问题描述
我正在使用 ElasticSearch 6.3 并且我正在处理具有多个子聚合的聚合，其中我需要根据较低级别的 reverse_nested 聚合的 doc_count 对顶级聚合存储桶进行排序。
我的索引是这样创建的：
PUT /myindex
{
"mappings": {
"default": {
"properties": {
"items": {
"type": "nested","properties": {
"subitems": {
"type": "nested","properties": {
"id": {
"type": "long"
},"name": {
"type": "keyword"
}
}
}
}
},"name": {
"type": "keyword"
}
}
}
}
}
这些是我编入索引的示例文档：
{
"name": "Document #1","items": [
{
"subitems": [
{
"id": 1,"name": "Subitem #1"
},{
"id": 2,"name": "Subitem #2"
}
]
},{
"subitems": [
{
"id": 2,"name": "Subitem #2"
},{
"id": 3,"name": "Subitem #3"
}
]
}
]
}
{
"name": "Document #2","items": [
{
"subitems": [
{
"id": 2,"name": "Subitem #2"
}
]
}
]
}
{
"name": "Document #3","items": [
{
"subitems": [
{
"id": 3,"name": "Subitem #3"
}
]
},"name": "Subitem #2"
}
]
}
]
}
{
"name": "Document #4",{
"id": 5,"name": "Subitem #5"
}
]
}
]
}
{
"name": "Document #5","name": "Subitem #2"
}
]
}
]
}
{
"name": "Document #6","name": "Subitem #3"
}
]
}
]
}
{
"name": "Document #7","name": "Subitem #3"
}
]
}
]
}
{
"name": "Document #8","name": "Subitem #3"
}
]
}
]
}
{
"name": "Document #9","name": "Subitem #3"
}
]
}
]
}
我需要我的聚合才能提取包含每个子项 ID/名称对的文档数量。 （考虑子项 ID 始终对应于相同的子项名称）。 即：
id | name | count
---+------------+------
2 | Subitem #2 | 5
3 | Subitem #3 | 6
1 | Subitem #1 | 1
5 | Subitem #5 | 1
这是原始的聚合查询：
GET /myindex/default/_search
{
"size": 0,"aggregations": {
"my_nested_agg": {
"nested": {
"path": "items.subitems"
},"aggregations": {
"subitem_id": {
"terms": {
"field": "items.subitems.id"
},"aggregations": {
"subitem_name": {
"terms": {
"field": "items.subitems.name"
},"aggregations": {
"my_rev_agg": {
"reverse_nested": {}
}
}
}
}
}
}
}
}
}
聚合似乎返回了我需要的所有数据：
{
"took": 0,"timed_out": false,"_shards": {
"total": 5,"successful": 5,"skipped": 0,"Failed": 0
},"hits": {
"total": 9,"max_score": 0.0,"hits": []
},"aggregations": {
"my_nested_agg": {
"doc_count": 19,"subitem_id": {
"doc_count_error_upper_bound": 0,"sum_other_doc_count": 0,"buckets": [
{
"key": 2,"doc_count": 11,"subitem_name": {
"doc_count_error_upper_bound": 0,"buckets": [
{
"key": "Subitem #2","my_rev_agg": {
"doc_count": 5
}
}
]
}
},{
"key": 3,"doc_count": 6,"buckets": [
{
"key": "Subitem #3","my_rev_agg": {
"doc_count": 6
}
}
]
}
},{
"key": 1,"doc_count": 1,"buckets": [
{
"key": "Subitem #1","my_rev_agg": {
"doc_count": 1
}
}
]
}
},{
"key": 5,"buckets": [
{
"key": "Subitem #5","my_rev_agg": {
"doc_count": 1
}
}
]
}
}
]
}
}
}
}
但是，存储桶根据“subitem_id”子聚合的 doc_count 按降序排列。
相反，我需要根据 reverse_nested 子聚合的 doc_count 按降序对存储桶进行排序。像这样：
id | name | count
---+------------+------
3 | Subitem #3 | 6
2 | Subitem #2 | 5
1 | Subitem #1 | 1
5 | Subitem #5 | 1
我尝试通过以下查询实现此目的：
GET /myindex/default/_search
{
"size": 0,"aggregations": {
"subitem_id": {
"terms": {
"field": "items.subitems.id","order": [
{
"subitem_name>my_rev_agg._count": "desc"
}
]
},"aggregations": {
"my_rev_agg": {
"reverse_nested": {}
}
}
}
}
}
}
}
}
}
但后来我得到了错误：
无效的聚合订单路径 [subitem_name>my_rev_agg._count]。存储桶只能在由路径内的零个或多个单存储桶聚合以及路径末端的最终单存储桶或指标聚合构建的子聚合器路径上排序。子路径 [subitem_name] 指向非单桶聚合
能否请您指教。 预先非常感谢您。
解决方法
我找到了一个可以满足我的要求的解决方案。关键是将 reverse_nested 聚合移到用于检索名称的术语子聚合之外：
GET /myindex/default/_search
{
"size": 0,"aggregations": {
"my_nested_agg": {
"nested": {
"path": "items.subitems"
},"aggregations": {
"subitem_id": {
"terms": {
"field": "items.subitems.id","order": [
{
"my_rev_agg": "desc"
}
]
},"aggregations": {
"subitem_name": {
"terms": {
"field": "items.subitems.name"
}
},"my_rev_agg": {
"reverse_nested": {}
}
}
}
}
}
}
}
这将返回正确排序的子项桶：
{
"took": 0,"timed_out": false,"_shards": {
"total": 5,"successful": 5,"skipped": 0,"failed": 0
},"hits": {
"total": 9,"max_score": 0.0,"hits": []
},"aggregations": {
"my_nested_agg": {
"doc_count": 19,"subitem_id": {
"doc_count_error_upper_bound": 0,"sum_other_doc_count": 0,"buckets": [
{
"key": 3,"doc_count": 6,"my_rev_agg": {
"doc_count": 6
},"subitem_name": {
"doc_count_error_upper_bound": 0,"buckets": [
{
"key": "Subitem #3","doc_count": 6
}
]
}
},{
"key": 2,"doc_count": 11,"my_rev_agg": {
"doc_count": 5
},"buckets": [
{
"key": "Subitem #2","doc_count": 11
}
]
}
},{
"key": 1,"doc_count": 1,"my_rev_agg": {
"doc_count": 1
},"buckets": [
{
"key": "Subitem #1","doc_count": 1
}
]
}
},{
"key": 5,"buckets": [
{
"key": "Subitem #5","doc_count": 1
}
]
}
}
]
}
}
}
}