在 postgresql 中重用子查询答案

【问题标题】：reusing subquery in postgresql在 postgresql 中重用子查询
【发布时间】：2022-01-24 18:07:38
【问题描述】：

我有一个运行良好的查询，但是我在其中两次使用了相同的子查询。我想知道是否有办法重用这个子查询。这是查询：

SELECT DISTINCT homeworks.*
FROM homeworks
INNER JOIN homework_messages ON homeworks.id = homework_messages.homework_id
WHERE homework_messages.message_type = 'submit'
  AND homework_messages.created_at::date <= (CURRENT_DATE - '5 days'::interval)
  AND (
    (
      SELECT MAX(homework_messages.created_at) FROM homework_messages
      WHERE homework_messages.homework_id = homeworks.id AND homework_messages.message_type != 'submit'
    ) < homework_messages.created_at 
    OR 
    (
      SELECT MAX(homework_messages.created_at) FROM homework_messages
      WHERE homework_messages.homework_id = homeworks.id AND homework_messages.message_type != 'submit'
    ) IS NULL
  )
GROUP BY homeworks.id

如您所见，我将这些行重复了两次：

(
  SELECT MAX(homework_messages.created_at) FROM homework_messages
  WHERE homework_messages.homework_id = homeworks.id AND homework_messages.message_type != 'submit'
)

我尝试使用 WITH 子句，但它似乎不适合这个问题，或者我用错了。不管怎样，谢谢你的帮助。

【问题讨论】：

你可以使用 PL/PgSQL 函数吗？
如果您检查执行计划，我敢打赌该语句实际上只执行了一次；考虑在横向连接中使用它
无关，但 distinct as 使用的是代码异味。您可以通过将内部连接转换为 exists 条件来摆脱这种情况。
homeworks.id 可能是主键，否则 PG 会产生函数依赖错误。这也表明由于GROUP BY homeworks.id 而不需要DISTINCT。

标签： sql postgresql subquery

【解决方案1】：

这是一个cross join lateral 的建议。我没有深入研究您的业务逻辑，只是试图保持其等效性。

SELECT DISTINCT homeworks.*
FROM homeworks
INNER JOIN homework_messages ON homeworks.id = homework_messages.homework_id
cross join lateral ( -- your subquery follows
  SELECT MAX(created_at) as created_at FROM homework_messages
  WHERE homework_id = homeworks.id AND message_type != 'submit'
) as lat
WHERE homework_messages.message_type = 'submit'
  AND homework_messages.created_at::date <= (CURRENT_DATE - '5 days'::interval)
  AND (lat.created_at < homework_messages.created_at OR lat.created_at IS NULL)
GROUP BY homeworks.id;

【讨论】：