Oracle LISTAGG和Coalesce

问题描述

我已经整理了下表EPS_PROPOSAL_EXT_T：

+--------------------+------------+------------+
| PROPOSAL_NUMBER    |ONR_OPTION_1|ONR_Option_2|   ...
+--------------------+------------+------------+
|        1234        |      N     |      N     |   ...
|        1235        |      Y     |      Y     |   ...
|        1236        |      N     |      Y     |   ... 
+--------------------+------------+------------+

这里的问题是我需要对所有这些不同的ONR_Options进行LISTAGG，但是我不能让它们显示为'Y'。我需要将'Y'的内容替换为人类可以理解的内容。请参阅下面的SELECT语句，我已尝试这样做：

SELECT eps.PROPOSAL_NUMBER,eps.TITLE,per.FULL_NAME,ext.NRP_IREF_CD,ext.RESEARCH_TYPE_CD,nsf.NSF_CODE,ext.NPS_THRUST_DESCRIPTION,CASE
        WHEN ext.ONR_Option_1 = 'Y' THEN 'Option 1'
        WHEN ext.ONR_Option_2 = 'Y' THEN 'Option 2'
        WHEN ext.ONR_Option_3 = 'Y' THEN 'Option 3'
        WHEN ext.ONR_Option_4 = 'Y' THEN 'Option 4'
        WHEN ext.ONR_Option_5 = 'Y' THEN 'Option 5'
        WHEN ext.ONR_Option_6 = 'Y' THEN 'Option 6'
        WHEN ext.ONR_Option_7 = 'Y' THEN 'Option 7'
        WHEN ext.ONR_Option_8 = 'Y' THEN 'Option 8'
        WHEN ext.ONR_Option_9 = 'Y' THEN 'Option 9' 
        WHEN ext.ONR_NOT_APPLICABLE = 'Y' THEN 'Not Applicable'
        ELSE ''
   END ONR
FROM EPS_PROPOSAL eps
LEFT JOIN EPS_PROPOSAL_EXT_T ext
    ON eps.PROPOSAL_NUMBER = ext.PROPOSAL_NUMBER
LEFT JOIN EPS_PROP_PERSON per
    ON eps.PROPOSAL_NUMBER = per.PROPOSAL_NUMBER AND
      (per.PROP_PERSON_ROLE_ID = 'PI' OR per.PROP_PERSON_ROLE_ID = 'PD')
LEFT JOIN EPS_PROP_ABSTRACT abs
    ON eps.PROPOSAL_NUMBER = abs.PROPOSAL_NUMBER
LEFT JOIN NSF_CODES nsf
    ON eps.NSF_CODE = nsf.NSF_SEQUENCE_NUMBER
WHERE eps.OWNED_BY_UNIT = '401' AND eps.requested_start_date_initial >= DATE '2019-10-01';

这可行，但是您可以看到问题所在，对吧？我设置CASE语句的方式，当一个特定的PROPOSAL_NUMBER包含多个TRUE的ONR_Option时，将不会考虑它。

在上面给出的示例中，如果我搜索编号为1235的提案， ONR的结果应为-“选项1，选项2”。

如何使用LISTAGG和COALESCE解决这个难题？还是解决该问题的方法？

解决方法

LISTAGG在这里没有帮助。为什么？因为您的数据模型错误。我认为应该是这样的：

SQL> with test (prop,onr,status) as
  2    (select 1234,'1','N' from dual union all
  3     select 1234,'2','N' from dual union all
  4     select 1234,'3','Y' from dual union all
  5     select 1235,'Y' from dual union all
  6     select 1235,'Y' from dual union all
  7     select 1235,'N' from dual
  8    )
  9  select prop,10    listagg(case when status = 'Y' then 'Option ' || onr end,',')
 11      within group (order by onr) as result
 12  from test
 13  group by prop;

      PROP RESULT
---------- ------------------------------
      1234 Option 3
      1235 Option 1,Option 2

SQL>

---

目前的方式，看看是否如此-串联的CASE列与表中的ONR_OPTION列一样多，可以像您想教的山羊一样缩放，加上删除多个逗号的痛苦-帮助：

SQL> with test (prop,onr_1,onr_2,onr_3) as
  2    (select 1234,'N','Y' from dual union all
  3     select 1235,'Y','N' from dual
  4    )
  5  select prop,6    trim(both ',' from
  7          case when onr_1 = 'Y' then 'Option 1' else null end ||','||
  8          case when onr_2 = 'Y' then 'Option 2' else null end ||','||
  9          case when onr_3 = 'Y' then 'Option 3' else null end
 10         ) as result
 11  from test;

      PROP RESULT
---------- ------------------------------
      1234 Option 3
      1235 Option 1,Option 2

SQL>