如何使用 group by 检索不同的数据

Question

我有以下 table 及其示例数据，

+------------------+-------+--------+---------------------+-------+
| PhoneCode        | SeqID | Active | Token               | CUSID |
+------------------+-------+--------+---------------------+-------+
| e29a5e1c695352b8 | 3898  | I      | JYN8CYRVzaKWL-l_K   | 50002 |
+------------------+-------+--------+---------------------+-------+
| e29a5e1c695352b8 | 3900  | I      | JYN8CYRVzaKWL-l_K   | 50002 |
+------------------+-------+--------+---------------------+-------+
| 741fb28bc72183e3 | 3899  | I      | RU3ReKEw0yin9LxZWCO | 50002 |
+------------------+-------+--------+---------------------+-------+
| 741fb28bc72183e4 | 3901  | A      | RU3ReKEw0yin9LxZWCO | 50002 |
+------------------+-------+--------+---------------------+-------+

我需要将 PhoneCode 与最新的 SeqID 区分开来。所以我尝试了以下查询。

SELECT UD.PHONECODE, UD.SeqID, UD.ACTIVE, UD.Token
FROM   DEVICE UD
WHERE UD.CUSID = '50002' AND UD.ACTIVE = 'I'
GROUP BY PHONECODE

但是它出错了。如何检索以下输出？

+------------------+-------+--------+---------------------+-------+
| PhoneCode        | SeqID | Active | Token               | CUSID |
+------------------+-------+--------+---------------------+-------+
| e29a5e1c695352b8 | 3900  | I      | JYN8CYRVzaKWL-l_K   | 50002 |
+------------------+-------+--------+---------------------+-------+
| 741fb28bc72183e3 | 3899  | I      | RU3ReKEw0yin9LxZWCO | 50002 |
+------------------+-------+--------+---------------------+-------+

Answer 1

您可以使用 MAX() 和一个连接：

SELECT UD.PHONECODE, UD.ACTIVE, UD.Token, UD.SeqID
FROM   DEVICE UD
INNER JOIN (
    SELECT PHONECODE, MAX(SeqID) as SEQID
    FROM DEVICE   
    WHERE CUSID = '50002' AND ACTIVE = 'I'
    GROUP BY PHONECODE
 ) T2 ON UD.PHONECODE = T2.PHONECODE AND UD.SEQID = T2.SEQID

Answer 2

没有 GROUP BY 和 self-joins 也可以：

SELECT PHONECODE,
       SeqID,
       ACTIVE,
       Token
FROM   (
  SELECT d.*,
         ROW_NUMBER() OVER (PARTITION BY phonecode ORDER BY seqid DESC) AS rn
  FROM   device d
  WHERE  cusid = 50002
  AND    active = 'I'
)
WHERE  rn = 1;

或者，如果你想使用 GROUP BY 那么你也可以使用 KEEP:

SELECT PHONECODE,
       MAX(SeqID) AS seqid,
       MAX(ACTIVE) KEEP (DENSE_RANK LAST ORDER BY SeqID) AS active,
       MAX(Token) KEEP (DENSE_RANK LAST ORDER BY SeqID) AS token
FROM   device
WHERE  cusid = 50002
AND    active = 'I'
GROUP BY phonecode;

其中，对于示例数据：

CREATE TABLE device (PhoneCode, SeqID, Active, Token, CUSID ) AS
SELECT 'e29a5e1c695352b8', 3898, 'I', 'JYN8CYRVzaKWL-l_K',   50002 FROM DUAL UNION ALL
SELECT 'e29a5e1c695352b8', 3900, 'I', 'JYN8CYRVzaKWL-l_K',   50002 FROM DUAL UNION ALL
SELECT '741fb28bc72183e3', 3899, 'I', 'RU3ReKEw0yin9LxZWCO', 50002 FROM DUAL UNION ALL
SELECT '741fb28bc72183e4', 3901, 'A', 'RU3ReKEw0yin9LxZWCO', 50002 FROM DUAL;

双输出：

PHONECODE SEQID ACTIVE TOKEN

741fb28bc72183e3 3899 I RU3ReKEw0yin9LxZWCO

e29a5e1c695352b8 3900 I JYN8CYRVzaKWL-l_K

db<>fiddle here

Answer 3

简单的子查询怎么样

SELECT UD.PHONECODE, UD.SeqID, UD.ACTIVE, UD.Token 
FROM device ud
WHERE (UD.PHONECODE, UD.SeqID) IN (
  SELECT PHONECODE, MAX(SeqID) 
  FROM device
  WHERE CUSID = '50002' AND ACTIVE = 'I'
  GROUP BY PHONECODE 
)

Answer 4

以下查询将起作用：

select phonecode,seqid,active,token,cusid
from
(
    select * , row_number() over(partition by phonecode order by seqid desc) as rownum
    from device
    where cusid = 50002 and active = 'I'
) as t1
where rownum = 1

Answer 5

如果您的数据库版本是 12c+ 并且显示 ROW_NUMBER() 函数的返回值无关紧要，那么没有子查询的作为选项的查询将是

SELECT d.*,
       ROW_NUMBER() OVER (PARTITION BY PhoneCode ORDER BY SeqID DESC) AS rn
  FROM device d
 WHERE cusid = 50002
   AND active = 'I'
 ORDER BY rn  
 FETCH FIRST 1 ROWS WITH TIES

其中 rn 的所有关系（相等值）都包含在结果集中。

Demo

如何使用 group by 检索不同的数据

How can I retrieve distinct data by using group by

sql

oracle

greatest-n-per-group

PHONECODE	SEQID	ACTIVE	TOKEN
741fb28bc72183e3	3899	I	RU3ReKEw0yin9LxZWCO
e29a5e1c695352b8	3900	I	JYN8CYRVzaKWL-l_K