删除按日期排序的组内的重复行
我怀疑我的头衔非常有道理，我会尽力解释我的要求。我需要清理一个审计表，该表跟踪对象的状态何时被修改。出于这样或那样的原因，在对象的状态仍然相同的情况下，使用新日期创建多个记录。我需要保留每个状态更改的第一个记录，然后删除状态相同的任何后续记录。哦，没有主键。是啊！ ：|
以下是一个示例数据集：
ObjectID ObjectState DateOfEntry
101144 1 2007-08-14 12:39:30.587
101144 1 2007-08-14 12:41:52.620
101144 1 2007-08-14 12:42:11.150
101144 1 2007-08-14 12:42:24.197
101144 3 2007-08-14 12:44:06.403
101144 3 2007-08-14 12:44:06.467
101144 3 2007-08-14 12:46:12.573
101144 3 2007-08-14 12:50:51.670
101144 3 2007-08-14 12:50:51.750
101144 3 2007-08-14 12:56:34.330
101144 4 2007-08-14 17:28:59.280
101144 3 2007-08-14 17:32:26.313
101144 3 2007-08-14 17:32:48.720
101144 3 2007-08-14 17:45:07.460
101144 3 2007-08-14 17:46:31.740
101144 3 2007-08-14 17:47:04.380
101144 3 2007-08-14 17:47:29.507
101144 3 2007-08-14 17:49:13.460
101144 3 2007-08-14 17:54:15.320
101144 3 2007-08-14 17:55:57.540
101144 3 2007-08-14 19:50:11.913
101144 3 2007-08-14 19:53:10.820
101144 3 2007-08-14 20:03:44.900
101144 3 2007-08-16 10:34:56.477
101144 3 2007-08-16 10:36:06.477
101144 3 2007-08-16 10:36:24.570
101144 3 2007-11-06 09:19:26.157
101144 3 2007-11-06 09:24:28.200
101144 4 2010-09-27 14:11:03.287
101144 4 2014-01-27 17:31:58.077
结束表结果应为：
ObjectID ObjectState DateOfEntry
101144 1 2007-08-14 12:39:30.587
101144 3 2007-08-14 12:44:06.403
101144 4 2007-08-14 17:28:59.280
101144 3 2007-08-14 17:32:26.313
101144 4 2010-09-27 14:11:03.287
我尝试过使用RANK()，但问题是我不能只对ObjectState进行排序，因为ObjectState值可能无序重复。我必须通过DateOfEntry订购它们。但如果我做RANK() OVER(ORDER BY DateOfEntry)那么我基本上会得到行号。
如何创建一个SQL查询，允许我按DateOfEntry排序，然后按ObjectState分组，这样我就可以删除该“对象状态组”中的所有行，除了最小的一个小组？
3 个答案:
答案 0 :(得分：3)
简短回答：
; WITH Records AS (
SELECT
ObjectId,
ObjectState,
DateOfEntry,
ROW_NUMBER() OVER (PARTITION BY ObjectID ORDER BY DateOfEntry) AS RowNum
FROM @Audits
)
DELETE R2
FROM Records R1
INNER JOIN Records R2
ON R1.ObjectId = R2.ObjectId
AND R1.ObjectState = R2.ObjectState
AND R1.RowNum + 1 = R2.RowNum
解决方案证明
DECLARE @Audits TABLE (ObjectID INT, ObjectState INT, DateOfEntry DATETIME)
INSERT @Audits
SELECT 101144,1,'2007-08-14 12:39:30.587' UNION ALL
SELECT 101144,1,'2007-08-14 12:41:52.620' UNION ALL
SELECT 101144,1,'2007-08-14 12:42:11.150' UNION ALL
SELECT 101144,1,'2007-08-14 12:42:24.197' UNION ALL
SELECT 101144,3,'2007-08-14 12:44:06.403' UNION ALL
SELECT 101144,3,'2007-08-14 12:44:06.467' UNION ALL
SELECT 101144,3,'2007-08-14 12:46:12.573' UNION ALL
SELECT 101144,3,'2007-08-14 12:50:51.670' UNION ALL
SELECT 101144,3,'2007-08-14 12:50:51.750' UNION ALL
SELECT 101144,3,'2007-08-14 12:56:34.330' UNION ALL
SELECT 101144,4,'2007-08-14 17:28:59.280' UNION ALL
SELECT 101144,3,'2007-08-14 17:32:26.313' UNION ALL
SELECT 101144,3,'2007-08-14 17:32:48.720' UNION ALL
SELECT 101144,3,'2007-08-14 17:45:07.460' UNION ALL
SELECT 101144,3,'2007-08-14 17:46:31.740' UNION ALL
SELECT 101144,3,'2007-08-14 17:47:04.380' UNION ALL
SELECT 101144,3,'2007-08-14 17:47:29.507' UNION ALL
SELECT 101144,3,'2007-08-14 17:49:13.460' UNION ALL
SELECT 101144,3,'2007-08-14 17:54:15.320' UNION ALL
SELECT 101144,3,'2007-08-14 17:55:57.540' UNION ALL
