AWS Redshift SQL - PIVOT 查询（一行/行多次计数）

Question

我正在尝试为 Amazon Redshift 中的多个状态计数生成一条记录（每个账户）。基本上，我想要一个数据透视表。但是我不喜欢这种做法，也不知道有没有更好的办法。

这就是我正在做的：

Select g1.account,
       sum(case when g1.status = 'PASS' then 1 else 0 end) as status_cnt_pass,
       sum(case when g1.status = 'NEW' then 1 else 0 end) as cnt_new,
       sum(case when g1.status = 'FAIL' then 1 else 0 end) as cnt_fail,
       sum(case when g1.status = 'TEST' then 1 else 0 end) as cnt_test,
       sum(case when coalesce(TRIM(g1.status ), '') != '' and g1.status not in('PASS', 'NEW', 'FAIL','TEST') then 1 else 0 end) as cnt_other,
       sum(case when coalesce(TRIM(g1.status ), '') = ''  then 1 else 0 end) as cnt_none

输出：

account cnt_new cnt_fail cnt_test cnt_other cnt_none
foo     41      3        16       2         0
bar     105     17       5        1         1

有没有更好的方法来计算 other 值而不维护列表？

不在('PASS', 'NEW', 'FAIL','TEST')

此外，如果先前的条件为真或 status = null （或空白），我想停止检查其他状态值。

我不拥有此表，也无法更改数据的填充方式。

Answer 1

我认为没有办法。但是你可以使用:: 来简化你的语法：

select g1.account,
       sum( (g1.status = 'PASS')::int ) as status_cnt_pass,
       sum( (g1.status = 'NEW')::int ) as cnt_new,
       sum( (g1.status = 'FAIL')::int ) as cnt_fail,
       sum( (g1.status = 'TEST')::int ) as cnt_test,
       sum( trim(g1.status) not in ('', 'PASS', 'NEW', 'FAIL', 'TEST')::int ) as cnt_other,
       sum( (coalesce(trim(g1.status ), '') = '')::int ) as cnt_none