在一个查询中为每篇文章选择文章 + 多条评论

Question

我的 postgresql 数据库中有 2 个表。简化后，它们看起来像这样：

// table: articles
+----+-----------+---------------+
| id | title     | body          |
+----+-----------+---------------+
| 1  | hello     | world         |
| ...                            |
+----+-----------+---------------+

// table: comments
+----+-----------+---------------+
| id | articleid | comment       |
+----+-----------+---------------+
| 1  | 1         | looks cool!   |
| ...                            |
+----+-----------+---------------+

我想制作一个包含所有文章的概览页面 + 每篇文章的最后 3 个 cmets。这可以通过一个查询来实现吗？

我的查询现在看起来像这样（nodeJS）：

let query, values, result;

let ret = [];

// query to select all articles, ordered by ID desc
query = 'SELECT id, title FROM articles ORDER BY id DESC';
result = await postgresql.query(query, []);

// store all rows in a constant
const articles = result.rows;

// loop over each article
articles.forEach(article => {
  // fetch the last 3 comments for each article
  query = 'SELECT id, comment FROM comments WHERE articleid = $1 ORDER BY id DESC LIMIT 3';
  values = [article.id];
  result = await postgresql.query(query, values);

  // store the comments in a constant
  const comments = result.rows;

  // push the article information (ID + title) and the comments in the ret array
  ret.push({
    articleID: article.id,
    title: article.title,
    comments: comments
  });
});

// ret now contains an array of objects of articles + comments
return ret;

我想知道是否可以将两个查询合并为 1，或（如果不可能），这是执行此操作的最佳性能方式。

请注意，这两个表都比我在此处描述的简化版本要大。目前我有 + 1,700 篇文章和 +100,000 个 cmets。

Answer 1

一个选项使用横向连接：

select a.id, a.title, c.id, c.comment
from articles a
cross join lateral (
    select c.*
    from comments c
    where c.articleid = a.id
    order by id desc
    limit 3
) c
order by a.id desc, c.id desc

Answer 2

我相信，一个简单的：

select id, comment from 
(select a.id, c.comment, dense_rank() over PARTITION BY a.id ORDER BY c.id DESC) as rn from articles a join comments c on a.id = c.article_id) t
where t.rn <= 3;

所以是的，你只需要使用“DENSE_RANK()”函数https://www.postgresqltutorial.com/postgresql-dense_rank-function/