SELECT 有多个 JOIN，包括 LISTAGG 和它自己的 JOIN

Question

我有以下 2 个查询，当运行分开时它们可以完美地工作。

查询 1

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 OPTION_1 = 'Y' THEN 'Option 1'
        WHEN OPTION_2 = 'Y' THEN 'Option 2'
        WHEN OPTION_3 = 'Y' THEN 'Option 3'
        WHEN OPTION_4 = 'Y' THEN 'Option 4'
        WHEN OPTION_5 = 'Y' THEN 'Option 5'
        WHEN OPTION_6 = 'Y' THEN 'Option 6'
        WHEN OPTION_7 = 'Y' THEN 'Option 7'
        WHEN OPTION_8 = 'Y' THEN 'Option 8'
        WHEN OPTION_9 = 'Y' THEN 'Option 9' 
        WHEN OPTION_0 = 'Y' THEN 'Not Applicable'
        ELSE ''
   END ONR,       
   abs.ABSTRACT_DETAILS
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';

结果

--------
PROPOSAL_NUMBER    TITLE    FULL_NAME    NRP_IREF_CD ... <br />
     2028          blah       ...           ....         <br />
     2029          blah2      ...           ....         <br />
     2030          blah3      ...           ....         <br />

查询 2

SELECT eps.PROPOSAL_NUMBER,
   LISTAGG(sk.DESCRIPTION, ', ') WITHIN GROUP (ORDER BY sk.DESCRIPTION) AS KEYWORDS 
FROM EPS_PROPOSAL eps
LEFT JOIN EPS_PROP_SCIENCE_KEYWORD key 
    ON eps.PROPOSAL_NUMBER = key.PROPOSAL_NUMBER
LEFT JOIN SCIENCE_KEYWORD sk
    ON key.SCIENCE_KEYWORD_CODE = sk.SCIENCE_KEYWORD_CODE
GROUP BY eps.PROPOSAL_NUMBER;

结果

--------
PROPOSAL_NUMBER    KEYWORDS
     2028          Keyword_A, Keyword_B, Keyword_C, Keyword_D
     2029          Keyword_Y, Keyword_Z
     2030          Keyword_W

所以你可以看到我正在处理的主要 table 是 EPS_PROPOSAL 而 PK 是 PROPOSAL_NUMBER。还有其他 table 将 PK 用作 FK，例如 EPS_PROP_SCIENCE_KEYWORD，您可以在其中有多行关键字映射到相同的 PROPOSAL_NUMBER。问题是只有 ID 或代码存储在 table 上，因此需要加入 SCIENCE_KEYWORD table 才能获得描述。

问题：我如何组合这 2 个查询，以便我的结果如下所示：

Results
--------
PROPOSAL_NUMBER    TITLE    FULL_NAME    NRP_IREF_CD ...  KEYWORDS
     2028          blah       ...           ....           Keyword_A, Keyword_B, Keyword_C, Keyword_D
     2029          blah2      ...           ....           Keyword_Y, Keyword_Z
     2030          blah3      ...           ....           Keyword_W

Answer 1

最简单的方法是进行小的修改 - 添加带有子查询的 OUTER APPLY：

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 OPTION_1 = 'Y' THEN 'Option 1'
        WHEN OPTION_2 = 'Y' THEN 'Option 2'
        WHEN OPTION_3 = 'Y' THEN 'Option 3'
        WHEN OPTION_4 = 'Y' THEN 'Option 4'
        WHEN OPTION_5 = 'Y' THEN 'Option 5'
        WHEN OPTION_6 = 'Y' THEN 'Option 6'
        WHEN OPTION_7 = 'Y' THEN 'Option 7'
        WHEN OPTION_8 = 'Y' THEN 'Option 8'
        WHEN OPTION_9 = 'Y' THEN 'Option 9' 
        WHEN OPTION_0 = 'Y' THEN 'Not Applicable'
        ELSE ''
   END ONR,       
   abs.ABSTRACT_DETAILS,
   k.keywords
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    
OUTER APPLY(
   SELECT 
      LISTAGG(sk.DESCRIPTION, ', ') WITHIN GROUP (ORDER BY sk.DESCRIPTION) AS KEYWORDS 
   FROM EPS_PROP_SCIENCE_KEYWORD key 
   LEFT JOIN SCIENCE_KEYWORD sk
       ON key.SCIENCE_KEYWORD_CODE = sk.SCIENCE_KEYWORD_CODE
   WHERE eps.PROPOSAL_NUMBER = key.PROPOSAL_NUMBER
) k
WHERE eps.OWNED_BY_UNIT = '401' AND eps.requested_start_date_initial >= DATE '2019-10-01';

Answer 2

一个选项是使用通用 Table 表达式：

WITH Q1 AS
(
  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 OPTION_1 = 'Y' THEN 'Option 1'
              WHEN OPTION_2 = 'Y' THEN 'Option 2'
              WHEN OPTION_3 = 'Y' THEN 'Option 3'
              WHEN OPTION_4 = 'Y' THEN 'Option 4'
              WHEN OPTION_5 = 'Y' THEN 'Option 5'
              WHEN OPTION_6 = 'Y' THEN 'Option 6'
              WHEN OPTION_7 = 'Y' THEN 'Option 7'
              WHEN OPTION_8 = 'Y' THEN 'Option 8'
              WHEN OPTION_9 = 'Y' THEN 'Option 9' 
              WHEN OPTION_0 = 'Y' THEN 'Not Applicable'
              ELSE ''
         END ONR,       
         abs.ABSTRACT_DETAILS
    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' 
), Q2 AS
(
  SELECT eps.PROPOSAL_NUMBER,
         LISTAGG(sk.DESCRIPTION, ', ') WITHIN GROUP(ORDER BY sk.DESCRIPTION) AS KEYWORDS
    FROM EPS_PROPOSAL eps
    LEFT JOIN EPS_PROP_SCIENCE_KEYWORD key
      ON eps.PROPOSAL_NUMBER = key.PROPOSAL_NUMBER
    LEFT JOIN SCIENCE_KEYWORD sk
      ON key.SCIENCE_KEYWORD_CODE = sk.SCIENCE_KEYWORD_CODE
   GROUP BY eps.PROPOSAL_NUMBER
)
SELECT Q1.*, Q2.KEYWORDS
  FROM Q1
  JOIN Q2
    ON Q2.PROPOSAL_NUMBER = Q1.PROPOSAL_NUMBER

SELECT 有多个 JOIN，包括 LISTAGG 和它自己的 JOIN

SELECT with multiple JOINS Including LISTAGG with it's own JOIN

oracle

group-by

left-join

listagg