如何基於其他列的值條件

Question

我在我的桌子下面的表中的數據

SQL> select * from mate2; 

    A_CD  S_ID  E_CD S_E S_NE E_IND 
---------- ---------- ---------- ----- ----- - 
     200   1   2 A  J  Y 
     200   2   3 B  A  N 
     200   3   1 C  D 
     200   4   2 A  C  Y 
     200   5   1 D  C 
     200   6   3 B  N  Y 
     200   7   3 N  K 

我需要a_cd和（S_E，S_NE），以獲得不同的S_ID組的數量爲顯示的列數用下面條件

1）取S_NE字母代碼時E_CD = 2或（E_CD = 3 AND E_IND = N）

2）取S_E當E_CD = 1或（E_CD = 3 AND E_IND = Y ）

3）取兩者S_E和S_NE當E_CD = 3和E_IND爲空

輸出應該像

200 A 3
200 B 2
200℃2
..
.. 。
我寫了這樣的事情，但它不工作

select a.a_cd, a.letter_code, COUNT (DISTINCT a.s_id) AS cnt 
      FROM (SELECT m.a_cd, 
         m.s_id, 
         CASE 
          WHEN (m.e_cd = 1) 
           OR (m.e_cd = 3 AND m.e_ind = 'Y') 
          THEN 
           m.S_E 
          WHEN (m.e_cd = 2) OR (m.e_cd = 3 AND m.e_ind = 'N') then m.S_NE 
          WHEN m.e_ind is null then m.S_NE      
         END 
          AS letter_code from mate2 m 
        union 
        SELECT m.a_cd, 
         m.s_id, 
         CASE 
          WHEN (m.e_cd = 1) 
           OR (m.e_cd = 3 AND m.e_ind = 'Y') 
          THEN 
           m.S_E 
          WHEN (m.e_cd = 2) OR (m.e_cd = 3 AND m.e_ind = 'N') then m.S_NE 
          WHEN m.e_ind is null then m.S_NE      
         END 
          AS letter_code from mate2 m) a group by a.ai_cd,a.letter_code 

Answer 1

我有點困惑你的groupping指標分析。您的示例輸出也不符合您的選擇標準。

但是，下面您會發現一個查詢應該根據您的規則選擇正確的中間結果集。只要根據這個做出正確的分組。

with 
mate2( a_cd, s_id, e_cd, s_e, s_ne, e_ind) as ( 
    select 200, 1, 2, 'A', 'J', 'Y' from dual union all 
    select 200, 2, 3, 'B', 'A', 'N' from dual union all 
    select 200, 3, 1, 'C', 'D', null from dual union all 
    select 200, 4, 2, 'A', 'C', 'Y' from dual union all 
    select 200, 5, 1, 'D', 'C', null from dual union all 
    select 200, 6, 3, 'B', 'N', 'Y' from dual union all 
    select 200, 7, 3, 'N', 'K', null from dual 
), 
rule_1(rule, a_cd, s_id, letter_code) as (
    -- take the S_NE when E_CD = 2 or (E_CD = 3 AND E_IND = N) 
    select 1, a_cd, s_id, s_ne from mate2 where e_cd = 2 or (e_cd = 3 and e_ind = 'N') 
), 
rule_2(rule, a_cd, s_id, letter_code) as (
    -- take the S_E when E_CD = 1 or (E_CD = 3 AND E_IND = Y) 
    select 2, a_cd, s_id, s_e from mate2 where e_cd = 1 or (e_cd = 3 and e_ind = 'Y') 
), 
rule_3(rule, a_cd, s_id, letter_code) as (
    -- take BOTH the S_E and S_NE when E_CD = 3 and E_IND is null 
    select 3, a_cd, s_id, letter_code from mate2 
    unpivot (letter_code for x in (s_ne as '1', s_e as '2')) 
    where e_cd = 3 and e_ind is null 
) 
-- group the way you like 
--select letter_code, count(*) from (
    select * from rule_1 
    union all 
    select * from rule_2 
    union all 
    select * from rule_3 
--) group by letter_code 
--order by 1 
;