有没有办法在这个 MYSQL 查询中找到总和和累积计数？

Question

我有这个 MYSQL 查询，我用它来查找表groups在其关联表members中的累积增长。

  SELECT DISTINCT g.name, m.created_at,
           COUNT(*) OVER(PARTITION by g.id ORDER BY m.created_at) count
    FROM members m
    INNER JOIN groups g on g.id = m.group_id

这给出了一个结果集，如：

[
  { 'group01', date: '2019-03-04', count: 5 },
  { 'group01', date: '2019-03-09', count: 12 }
  { 'group01', date: '2019-03-15', count: 23 }
  { 'group02', date: '2019-03-04', count: 3 }
  { 'group01', date: '2019-03-10', count: 19 }
  { 'group01', date: '2019-03-17', count: 27 }
]

我还将检索组成员的总数。 这样数据集将如下所示：

[
  { name: 'group01', date: '2019-03-04', count: 5, total: 23 },
  { name: 'group01', date: '2019-03-09', count: 12, total: 23 },
  { name: 'group01', date: '2019-03-15', count: 23, total: 23 },
  { name: 'group02', date: '2019-03-04', count: 3, total: 27 },
  { name: 'group02', date: '2019-03-10', count: 19, total: 27 },
  { name: 'group02', date: '2019-03-17', count: 27, total: 27 }
]

我需要这样做，以便稍后我可以轻松订购并找到成员最多的组。

有没有办法做到这一点？

Answer 1

只需在结果集中添加另一列，即对同一partition进行 window 计数，但没有order by子句； 这会导致数据库计算分区中的总行数：

SELECT g.name, m.created_at,
       COUNT(*) OVER(PARTITION by g.id ORDER BY m.created_at) count,
       COUNT(*) OVER(PARTITION by g.id) total
FROM members m
INNER JOIN groups g on g.id = m.group_id

注意：我对在SELECT子句中使用DISTINCT持怀疑态度，因此我将其删除（尽管存在极端情况，但将其与 window 函数一起使用几乎没有意义） - 如果需要，您可以将其添加回来。

Answer 2

我会用显式聚合来写这个：

SELECT g.name, m.created_at,
       SUM(COUNT(*)) OVER (PARTITION BY g.id ORDER BY m.created_at) as count_on_day,
       SUM(COUNT(*)) OVER (PARTITION BY g.id) as count_for_group
FROM members m INNER JOIN
     groups g 
     ON g.id = m.group_id
GROUP BY g.name, m.created_at