限制父表的has-many关系答案

【问题标题】：Limit a has-many relationship by the parent table限制父表的has-many关系
【发布时间】：2020-08-28 03:41:56
【问题描述】：

使用 Postgres 9.6，user has_many posts，其中帖子可以是 public 或 private。我需要返回加入他们帖子的用户列表其中用户拥有 > 0 个公共帖子，并根据 user 限制返回的记录，同时还限制加入 posts 的数量。

作为一个表示为模拟 ORM 的示例：

// User A has 4 public posts
// User B has 0 public posts
// User C has 1 public post
// User D has 5 public posts

// I am limiting the posts fetched per-user to 3 and the users fetched to 2:

[
  {
    user: "A",
    posts: [
      { id: 10 },
      { id: 28 },
      { id: 33 }
      // Fourth post omitted since we are limiting per-user to 3
    ]
  },
  // B is skipped since they have no posts
  {
    user: "C",
    posts: [
      { id: 45 }
    ]
  }
  // D is skipped because we are limiting total users to 2
]

我能够让它按预期工作，除了能够通过此查询限制每个用户返回的帖子：

SELECT *
FROM users
LEFT OUTER JOIN posts
ON posts.user_id = users.id
WHERE users.id IN (
    SELECT users.id
    FROM users
    LEFT OUTER JOIN posts
    ON posts.user_id = users.id AND posts.visibility = "public"
    GROUP BY users.id
    HAVING COUNT(posts.id) > 0
    LIMIT 1 -- Limits users fetched regardless of how many posts, but doesn't limit posts fetched themselves
)

但这感觉不必要的复杂，甚至不能正确解决问题。由于我们使用的是 Postgres，因此我尝试了一些涉及 LEFT OUTER JOIN LATERAL 的尝试，这似乎很有希望限制每个用户的帖子，但在设置返回用户的限制时让我感到困惑。

【问题讨论】：

LIMITing 没有ORDERing 很少有意义。
会有订单，但我试图将其保留为最小示例

标签： sql postgresql

【解决方案1】：

您可以这样做（假设posts 有一个名为id 的列）：

select *
from (
  select *, row_number(partition by p.user_id order by p.id) as rn
  from users u
  join posts p on p.user_id = u.id
  where u.user_id in (
    select user_id
    from (select distinct user_id from posts where visibility = 'public') x
    order by user_id
    limit 2 -- limits users to 2 (with public posts)
  )
) y
where rn <= 3 -- limits posts per user to 3

当帖子数量庞大且其中大部分是公共帖子时，可以通过使用横向表格表达式来提高性能。

【讨论】：