Elasticsearch：获取 nested 类型数组中的所有元素

在我之前的文章 “Elasticsearch: object 及 nested 数据类型” 对 nested 数据类型做了一个比较详细的介绍。在实际使用中，你在构建查询时肯定会遇到一些问题。根据官方文档介绍，nested 类型字段在隐藏数组中索引其每个项目，这允许独立于索引文档搜索这些项目。今天我带来一个问题：问如何获得一个文档，其整个 item 列表必须与一个子句匹配。 让我举个例子让你更清楚。

我有一个 item 字段设置为 nested 的索引。 我索引了两个文档，其中一个所有项目的 status 都为 active，另一个则不全是。

PUT nested_index
{"mappings": {"properties": {"description": {"type": "text"},"item": {"type": "nested","properties": {"value": {"type": "text"},"status": {"type": "keyword"}}}}}
}

POST nested_index/_doc
{"description": "nested A","item": [{"value": "a","status": "active"},{"value": "ab","status": "active"}]
}POST nested_index/_doc
{"description": "nested B","item": [{"value": "b","status": "inactive"},{"value": "bc","status": "active"}]
}

如上所示，第一个文档里的 status 都是 active 的状态，而第二个文档的其中一个 status 状态是 active，另外一个是 inactive 状态。

现在我们想要查询的问题是：搜索到 status 都是 active 的文档。

也许你会想到使用如下的查询：

GET nested_index/_search
{"query": {"nested": {"path": "item","query": {"term": {"item.status": {"value": "active"}}}}}
}

上面命令查询的结果是：

{"took": 1,"timed_out": false,"_shards": {"total": 1,"successful": 1,"skipped": 0,"failed": 0},"hits": {"total": {"value": 2,"relation": "eq"},"max_score": 0.35667494,"hits": [{"_index": "nested_index","_id": "xRoaRIYB2XodIZsbUfzi","_score": 0.35667494,"_source": {"description": "nested A","item": [{"value": "a","status": "active"},{"value": "ab","status": "active"}]}},{"_index": "nested_index","_id": "xhoaRIYB2XodIZsbWvzm","_score": 0.35667494,"_source": {"description": "nested B","item": [{"value": "b","status": "inactive"},{"value": "bc","status": "active"}]}}]}
}

显然两个文档都被查询到了。这个不是我们想要的结果。

解决它的方法是使用两个 must_not 子句。 最里面的 must_not 子句将过滤文档 A，因为所有项目都具有 active 状态。 最外层的 must_not 将反转操作，迄今为止省略的文档 A 将可用，而文档 B 将不可用且不会出现在响应中。

GET nested_index/_search
{"query": {"bool": {"must_not": [{"nested": {"path": "item","query": {"bool": {"must_not": [{"term": {"item.status": {"value": "active"}}}]}}}}]}}
}

上面的命令结果为：

{"took": 2,"timed_out": false,"_shards": {"total": 1,"successful": 1,"skipped": 0,"failed": 0},"hits": {"total": {"value": 1,"relation": "eq"},"max_score": 0,"hits": [{"_index": "nested_index","_id": "xRoaRIYB2XodIZsbUfzi","_score": 0,"_source": {"description": "nested A","item": [{"value": "a","status": "active"},{"value": "ab","status": "active"}]}}]}
}

这次显然只有 nested A 文档被搜索到。