到多加一个filter怎么样：
POST /medicalrecord/my_type/_search?size=0
{
"size": 0,
"aggs": {
"department_group": {
"terms": {
"field": "department_secondary",
"size": 1000000
},
"aggs": {
"levels_2_3": {
"filter": {
"terms": {
"has_bug": [
"2",
"3"
]
}
}
},
"level_2_3_percentage": {
"bucket_script": {
"buckets_path": {
"level_count": "levels_2_3 > _count",
"department_total_record": "_count"
},
"format": "#.####",
"script": "params.level_count / params.department_total_record"
}
},
"levels_1": {
"filter": {
"terms": {
"has_bug": [
"1"
]
}
}
},
"level_1_percentage": {
"bucket_script": {
"buckets_path": {
"level_count": "levels_1 > _count",
"department_total_record": "_count"
},
"format": "#.####",
"script": "params.level_count / params.department_total_record"
}
}
}
}
}
}
返回结果：
{
"took": 390,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"failed": 0
},
"hits": {
"total": 13212,
"max_score": 0,
"hits": []
},
"aggregations": {
"department_group": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "乳腺甲状腺科一病区",
"doc_count": 770,
"levels_2_3": {
"doc_count": 770
},
"levels_1": {
"doc_count": 0
},
"level_2_3_percentage": {
"value": 1,
"value_as_string": "1"
},
"level_1_percentage1": {
"value": 0,
"value_as_string": "0"
}
},
{
"key": "内三",
"doc_count": 710,
"levels_2_3": {
"doc_count": 708
},
"levels_1": {
"doc_count": 2
},
"level_2_3_percentage": {
"value": 0.9971830985915493,
"value_as_string": "0.9972"
},
"level_1_percentage1": {
"value": 0.0028169014084507044,
"value_as_string": "0.0028"
}
},
{
"key": "康复医学科病区",
"doc_count": 668,
"levels_2_3": {
"doc_count": 668
},
"levels_1": {
"doc_count": 0
},
"level_2_3_percentage": {
"value": 1,
"value_as_string": "1"
},
"level_1_percentage1": {
"value": 0,
"value_as_string": "0"
}
},
{
"key": "内一西病区",
"doc_count": 604,
"levels_2_3": {
"doc_count": 602
},
"levels_1": {
"doc_count": 2
},
"level_2_3_percentage": {
"value": 0.9966887417218543,
"value_as_string": "0.9967"
},
"level_1_percentage1": {
"value": 0.0033112582781456954,
"value_as_string": "0.0033"
}
}
...es6.3有一个buckets_sort 功能 可以 看看 能不能将聚合后的结果排序。
另外jestAPI的写法放在下面：
@Test
public void test06() {
HashMap bucketsPathHashMap23 = new HashMap<>(2);
bucketsPathHashMap23.put("level_count", "levels_2_3 > _count");
bucketsPathHashMap23.put("department_total_record", "_count");
HashMap bucketsPathHashMap1 = new HashMap<>(2);
bucketsPathHashMap1.put("level_count", "levels_1 > _count");
bucketsPathHashMap1.put("department_total_record", "_count");
TermsAggregationBuilder termsAggregationBuilder = AggregationBuilders.terms("department_group").field("department_secondary").size(Integer.MAX_VALUE)
.subAggregation(AggregationBuilders.filter("levels_2_3", QueryBuilders.termsQuery("has_bug", "2", "3")))
.subAggregation(PipelineAggregatorBuilders.bucketScript("level_2_3_percentage", bucketsPathHashMap23, new Script("params.level_count / params.department_total_record")).format("#.####"))
.subAggregation(AggregationBuilders.filter("levels_1", QueryBuilders.termsQuery("has_bug", "1")))
.subAggregation(PipelineAggregatorBuilders.bucketScript("level_1_percentage", bucketsPathHashMap1, new Script("params.level_count / params.department_total_record")).format("#.####"));
SearchSourceBuilder searchSourceBuilder = new SearchSourceBuilder()
//聚合
.aggregation(termsAggregationBuilder)
// 设置 不排序 按照文档存入顺序返回
.sort("_doc");
Search.Builder builder = new Search.Builder(searchSourceBuilder.toString())
.addIndex("index")
.addType("type")
.setParameter(Parameters.SIZE, GlobalConstants.HITS_SIZE)
.setParameter(Parameters.SCROLL, "5m")
//可以屏蔽元数据
.setParameter("filter_path", "hits.hits._source,hits.hits._id,hits.hits,_scroll_id");
System.out.println(searchSourceBuilder.toString());
}