Question

如何计算具有特定值的所有行的总长度？

我们假设有下表：

 id | unit_id | length | column to be filled with total length
 1  |   1     | 10
 2  |   1     |  4
 3  |   1     |  5
 4  |   2     |  3
 5  |   3     |  3
 6  |   3     |  6

在这种情况下，如何更新表，使unit_id为1的所有行都具有unit_id为1（10 + 4 + 5 = 19）的所有行长度的总和，然后是两行有一个unit_id为3有9。

我试过

update test.routes 
set total_length = (select sum(length) from test.routes where unit_id = unit_id) where unit_id = unit_id

但它的作用是它只是用相同的值更新整个表，如何更新每个unit_id的正确总和？

Answer 1

尝试CTE：

t=# with a as (select *, sum(length) over (partition by unit_id) from routes)
t-# update routes u set total_length = a.sum
t-# from a
t-# where a.id = u.id;
UPDATE 6
Time: 0.520 ms
t=# select * from routes ;
 id | unit_id | length | total_length
----+---------+--------+--------------
  1 |       1 |     10 |           19
  2 |       1 |      4 |           19
  3 |       1 |      5 |           19
  4 |       2 |      3 |            3
  5 |       3 |      3 |            3
  6 |       4 |      6 |            6
(6 rows)

Answer 2

您需要限定对属性unit_id的引用。否则，像where unit_id = unit_id这样的约束（除了空值之外）始终为真，因此将总结所有内容：

update test.routes r1 set total_length = (select sum(length) from test.routes r2 where r2.unit_id = r1.unit_id)

Answer 3

这应该做的工作。

update 
    routes as s 
inner join (
    select unit_id, sum(length) as total_length from routes group by unit_id
) as g 
set 
    s.total_length = g.total_length 
where 
    s.unit_id = g.unit_id

这里我们创建一个临时表，其中unit_id的总长度。通过使用2个表之间的连接，我们可以有效地使用子查询

如何在Postgresql中对具有共同字段的列求和？

3 个答案: