按列名分組的SELECT distinct行

Question

我有一個場景，我需要選擇員工的最新ID。該ID可以是一個varchar或INT

id  mail 
------------- 
1234 xyz@abc.com 
5678 xyz@abc.com 
C123 xyz@abc.com 
C234 def@abc.com 

現在，我的查詢結果只是必須有

id  mail 
------------- 
5678 xyz@abc.com 
C234 def@abc.com 

的邏輯是，如果沒有針對郵件ID的整數，最高整數必須被挑選。如果沒有整數，則可以選取varchar值。

試圖

select max(id), mail from worker group by mail 

但如果有重複的郵件ID則會選取VARCHAR值。

Answer 1

我會使用isnumeric值和ROW_NUMBER來選擇正確的值。

DECLARE @T TABLE (id VARCHAR(10), mail VARCHAR(100)) 
INSERT @T VALUES 
    ('1234', 'xyz@abc.com'), 
    ('5678', 'xyz@abc.com'), 
    ('C123', 'xyz@abc.com'), 
    ('C234', 'def@abc.com') 

SELECT id, mail 
FROM (
    SELECT 
     id, 
     mail, 
     ROW_NUMBER() OVER (
      PARTITION BY 
       mail 
      ORDER BY 
       CASE WHEN ISNUMERIC(id) = 1 THEN CONVERT(INT, id) ELSE 0 END DESC, 
       id) AS RowNum 
    FROM @T 
    ) T 
WHERE RowNum = 1 

輸出產量

id   mail 
---------- ------------- 
C234  def@abc.com 
5678  xyz@abc.com 

Answer 2

一個解決方案，我能想到的是：

select MAX(t.id), t.email 
from ( 
    select * from Emails where ISNUMERIC(id) = 1 
     union 
    select * from Emails where ISNUMERIC(id) <> 1 and email not in 
           (select email from Emails where ISNUMERIC(id) = 1) 
) as t 
group by t.email