SPLIT_PART 为负值 [Postgres 9.5]

SPLIT_PART with a negative value [Postgres 9.5]

我需要对该查询使用 split_part 函数:

CREATE TABLE client_group_by_group_test AS SELECT *, SPLIT_PART( groupe,
 ',', 1 ) AS group1, SPLIT_PART(SPLIT_PART(groupe,',',2),',',-1) AS 
 group2, SPLIT_PART(SPLIT_PART(groupe,'',3),'',-3) AS group3, 
 SPLIT_PART(groupe,'',-4) AS group4 FROM planification_client

但它给了我以下错误:

ERROR: field position must be greater than zero

那么,我该如何处理这里的负值?

这种reverse(split_part(reverse(col_A), '_'::text, 1))的说法行得通吗?我指的是 that question.

编辑:我完全被这个查询困住了。

更多细节:我有一列 "server name" 和另一列不同的组用逗号分隔。

server name| group                        |
-----------+------------------------------+
XPTERTBIEP9|GRNW_SPO_S_F_H, GRNW_SPO_S_I_J|

我需要得到的输出是如果服务器有多个组,它们需要在不同的列中,如 group1、group2...

server name| group                        |group1        |group 2
-----------+------------------------------+--------------+--------------
XPTERTBIEP9|GRNW_SPO_S_F_H, GRNW_SPO_S_I_J|GRNW_SPO_S_F_H|GRNW_SPO_S_I_J

如果负数应该表示距末尾的偏移量,则两步法可能更好:

CREATE TABLE client_group_by_group_test 
AS 
SELECT ..., 
       agroups[1] as group1,
       agroups2[cardinality(agroups2) - 1] as groups2,
       agroups3[cardinality(agroups3) - 3] as groups3,
       agroups[cardinality(agroups) - 4] as group4
from (
   select *,
          string_to_array(groupe, ',') as agroups, 
          (string_to_array(string_to_array(groupe, ','), ',')[2]) as agroups2,
          (string_to_array(string_to_array(groupe, ','), ',')[3]) as agroups3,
   from planification_client
) t

请注意,您需要在最外面的 SELECT 中列出所需的列,以排除中间的 "agroups" 列。