SELECT 101144,3,'2007-08-14 19:50:11.913' UNION ALL
SELECT 101144,3,'2007-08-14 19:53:10.820' UNION ALL
SELECT 101144,3,'2007-08-14 20:03:44.900' UNION ALL
SELECT 101144,3,'2007-08-16 10:34:56.477' UNION ALL
SELECT 101144,3,'2007-08-16 10:36:06.477' UNION ALL
SELECT 101144,3,'2007-08-16 10:36:24.570' UNION ALL
SELECT 101144,3,'2007-11-06 09:19:26.157' UNION ALL
SELECT 101144,3,'2007-11-06 09:24:28.200' UNION ALL
SELECT 101144,4,'2010-09-27 14:11:03.287' UNION ALL
SELECT 101144,4,'2014-01-27 17:31:58.077'
; WITH Records AS (
SELECT
ObjectId,
ObjectState,
DateOfEntry,
ROW_NUMBER() OVER (PARTITION BY ObjectID ORDER BY DateOfEntry) AS RowNum
FROM @Audits
)
DELETE R2
FROM Records R1
INNER JOIN Records R2
ON R1.ObjectId = R2.ObjectId
AND R1.ObjectState = R2.ObjectState
AND R1.RowNum + 1 = R2.RowNum
SELECT * FROM @Audits
产生此输出
ObjectID ObjectState DateOfEntry
----------- ----------- -----------------------
101144 1 2007-08-14 12:39:30.587
101144 3 2007-08-14 12:44:06.403
101144 4 2007-08-14 17:28:59.280
101144 3 2007-08-14 17:32:26.313
101144 4 2010-09-27 14:11:03.287
答案 1 :(得分：0)
如果@table是你的表，可能是下面的sql会帮助你。我假设DateOfEntry是排序的。
DELETE B
FROM (SELECT *,ROW_NUMBER() OVER(ORDER BY DateOfEntry) [ROW] FROM @table) A
LEFT JOIN (SELECT *,ROW_NUMBER() OVER(ORDER BY DateOfEntry) [ROW] FROM @table) B
ON A.[Row] = B.[Row] - 1
AND ABS(a.ObjectState - b.ObjectState) = 0
我之前和之后的结果
注意：请不要介意DateOfEntry列中的数据。为方便起见，我保留了它的号码。
答案 2 :(得分：0)
我最终想到使用LAG（）的替代解决方案将消除CTE。
DELETE @Audits
FROM @Audits a1
INNER JOIN (SELECT ObjectID, DateOfEntry
FROM (SELECT ObjectID, DateOfEntry, ObjectState,
LAG(ObjectState) OVER(PARTITION BY ObjectID ORDER BY DateOfEntry) AS [PreviousObjectState]
FROM @Audits) AS Audits
WHERE Audits.ObjectState = PreviousObjectState
) a2
ON a2.ObjectID = a1.ObjectID AND a2.DateOfEntry = a1.DateOfEntry
SELECT * FROM @Audits
带证明的长版本（我已经使用不同的ID复制数据集以验证分区是否按预期工作）
DECLARE @Audits TABLE (ObjectID INT, ObjectState INT, DateOfEntry DATETIME)
INSERT @Audits
SELECT 101144,1,'2007-08-14 12:39:30.587' UNION ALL
SELECT 101144,1,'2007-08-14 12:41:52.620' UNION ALL
SELECT 101144,1,'2007-08-14 12:42:11.150' UNION ALL
SELECT 101144,1,'2007-08-14 12:42:24.197' UNION ALL
SELECT 101144,3,'2007-08-14 12:44:06.403' UNION ALL
SELECT 101144,3,'2007-08-14 12:44:06.467' UNION ALL
SELECT 101144,3,'2007-08-14 12:46:12.573' UNION ALL
SELECT 101144,3,'2007-08-14 12:50:51.670' UNION ALL
SELECT 101144,3,'2007-08-14 12:50:51.750' UNION ALL
SELECT 101144,3,'2007-08-14 12:56:34.330' UNION ALL
SELECT 101144,4,'2007-08-14 17:28:59.280' UNION ALL
SELECT 101144,3,'2007-08-14 17:32:26.313' UNION ALL
SELECT 101144,3,'2007-08-14 17:32:48.720' UNION ALL
SELECT 101144,3,'2007-08-14 17:45:07.460' UNION ALL
SELECT 101144,3,'2007-08-14 17:46:31.740' UNION ALL
SELECT 101144,3,'2007-08-14 17:47:04.380' UNION ALL
SELECT 101144,3,'2007-08-14 17:47:29.507' UNION ALL
SELECT 101144,3,'2007-08-14 17:49:13.460' UNION ALL
SELECT 101144,3,'2007-08-14 17:54:15.320' UNION ALL
SELECT 101144,3,'2007-08-14 17:55:57.540' UNION ALL
SELECT 101144,3,'2007-08-14 19:50:11.913' UNION ALL
SELECT 101144,3,'2007-08-14 19:53:10.820' UNION ALL
