IF OBJECT_ID('Tempdb..#TempTable') IS NOT NULL
DROP TABLE #TempTable
CREATE TABLE #TempTable
(
[ID] INT NOT NULL ,
[Value] VARCHAR(50) NULL ,
[Date] DATE NULL ,
[Time] TIME(7) NULL ,
[Duration] INT NULL ,
[srcFile] VARCHAR(50) NULL,
)
INSERT #TempTable
( [ID], [Value], [Date], [Time], [Duration], [srcFile] )
VALUES ( 1, N'One', CAST(N'2014-07-29' AS DATE), CAST(N'23:34:00' AS TIME), 1710, N'sF1' ),
( 2, N'One', CAST(N'2014-07-30' AS DATE), CAST(N'00:00:10' AS TIME), 1710, N'sF1' ),
( 3, N'One', CAST(N'2014-07-30' AS DATE), CAST(N'01:30:00' AS TIME), 1710, N'sF1' ),
( 4, N'One', CAST(N'2014-07-30' AS DATE), CAST(N'01:54:00' AS TIME), 1710, N'sF1' ),
( 5, N'One', CAST(N'2014-07-30' AS DATE), CAST(N'13:30:00' AS TIME), 1710, N'sF1' ),
( 6, N'One', CAST(N'2014-07-30' AS DATE), CAST(N'13:57:00' AS TIME), 1710, N'sF2' ),
( 7, N'One', CAST(N'2014-07-30' AS DATE), CAST(N'23:34:00' AS TIME), 1710, N'sF1' ),
( 8, N'One', CAST(N'2014-07-31' AS DATE), CAST(N'00:00:10' AS TIME), 1710, N'sF2' ),
( 9, N'One', CAST(N'2014-07-31' AS DATE), CAST(N'00:10:10' AS TIME), 1710, N'sF3' ),
( 10, N'One', CAST(N'2014-08-01' AS DATE), CAST(N'00:00:00' AS TIME), 1710, N'sF2' ),
( 11, N'One', CAST(N'2014-08-01' AS DATE), CAST(N'00:00:00' AS TIME), 1710, N'sF1' ),
( 12, N'One', CAST(N'2014-08-01' AS DATE), CAST(N'01:00:00' AS TIME), 1710, N'sF3' ),
( 13, N'One', CAST(N'2014-08-01' AS DATE), CAST(N'01:00:00' AS TIME), 1710, N'sF4' ),
( 14, N'Two', CAST(N'2014-08-01' AS DATE), CAST(N'00:01:00' AS TIME), 1710, N'sF2' )
SELECT *
FROM #TempTable
基准表
ID Value Date Time Duration srcFile
1 One 7/29/2014 23:34:00 1710 sF1
2 One 7/30/2014 0:00:10 1710 sF1
3 One 7/30/2014 1:30:00 1710 sF1
4 One 7/30/2014 1:54:00 1710 sF1
5 One 7/30/2014 13:30:00 1710 sF1
6 One 7/30/2014 13:57:00 1710 sF2
7 One 7/30/2014 23:34:00 1710 sF1
8 One 7/31/2014 0:00:10 1710 sF2
9 One 8/1/2014 0:00:00 1710 sF2
10 Two 8/1/2014 0:01:00 1710 sF2
11 One 8/1/2014 0:00:00 1710 sF1
要求：
当[Value] + [Date] + [Time]匹配时，Dup 输出：标记isDup标志1和dupFIle与srcFile，用于两个或多个记录，其中dup条件匹配。
当[值]匹配且任何两个或多个记录的[日期] + [时间]落在[日期] + [时间]加号（+）[持续时间]内时，然后重叠（注意：当所有匹配的记录都是DUP时）。 ..它们也不能重叠。但重叠可以有至少一个独特的记录和多个重复，它们都在持续时间范围内）。 输出：标记isOverlap标记为1，overlapFile为srcFile，用于两个或多个重叠条件匹配的记录。
这就是我试过的
;WITH dupCTE AS (
SELECT ID, Value, [Date], [Time], Duration, srcFile
,CASE
WHEN COUNT(*) OVER (PARTITION BY Value, [Date], [Time]) > 1 THEN 1
ELSE 0
END AS isDup
,CASE WHEN COUNT(*) OVER (PARTITION BY Value, [Date], [Time]) > 1 THEN STUFF((SELECT ' - ' + srcFile
FROM #TempTable T
WHERE T.Value = TT.Value
AND T.[Date] = TT.[Date]
AND T.[Time] = TT.[Time]
FOR XML PATH('')), 1, 3, '')
ELSE NULL
END AS dupFIle
FROM #TempTable TT
)
, overlapCTE AS (
SELECT A. ID, A.Value, A.[Date], A.[Time], A.Duration, A.srcFile, A.isDup, A.dupFIle
,CASE WHEN B.ID IS NOT NULL THEN 1
ELSE 0
END AS 'isOverlap'
,CASE WHEN b.ID IS NOT NULL THEN STUFF((SELECT ' - ' + srcFile
FROM #TempTable T
WHERE T.Value = A.Value
