Question

我有一个名为profile的表，其中包含用户信息。我需要对此表进行过滤查询并获取：

与此查询匹配的行数。
仅限前5000个匹配行的数据。

我正在寻找一种最佳方法。显然，至少需要进行一次扫描来进行计数，但理想情况下，DB可以在进行计数时获取最高匹配。

以下查询给出了正确的结果，但看起来有点hacky。我想知道它能否做得更好？

WITH total AS (
     SELECT COUNT(*) AS total FROM profile 
         WHERE project_id = 1 and some_prop = 100)
SELECT total.total, full_name, other_prop 
    FROM profile
    INNER JOIN total ON 1 = 1
    WHERE project_id = 1 and some_prop = 100
    ORDER BY full_name ASC
    LIMIT 5000

有更有效的方法吗？

Answer 1

您正在扫描同一个表两次以应用过滤器。使用下面的代码，您只需在应用过滤器后扫描表格，然后执行总计并在过滤后的表格中列出两者。

with s as (
    select *
    from profile 
    where project_id = 1 and some_prop = 100
), t as (
    select count(*) as total from s
)
select total, full_name, other_prop 
from s cross join t
order by full_name asc
limit 5000

窗口功能版

select
    count(*) over() as total,
    full_name,
    other_prop 
from profile
where project_id = 1 and some_prop = 100
order by full_name asc
limit 5000

将LIMITed SELECT和总COUNT组合在一起？

1 个答案: