基于公共列值从同一个表中聚合一对唯一的列值答案

【问题标题】：Aggregating a unique pair of column values from the same table based on a common column value基于公共列值从同一个表中聚合一对唯一的列值
【发布时间】：2014-07-26 00:42:16
【问题描述】：

我有下表：

my_table
------------------------
| common_id | uniq_val |
------------------------
|    1      |    foo   |          
------------------------
|    1      |    bar   |          
------------------------

我想从中聚合值，使得结果查询看起来像：

DESIRED RESULT
---------------------------------------
| common_id | uniq_val_1 | uniq_val_2 |
---------------------------------------
|    1      |     foo    |      bar   |
---------------------------------------

OR

---------------------------------------
| common_id | uniq_val_1 | uniq_val_2 |
---------------------------------------
|    1      |     bar    |      foo   |
---------------------------------------

所以我写了查询：

SELECT t1.common_id, t1.uniq_val, t2.uniq_val
FROM my_table t1 JOIN my_table AS t2 
ON t1.common_id=t2.common_id 
WHERE t1.uniq_val!=t2.uniq_val;

结果

RESULTING SELECT
---------------------------------------
| common_id | uniq_val_1 | uniq_val_2 |
---------------------------------------
|    1      |     foo    |      foo   |
---------------------------------------
|    1      |     bar    |      bar   |
---------------------------------------

但我只需要其中一列，所以我应该能够执行 GROUP BY t1.common_id，例如：

SELECT t1.common_id, t1.uniq_val, t2.uniq_val
FROM my_table t1 JOIN my_table AS t2 
ON t1.common_id=t2.common_id 
WHERE t1.uniq_val!=t2.uniq_val
GROUP BY t1.common_id;

不幸的是，这会返回错误：

ERROR:  column "t1.uniq_val" must appear in the GROUP BY clause or be used in an aggregate function

谁能指出我的逻辑错误？

【问题讨论】：

每个 common_id 值是否总是只有 2 个唯一值？有的有 2 个，有的有 3 个或 4 个吗？

标签： sql postgresql join group-by

【解决方案1】：

简单聚合怎么样？

select common_id, min(uniq_val) as uniq_val_1, max(uniq_val) as uniq_val_2
from my_table
group by common_id;

【讨论】：

【解决方案2】：

你可以试试distinct on

SELECT distinct on (t1.common_id) t1.common_id, t1.uniq_val, t2.uniq_val FROM my_table t1 JOIN my_table AS t2 ON t1.common_id=t2.common_id WHERE t1.uniq_val!=t2.uniq_val;

我认为它会产生你需要的东西！

【讨论】：

给出错误：在“，”或附近出现语法错误，你确定这适用于 postgres 吗？
我刚刚使用了您的查询，您说的那个有效并添加了distinct on (t1.common_id)。请注意，此字段与您选择的第一个字段 (t1.common_id) 之间没有逗号！

【解决方案3】：

这将处理每个 common_id 最多 10 个 uniq_val 值。如果需要，您可以删除或添加更少或更多的 uniq_val 值。

在此处查看具有不同计数的 uniq_val 值的 common_id 值的演示： http://sqlfiddle.com/#!15/e2c87/1/0

select common_id,
       max(case when rn = 1 then uniq_val else null end) as uniq_val_1,
       max(case when rn = 2 then uniq_val else null end) as uniq_val_2,
       max(case when rn = 3 then uniq_val else null end) as uniq_val_3,
       max(case when rn = 4 then uniq_val else null end) as uniq_val_4,
       max(case when rn = 5 then uniq_val else null end) as uniq_val_5,
       max(case when rn = 6 then uniq_val else null end) as uniq_val_6,
       max(case when rn = 7 then uniq_val else null end) as uniq_val_7,
       max(case when rn = 8 then uniq_val else null end) as uniq_val_8,
       max(case when rn = 9 then uniq_val else null end) as uniq_val_9,
       max(case when rn = 10 then uniq_val else null end) as uniq_val_10
from(
select row_number() over (partition by common_id order by common_id, uniq_val) as rn,
       common_id,
       uniq_val
  from my_table
 order by common_id, uniq_val) x
 group by common_id
 order by common_id

【讨论】：