AND ((CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2) > CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2) AND CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2) < DATEADD(SECOND, A.Duration, CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2)))
OR (CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2) > CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2) AND CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2) < DATEADD(SECOND, B.Duration, CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2))))
FOR XML PATH('')), 1, 3, '')
ELSE NULL
END AS 'overlapFiles'
FROM dupCTE A LEFT JOIN dupCTE B
ON A.Value = B.Value
AND ((CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2) > CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2) AND CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2) < DATEADD(SECOND, A.Duration, CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2)))
OR (CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2) > CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2) AND CAST(CAST(A.[Date] AS VARCHAR(10)) + ' ' + CAST(A.[Time] AS VARCHAR(16)) AS DateTime2) < DATEADD(SECOND, B.Duration, CAST(CAST(B.[Date] AS VARCHAR(10)) + ' ' + CAST(B.[Time] AS VARCHAR(16)) AS DateTime2))))
WHERE A.isDup = 1 OR
B.ID IS NOT NULL
)
SELECT * FROM overlapCTE
DROP TABLE #TempTable
当前输出
ID Value Date Time Duration srcFile isDup dupFIle isOverlap overlapFiles
1 One 2014-07-29 23:34:00 1710 sF1 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
2 One 2014-07-30 00:00:10 1710 sF1 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
3 One 2014-07-30 01:30:00 1710 sF1 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
4 One 2014-07-30 01:54:00 1710 sF1 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
5 One 2014-07-30 13:30:00 1710 sF1 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
6 One 2014-07-30 13:57:00 1710 sF2 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
7 One 2014-07-30 23:34:00 1710 sF1 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
8 One 2014-07-31 00:00:10 1710 sF2 0 NULL 1 sF1 - sF1 - sF1 - sF1 - sF1 - sF2 - sF1 - sF2 - sF2 - sF1
9 One 2014-08-01 00:00:00 1710 sF2 1 sF2 - sF1 0 NULL
11 One 2014-08-01 00:00:00 1710 sF1 1 sF2 - sF1 0 NULL
期望输出
ID Value Date Time Duration srcFile isDup dupFIle isOverLap overlapFile
1 One 2014-07-29 24:34:00 1710 sF1 0 NULL 1 sF1 - sF1
2 One 2014-07-30 00:00:10 1710 sF1 0 NULL 1 sF1 - sF1
3 One 2014-07-30 01:30:00 1710 sF1 0 NULL 1 sF1 - sF1
4 One 2014-07-30 01:54:00 1710 sF1 0 NULL 1 sF1 - sF1
5 One 2014-07-30 13:30:00 1710 sF1 0 NULL 1 sF1 - sF2
6 One 2014-07-30 13:57:00 1710 sF2 0 NULL 1 sF2 - sF1
7 One 2014-07-30 24:34:00 1710 sF1 0 NULL 1 sF1 - sF2
8 One 2014-07-31 00:00:10 1710 sF2 0 NULL 1 sF2 - sF1
9 One 2014-08-01 00:00:00 1710 sF2 1 sF2 - sF1 0 NULL
10 Two 2014-08-01 00:01:00 1710 sF2 0 NULL 0 NULL
11 One 2014-08-01 00:00:00 1710 sF1 1 sF1 - sF2 0 NULL
我不符合要求。任何帮助将不胜感激。
谢谢
更新： 添加了电流输出
UPDATE2： 在Dup CTE中发现错误（使用ID代替Value）。 仍然需要所需的输出。
UPDATE3： 进步人士，我们非常接近。现在，重叠逻辑正在“正常工作”。一个主要问题是overlapFIles。它应该只列出彼此重叠的记录的文件（现在，它列出了来自overlapCTE输出的所有文件，而不是仅列出那些符合STUFF查询中的WHERE的文件）。还有，有办法列出这个独特的记录吗？
