Запрос SQL LISTAGG не создает объединенную строку

Question

Я пытаюсь создать объединенную строку значений на основе строк с использованием Oracle LISTAGG.Вот упрощенный код, который все еще показывает проблему, которую я пытаюсь решить.

SELECT S_SUBJECT.UNIQRECNUM,                      
(SELECT LISTAGG(S_STUDY.U_PRIMARYRESEARCHCOHORT,',') WITHIN GROUP ( 
     ORDER BY U_PRIMARYRESEARCHCOHORT)
     FROM S_STUDY
     WHERE S_STUDY.S_STUDYID = S_PARTICIPANT.SSTUDYID) COHORTS

FROM S_SUBJECT
JOIN S_PARTICIPANT ON S_PARTICIPANT.SUBJECTID = S_SUBJECT.S_SUBJECTID
WHERE S_SUBJECT.UNIQRECNUM IN ('9','8','7','6','5','2');


These are the results of this query;
UNIQRECNUM | COHORTS
---------------------
    2      | Gastro
    5      | Metabolic
    6      | Cardio
    7      | Cardio
    8      | Gastro
    8      | Cardio
    9      | Gastro
    9      | Gastro
    9      | Gastro


What I am expecting to see from LISTAGG would be
UNIQRECNUM  | COHORTS
--------------------
    2       | Gastro
    5       | Metabolic
    6       | Cardio
    7       | Cardio
    8       | Cardio,Gastro
    8       | Cardio,Gastro
    9       | Gastro,Gastro,Gastro
    9       | Gastro,Gastro,Gastro
    9       | Gastro,Gastro,Gastro

Answer 1

Я думаю, что вы хотите функцию окна:

SELECT su.UNIQRECNUM,                      
       LISTAGG(st.U_PRIMARYRESEARCHCOHORT, ',') WITHIN GROUP
                   (ORDER BY U_PRIMARYRESEARCHCOHORT) OVER (PARTITION BY su.UNIQRECNUM) as COHORTS
FROM S_SUBJECT su JOIN
     S_PARTICIPANT p
     ON p.SUBJECTID = su.S_SUBJECTID JOIN
     S_STUDY st
     ON st.S_STUDYID = p.SSTUDYID
WHERE su.UNIQRECNUM IN ('9', '8', '7', '6', '5', '2');

Answer 2

Попробуйте:

Пример данных:

UNIQRECNUM COHORTS  
---------- ---------
         2 Metabolic
         5 Cardio   
         6 Cardio   
         7 Gastro   
         8 Gastro   
         8 Cardio   
         9 Gastro   
         9 Gastro   
         9 Gastro   

WITH T AS 
(
    SELECT 2 AS  UNIQRECNUM,'Metabolic' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 5 AS  UNIQRECNUM,'Cardio' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 6 AS  UNIQRECNUM,'Cardio' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 7 AS  UNIQRECNUM,'Gastro' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 8 AS  UNIQRECNUM,'Gastro' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 8 AS  UNIQRECNUM,'Cardio' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 9 AS  UNIQRECNUM,'Gastro' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 9 AS  UNIQRECNUM,'Gastro' AS COHORTS FROM DUAL
    UNION ALL
    SELECT 9 AS  UNIQRECNUM,'Gastro' AS COHORTS FROM DUAL
)
SELECT UNIQRECNUM,
     LISTAGG(COHORTS,',') WITHIN GROUP ( 
     ORDER BY COHORTS) OVER (PARTITION BY UNIQRECNUM) AS COHORTS 
FROM T;          

Вывод:

UNIQRECNUM COHORTS                                                                                             
---------- ----------------------
         2 Metabolic                                                                                           
         5 Cardio                                                                                              
         6 Cardio                                                                                              
         7 Gastro                                                                                              
         8 Cardio,Gastro                                                                                       
         8 Cardio,Gastro                                                                                       
         9 Gastro,Gastro,Gastro                                                                                
         9 Gastro,Gastro,Gastro                                                                                
         9 Gastro,Gastro,Gastro              

Answer 3

Я считаю, что ваш запрос будет выглядеть так:

SELECT S_SUBJECT.UNIQRECNUM,                      
  LISTAGG(S_STUDY.U_PRIMARYRESEARCHCOHORT,',') WITHIN GROUP ( 
     ORDER BY U_PRIMARYRESEARCHCOHORT) COHORTS
FROM S_SUBJECT
JOIN S_PARTICIPANT ON S_PARTICIPANT.SUBJECTID = S_SUBJECT.S_SUBJECTID
JOIN S_STUDY ON S_STUDY.S_STUDYID = S_PARTICIPANT.SSTUDYID
WHERE S_SUBJECT.UNIQRECNUM IN ('9','8','7','6','5','2')
GROUP BY S_SUBJECT.UNIQRECNUM;

Answer 4

Почему бы не

select u.uniqrecnum,
  listagg(t.u_primarysearchcohort, ',') 
    within group (order by t.u_primarysearchcohort) cohorts
from s_subject u join s_study t on t.s_studyid = p.sstudyid
join s_participant p on p.subjectid = u.s_subjectid
where u.uniqrecnum in ('9','8','7','6','5','2')
group by u.uniqrecnum;