如何从 N1QL 查询中的嵌套数组中提取最大值和最小值？

Question

背景

我正在运行以下查询（我知道它可能很难阅读，所以我会在之后澄清它在做什么）：

SELECT *
FROM `my_bucket` AS a
    NEST (
    SELECT c.toNode, c.fromNode, d.endDateTime, d.startDateTime
    FROM `my_bucket` AS c
        JOIN (
        SELECT id, customAttributes.endDateTime, customAttributes.startDateTime
        FROM `my_bucket`
        WHERE type='airNode'
            AND customAttributes.endDateTime >= CLOCK_UTC()
            AND customAttributes.endDateTime <= DATE_ADD_STR(CLOCK_UTC(), 14, 'day')) AS d ON c.toNode = d.id
        AND c.type='relNode') b ON a.id = b.fromNode AND a.type='edNode';

本质上，这里发生的事情是我们从数据库中提取相关的 airNodes，并将它们与 relNodes.toNode = airNodes.id 的 relNodes 连接起来（toNode 和 id 是来自它们各自节点的字段）。然后，我们将这些 relAirNodes 与 edNodes 连接起来，语句返回以下结构：

[ { 'edNode':..., 'relAirNodes':[{...},{...}] },...]

所以本质上，我们得到一个对数组，其中每一对由一个 edNode 和一个 relAirNode 数组组成。

问题

relAirNodes 数组中的每个 relAirNode 都有一个 endTime 和 startTime 字段。有没有办法运行查询，使得返回结果是一个对数组，对的第一个元素是一个 editNode，对的第二个元素是一个单一的结构，数组的最小 startTime relAirNodes 和 relAirNodes 数组的最大 endTime？ 换句话说，返回对象如下所示：

[ { 'edNode':..., 'relAirNodes':{'startTime':..., 'endTime':...}}, ...]

Answer 1

Couchbase JOINS 从左到右。如果您只查看 INNER JOINS，则只需重新排列 JOIN，以便更早地消除。

创建此处描述的索引https://blog.couchbase.com/ansi-join-support-n1ql/

SELECT en.*, c AS relAirNodes
FROM (SELECT rn.fromNode, MIN(b.startDateTime) AS startDateTime, MAX(b.endDateTime) AS endDateTime
      FROM (SELECT an.id, an.customAttributes.endDateTime, an.customAttributes.startDateTime
            FROM my_bucket AS an
            WHERE an.type = "airNode"
                  AND an.customAttributes.endDateTime BETWEEN CLOCK_UTC() AND DATE_ADD_STR(CLOCK_UTC(), 14, "day")
            ) AS b
      JOIN my_bucket AS rn ON rn.type = "relNode" AND rn.toNode = b.id
      GROUP BY rn.fromNode) AS c
JOIN my_bucket AS en ON en.type = "edNode" AND en.id = c.fromNode
;

CREATE INDEX ix1 ON my_bucket(customAttributes.endDateTime, customAttributes.startDateTime, id) WHERE type = "airNode";
CREATE INDEX ix2 ON my_bucket(toNode, fromNode) WHERE type = "relNode";
CREATE INDEX ix3 ON my_bucket(id) WHERE type = "edNode";

如果需要外部连接（需要 CB 6.5.0+ https://blog.couchbase.com/ansi-join-enhancements-and-ansi-merge/）

SELECT en.*, c AS relAirNodes
FROM my_bucket AS en
LEFT JOIN (SELECT rn.fromNode, MIN(b.startDateTime) AS startDateTime, MAX(b.endDateTime) AS endDateTime
           FROM (SELECT an.id, an.customAttributes.endDateTime, an.customAttributes.startDateTime
                 FROM my_bucket AS an
                 WHERE an.type = "airNode"
                       AND an.customAttributes.endDateTime BETWEEN CLOCK_UTC() AND DATE_ADD_STR(CLOCK_UTC(), 14, "day")
            ) AS b
           JOIN my_bucket AS rn ON rn.type = "relNode" AND rn.toNode = b.id
           GROUP BY rn.fromNode) AS c ON en.id = c.fromNode
WHERE en.type = "edNode";

还有 ARRAY_MIN()、ARRAY_MAX() 函数。例如：对象数组上的 ARRAY_MIN(relAirNodes[*].startDateTime)。