Update4：添加了更多记录，以查看重复和重叠查询是否可以容纳多于两条记录。
答案 0 :(得分：2)
代码符合您的要求（希望如此）。 我测试了它添加了更多的重叠和重复，它不仅有2个重复，重叠文件（例如srcFile =＆＃39; sF3＆＃39;），但有这些观察结果：
with rows
(
select [ID],[Value], [Date], [Time], [Duration], [srcFile],
cast(cast([date] as varchar(10))+' ' +cast(time as varchar(8)) as datetime) as datetime,
dateadd(ss,-duration,cast(cast([date] as varchar(10))+' ' +cast(time as varchar(8)) as datetime)) as date_from,
dateadd(ss,duration,cast(cast([date] as varchar(10))+' ' +cast(time as varchar(8)) as datetime)) as date_to
from #TempTable
)
, dups
as
(
SELECT [value], [Date], [Time]
FROM rows
group by [value], [Date], [Time]
having count([ID])>1
)
, dups_files
as
(
select r.*
,
STUFF((select ' - '+d.srcFile
FROM rows as d
WHERE (r.[value]=d.value and r.time=d.time
and r.date=d.date)
order by d.srcFile
FOR xml path('')),1,3,'') as dupFile
FROM dups r
)
, duplicities
as
(
select a.id, d.dupFile
from rows a join dups_files d
on ( a.value=d.value and a.date=d.date
and a.time=d.time)
)
, overlaps_pairs
as
(
select f.id as id_a, d.id as id_b, f.srcfile as srcfile_a, d.srcfile as srcfile_b
from rows f JOIN rows d
ON (f.id<>d.id
AND f.value=d.value
and f.datetime between d.date_from and d.date_to)
where not exists
(select 1 FROM duplicities du
where f.id=du.id)
)
, overlaps
as
(
select DISTINCT op.id_a as id, 1 as isoverlap
, STUFF((select distinct ' - ' +aa.srcFile_b
from overlaps_pairs aa
where aa.id_a = op.id_a
or op.id_b=aa.id_a
for xml path ('')),1,3,'') as overlapfiles
from overlaps_pairs op
)
select a.id, a.value, a.date, a.time, a.duration, a.srcFile,
case when d.id is not null then 1 else 0 end as isDup, d.dupFile,
o.isoverlap, o.overlapfiles
from rows a LEFT OUTER JOIN overlaps o
on (a.id=o.id)
LEFT OUTER JOIN duplicities d
on (a.id=d.id);
答案 1 :(得分：1)
我觉得自己厌倦了使用可能不需要的CTE。我不知道你的用例，但这个表结构很奇怪IMO;这是家庭作业吗？无论如何，这应该可以解决问题：
WITH
tFullCTE AS (
SELECT ID as ID, Value as Value, [Date] as Date, [Time] as Time, CAST(CAST([Date] AS VARCHAR(10)) + ' ' + CAST([Time] AS VARCHAR(16)) AS DateTime2) as DateTime, Duration as Duration, srcFile as srcFile
FROM #TempTable TT
)
,dupCTE AS (
SELECT main.ID as FirstID, dups.ID as SecondID
FROM tFullCTE main
INNER join tFullCTE dups on main.value = dups.value and main.DateTime = dups.DateTime and main.id <> dups.id
)
,
overlapCTE AS (
SELECT main.ID as FirstID, ovlp.ID as SecondID
FROM tFullCTE main
INNER JOIN tFullCTE ovlp
ON main.Value = ovlp.Value
AND ((ovlp.DateTime > main.DateTime AND ovlp.DateTime < DATEADD(SECOND, main.Duration, main.DateTime))
OR (main.DateTime > ovlp.DateTime AND main.DateTime < DATEADD(SECOND, ovlp.Duration, ovlp.DateTime)))
AND main.ID <> ovlp.ID
)
SELECT main.*,
CASE WHEN EXISTS (SELECT dup.SecondID FROM dupCTE dup WHERE dup.FirstID = main.ID) THEN 1 ELSE 0 END as isDup,
STUFF(( SELECT DISTINCT ' - ' + dupDetails.srcFile
FROM dupCTE dup
INNER JOIN tFullCTE dupDetails on dup.SecondID = dupDetails.ID
WHERE dup.FirstID = main.ID
FOR XML PATH('')), 1, 3, '') as dupFile,
