我有以下数据:
SUM_OF_ALL_PUSHES | USER_LOGIN | COMPONENT_NAME | ROW_LST_UPD_TS
-------------------|---------------|-------------------|-------------------------
61 | DOMAIN\abc | Component A | 22/02/12 12:58:26.325 PM
14 | DOMAIN\xyz | Component B | 22/02/12 05:20:52.565 PM
17 | DOMAIN\xyz | Component A | 22/02/12 05:21:58.045 PM
34 | DOMAIN\abc | Component A | 22/02/12 05:44:20.274 PM
38 | DOMAIN\abc | Component A | 22/02/12 06:28:41.465 PM
我使用以下方法导出数据:
SELECT SUM(Q.SUM_OF_ALL_PUSHES) AS SUM_OF_ALL_PUSHES, Q.USER_LOGIN, Q.COMPONENT_NAME, Q.ROW_LST_UPD_TS FROM( SELECT
SUM(REGEXP_SUBSTR(SUBSTR(EVENT_MSG, INSTR(EVENT_MSG, ' ', 1,1), INSTR(EVENT_MSG, ' ', 1,1)) , '[0-9]+')) AS "SUM_OF_ALL_PUSHES",
USER_LOGIN,
COMPONENT_NAME,
EVENT_MSG,
ROW_LST_UPD_TS
FROM EVENT_MGT.EVENT_LOG
WHERE
ROW_LST_UPD_TS BETWEEN TRUNC(SYSDATE - 1) AND TRUNC(SYSDATE) - 1/86400
AND SUBSTR(EVENT_MSG,1 ,INSTR(EVENT_MSG, 'd', 1,1)) = 'Received'
GROUP BY
USER_LOGIN,
COMPONENT_NAME,
EVENT_MSG,
ROW_LST_UPD_TS) Q
GROUP BY COMPONENT_NAME, USER_LOGIN, ROW_LST_UPD_TS
ORDER BY ROW_LST_UPD_TS ASC
问题是我需要按组件对每个用户的数字求和。因此,在上面的结果集中,我需要聚合第1,4和5列,并将SUM_OF_ALL_PUSHES的值相加,因为每个记录都有用户abc和组件A.
总而言之,我需要删除其他列中的数据重复,并根据用户和组件对第一列中的值求和。所以结果集应如下所示:
SUM_OF_ALL_PUSHES | USER_LOGIN | COMPONENT_NAME | ROW_LST_UPD_TS
-------------------|---------------|-------------------|-------------------------
133 | DOMAIN\abc | Component A | 22/02/12 12:58:26.325 PM
14 | DOMAIN\xyz | Component B | 22/02/12 05:20:52.565 PM
17 | DOMAIN\xyz | Component A | 22/02/12 05:21:58.045 PM
请帮忙。
答案 0 :(得分:1)
听起来你不想按时间戳分组,而是将聚合函数(我猜是MAX
)应用于它。
SELECT SUM(Q.SUM_OF_ALL_PUSHES) AS SUM_OF_ALL_PUSHES, Q.USER_LOGIN, Q.COMPONENT_NAME, MAX(Q.ROW_LST_UPD_TS) FROM( SELECT
SUM(REGEXP_SUBSTR(SUBSTR(EVENT_MSG, INSTR(EVENT_MSG, ' ', 1,1), INSTR(EVENT_MSG, ' ', 1,1)) , '[0-9]+')) AS "SUM_OF_ALL_PUSHES",
USER_LOGIN,
COMPONENT_NAME,
EVENT_MSG,
ROW_LST_UPD_TS
FROM EVENT_MGT.EVENT_LOG
WHERE
ROW_LST_UPD_TS BETWEEN TRUNC(SYSDATE - 1) AND TRUNC(SYSDATE) - 1/86400
AND SUBSTR(EVENT_MSG,1 ,INSTR(EVENT_MSG, 'd', 1,1)) = 'Received'
GROUP BY
USER_LOGIN,
COMPONENT_NAME,
EVENT_MSG,
ROW_LST_UPD_TS) Q
GROUP BY COMPONENT_NAME, USER_LOGIN
ORDER BY MAX(Q.ROW_LST_UPD_TS) ASC