SELECT 101144,3,'2007-08-14 20:03:44.900' UNION ALL
SELECT 101144,3,'2007-08-16 10:34:56.477' UNION ALL
SELECT 101144,3,'2007-08-16 10:36:06.477' UNION ALL
SELECT 101144,3,'2007-08-16 10:36:24.570' UNION ALL
SELECT 101144,3,'2007-11-06 09:19:26.157' UNION ALL
SELECT 101144,3,'2007-11-06 09:24:28.200' UNION ALL
SELECT 101144,4,'2010-09-27 14:11:03.287' UNION ALL
SELECT 101144,4,'2014-01-27 17:31:58.077' UNION ALL
SELECT 101145,1,'2007-08-14 12:39:30.587' UNION ALL
SELECT 101145,1,'2007-08-14 12:41:52.620' UNION ALL
SELECT 101145,1,'2007-08-14 12:42:11.150' UNION ALL
SELECT 101145,1,'2007-08-14 12:42:24.197' UNION ALL
SELECT 101145,3,'2007-08-14 12:44:06.403' UNION ALL
SELECT 101145,3,'2007-08-14 12:44:06.467' UNION ALL
SELECT 101145,3,'2007-08-14 12:46:12.573' UNION ALL
SELECT 101145,3,'2007-08-14 12:50:51.670' UNION ALL
SELECT 101145,3,'2007-08-14 12:50:51.750' UNION ALL
SELECT 101145,3,'2007-08-14 12:56:34.330' UNION ALL
SELECT 101145,4,'2007-08-14 17:28:59.280' UNION ALL
SELECT 101145,3,'2007-08-14 17:32:26.313' UNION ALL
SELECT 101145,3,'2007-08-14 17:32:48.720' UNION ALL
SELECT 101145,3,'2007-08-14 17:45:07.460' UNION ALL
SELECT 101145,3,'2007-08-14 17:46:31.740' UNION ALL
SELECT 101145,3,'2007-08-14 17:47:04.380' UNION ALL
SELECT 101145,3,'2007-08-14 17:47:29.507' UNION ALL
SELECT 101145,3,'2007-08-14 17:49:13.460' UNION ALL
SELECT 101145,3,'2007-08-14 17:54:15.320' UNION ALL
SELECT 101145,3,'2007-08-14 17:55:57.540' UNION ALL
SELECT 101145,3,'2007-08-14 19:50:11.913' UNION ALL
SELECT 101145,3,'2007-08-14 19:53:10.820' UNION ALL
SELECT 101145,3,'2007-08-14 20:03:44.900' UNION ALL
SELECT 101145,3,'2007-08-16 10:34:56.477' UNION ALL
SELECT 101145,3,'2007-08-16 10:36:06.477' UNION ALL
SELECT 101145,3,'2007-08-16 10:36:24.570' UNION ALL
SELECT 101145,3,'2007-11-06 09:19:26.157' UNION ALL
SELECT 101145,3,'2007-11-06 09:24:28.200' UNION ALL
SELECT 101145,4,'2010-09-27 14:11:03.287' UNION ALL
SELECT 101145,4,'2014-01-27 17:31:58.077'
DELETE @Audits
FROM @Audits a1
INNER JOIN (SELECT ObjectID, DateOfEntry
FROM (SELECT ObjectID, DateOfEntry, ObjectState,
LAG(ObjectState) OVER(PARTITION BY ObjectID ORDER BY DateOfEntry) AS [PreviousUserState]
FROM @Audits) AS Audits
WHERE Audits.ObjectState = PreviousUserState
) a2
ON a2.ObjectID = a1.ObjectID AND a2.DateOfEntry = a1.DateOfEntry
SELECT * FROM @Audits
产生此输出
ObjectID ObjectState DateOfEntry
----------- ----------- -----------------------
101144 1 2007-08-14 12:39:30.587
101144 3 2007-08-14 12:44:06.403
101144 4 2007-08-14 17:28:59.280
101144 3 2007-08-14 17:32:26.313
101144 4 2010-09-27 14:11:03.287
101145 1 2007-08-14 12:39:30.587
101145 3 2007-08-14 12:44:06.403
101145 4 2007-08-14 17:28:59.280
101145 3 2007-08-14 17:32:26.313
101145 4 2010-09-27 14:11:03.287
相关问题
最新问题
- 我写了这段代码，但我无法理解我的错误
- 我无法从一个代码实例的列表中删除 None 值，但我可以在另一个实例中。为什么它适用于一个细分市场而不适用于另一个细分市场？
- 是否有可能使 loadstring 不可能等于打印？卢阿
- java中的random.expovariate()
- Appscript 通过会议在 Google 日历中发送电子邮件和创建活动
- 为什么我的 Onclick 箭头功能在 React 中不起作用？
- 在此代码中是否有使用“this”的替代方法？
- 在 SQL Server 和 PostgreSQL 上查询，我如何从第一个表获得第二个表的可视化
- 每千个数字得到
- 更新了城市边界 KML 文件的来源？