CASE WHEN EXISTS (SELECT ovlp.SecondID FROM overlapCTE ovlp WHERE ovlp.FirstID = main.ID) THEN 1 ELSE 0 END as isOverlap,
STUFF(( SELECT DISTINCT ' - ' + ovlpDetails.srcFile
FROM overlapCTE ovlp
INNER JOIN tFullCTE ovlpDetails on ovlp.SecondID = ovlpDetails.ID
WHERE ovlp.FirstID = main.ID
FOR XML PATH('')), 1, 3, '') as overlapFile
FROM tFullCTE main
答案 2 :(得分：1)
;WITH ModifiedData AS(
SELECT
*,
[DateTime] = DATEADD(S, DATEDIFF(S, 0, [Time]), CAST([Date] AS DATETIME)),
DateTimeWithDuration = DATEADD(S, Duration, DATEADD(S, DATEDIFF(S, 0, [Time]), CAST([Date] AS DATETIME)))
FROM #TempTable
)
, Flags AS(
SELECT
m.*,
isDup = CASE
WHEN COUNT(*) OVER(PARTITION BY m.Value, m.[DateTime]) > 1 THEN 1
ELSE 0
END,
dupFile = CASE
WHEN COUNT(*) OVER(PARTITION BY m.Value, m.[DateTime]) > 1 THEN
STUFF((
SELECT ' - ' + srcFile
FROM ModifiedData
WHERE
Value = m.Value
AND m.DateTime = [DateTime]
FOR XML PATH('')), 1, 3, '')
ELSE NULL
END,
isOverlap = CASE
WHEN COUNT(m2.Value) > 0 THEN 1
ELSE 0
END
FROM ModifiedData m
LEFT JOIN ModifiedData m2
ON m2.value = m.Value
AND
(
(m2.DateTime > m.DateTime AND m2.DateTime < m.DateTimeWithDuration)
OR (m.DateTime > m2.DateTime AND m.DateTime < m2.DateTimeWithDuration)
)
AND m2.ID <> m.ID
GROUP BY
m.ID, m2.ID, m.Value, m.Date, m.Time, m.Duration, m.srcFile, m.DateTime, m.DateTimeWithDuration
)
SELECT
f.ID,
f.Value,
f.[Date],
f.[Time],
f.Duration,
f.SrcFile,
dupFile = CASE
WHEN isOverlap = 1 THEN
(SELECT f.srcFile + ' - ' + srcFile
FROM ModifiedData
WHERE
Value = f.Value
AND
(
([DateTime] > f.DateTime AND [DateTime] < f.DateTimeWithDuration)
OR (f.DateTime > [DateTime] AND f.DateTime < DateTimeWithDuration)
)
AND ID <> f.ID
FOR XML PATH(''))
ELSE NULL
END
FROM Flags f
ORDER BY f.ID
答案 3 :(得分：1)
这应该可以准确地告诉你：
With CTE as (Select T.ID ID1, T.srcFile + ' - ' + c.srcFile over1, '1' as isDup from
#TempTable T
INNER JOIN #TempTable c on T.Value = c.Value and c.ID <> T.ID and (Cast(C.Date as datetime) + Cast(C.Time as datetime)) = (Cast(T.Date as datetime) + Cast(T.Time as datetime))),
CTE2 as
(Select T.ID ID1, c.ID ID2, T.srcFile + ' - ' + c.srcFile over1, c.srcFile + ' - ' + T.srcFile over2, '1' as isOverLap from
#TempTable T
INNER JOIN #TempTable c on T.Value = c.Value and c.ID <> T.ID
Where DateAdd(second, c.Duration, Cast(C.Date as datetime) + Cast(C.Time as datetime)) > (Cast(T.Date as datetime) + Cast(T.Time as datetime)) and (Cast(C.Date as datetime) + Cast(C.Time as datetime)) < (Cast (T.Date as datetime) + Cast(T.Time as datetime)))
Select T.*, ISNULL((Select top 1 c.isDup from CTE c where c.ID1 = T.ID) ,0) isDup
,(Select substring((select ',' + c1.over1 as [text()] from CTE c1 where c1.ID1 = T.ID for xml path ('')),2,1000)) dupFile
,ISNULL((select Top 1 case isOverLap when 1 then 1 else 0 end from CTE2 c where c.ID1 = T.ID or C.ID2 = T.ID),0) isOverLap
,(Select substring((select case when T.ID = C.ID1 then ',' + c.over1 else ',' + c.over2 end as [text()] from CTE2 c where c.ID1 = T.ID or C.ID2 = T.ID for xml path('')),2,1000)) OverlapFile
from #TempTable T