跨行的 SQL DISTINCT 值

Question

我在列中有重复的agreementnumber 和重复的telephone，我想在列中获得唯一的agreementnumber，它对应的唯一的telephone。

我在 SQL 中编写了查询，它给了我唯一的 agreementnumber，但行中的 telephone 是重复的，但我想要唯一的电话号码。

代码：

select agreementnumber,

  max(case when rn = 1 then telephone end) telephone1,

  max(case when rn = 2 then telephone end) telephone2,

  max(case when rn = 3 then telephone end) telephone3,

  max(case when rn = 4 then telephone end) telephone4,

  max(case when rn = 5 then telephone end) telephone5

from
(
  select agreementnumber, telephone,

    row_number() over(partition by agreementnumber order by telephone) rn
  from alternate_mobile 

) src
group by agreementnumber;

我想要以下输出。 col1 和 col2,col3,col4,col4 中的唯一值。

col1 col2 col3 col4`` AGMTNO phone1 phone2 phone3

Answer 1

尝试在您的查询中进行这个小改动：

select agreementnumber,
  max(case when rn = 1 then telephone end) telephone1,
  max(case when rn = 2 then telephone end) telephone2,   
  max(case when rn = 3 then telephone end) telephone3,
  max(case when rn = 4 then telephone end) telephone4,
  max(case when rn = 5 then telephone end) telephone5

from
(
  select x.*,
    row_number() over(partition by x.agreementnumber order by x.telephone) rn
  from (
    select distinct agreementnumber, telephone
    from alternate_mobile 
  ) x
) src
group by agreementnumber;

如果您收到重复电话是因为您在 alternate_mobile 表中重复了 agreementnumber/telephone。

编辑：

我将查询更改为只保留电话中的数字，删除所有其余字符：

select agreementnumber,
  max(case when rn = 1 then telephone end) telephone1,
  max(case when rn = 2 then telephone end) telephone2,   
  max(case when rn = 3 then telephone end) telephone3,
  max(case when rn = 4 then telephone end) telephone4,
  max(case when rn = 5 then telephone end) telephone5

from
(
  select x.*,
    row_number() over(partition by x.agreementnumber order by x.telephone) rn
  from (
    select distinct agreementnumber, regexp_replace(telephone,'[^0-9]', '') as telephone
    from alternate_mobile 
  ) x
) src
group by agreementnumber;

Answer 2

请注意，您可以使用rank() 而不是row_number() 来减少子查询的数量：

select agreementnumber,
       max(case when rn = 1 then telephone end) as telephone1,
       max(case when rn = 2 then telephone end) as telephone2,   
       max(case when rn = 3 then telephone end) as telephone3,
       max(case when rn = 4 then telephone end) as telephone4,
       max(case when rn = 5 then telephone end) as telephone5
from (select am.*,
             rank() over (partition by am.agreementnumber order by am.telephone) as rn
      from alternate_mobile am
     ) am
group by agreementnumber;