0
我有以下映射:ElasticSearch:从嵌套聚集查询中访问外文档字段
{
"dynamic": "strict",
"properties": {
"id": {
"type": "string"
},
"title": {
"type": "string"
},
"things": {
"type": "nested",
"properties": {
"id": {
"type": "long"
},
"something": {
"type": "long"
}
}
}
}
}
我插入文档如下(Python脚本):
body = {"id": 1, "title": "one", "things": [{"id": 1000, "something": 33}, {"id": 1001, "something": 34}, ]}
es.create(index_name, doc_type=doc_type, body=body, id=1)
body = {"id": 2, "title": "two", "things": [{"id": 1000, "something": 43}, {"id": 1001, "something": 44}, ]}
es.create(index_name, doc_type=doc_type, body=body, id=2)
body = {"id": 3, "title": "three", "things": [{"id": 1000, "something": 53}, {"id": 1001, "something": 54}, ]}
es.create(index_name, doc_type=doc_type, body=body, id=3)
我运行以下聚合查询:
{
"query": {
"match_all": {}
},
"aggs": {
"things": {
"aggs": {
"num_articles": {
"terms": {
"field": "things.id",
"size": 0
},
"aggs": {
"articles": {
"top_hits": {
"size": 50
}
}
}
}
},
"nested": {
"path": "things"
}
}
},
"size": 0
}
(所以,我要计算每个“事物”出现的次数,并对每个事物列出一个列表在其中出现每个事物的文章)的
查询生成:
"key": 1000,
"doc_count": 3,
"articles": {
"hits": {
"total": 3,
"max_score": 1,
"hits": [{
"_index": "test",
"_type": "article",
"_id": "2",
"_nested": {
"field": "things",
"offset": 0
},
"_score": 1,
"_source": {
"id": 1000,
"something": 43
}
}, {
"_index": "test",
"_type": "article",
"_id": "1",
"_nested": {
"field": "things",
"offset": 0
},
"_score": 1,
"_source": {
"id": 1000,
"something": 33
}
......(依此类推)
我想什么是每个命中列出所有来自“外部”或顶级文档的字段,即在这种情况下是id和标题。
这实际上是可能的.....如果是这样如何?
非常近... .. –
问题是''''reverse_things'''部分列出了ID和标题,但不是以相同的顺序。所以,密钥对ID是1,2,3 “ID”:{ “doc_count_error_upper_bound”:0, “sum_other_doc_count”:0, “桶”:[{ “钥匙”: “1”, “doc_count”:1 },{ “钥匙”: “2”, “doc_count”:1 },{ “键”: “3”, “doc_count”:1 }] }, –
但是标题的关键是一,三,二。 “标题”:{ “doc_count_error_upper_bound”:0, “sum_other_doc_count”:0, “桶”:[{ “键”: “一个”, “doc_count”:1 },{ “键“: “三化”, “doc_count”:1 },{ “钥匙”: “两节”, “doc_count”:1 }] } 如果排序可能会被迫以配合原创文章,这将工作。谢谢@ kristian-ferkić顺便说一句... –