我在SQL Server 2008 R2的表中有以下数据：
ID Code StartDate EndDate
10001 3 2014-07-25 2014-07-28
10001 3 2014-07-29 2014-10-06
10001 3 2014-10-07 2014-10-10
10001 1 2014-10-11 2014-10-31
10001 1 2014-11-01 2014-11-15
10001 3 2014-11-16 2014-11-25
10001 3 2014-11-26 NULL
20002 3 2014-07-25 2014-07-28
20002 3 2014-07-29 2014-10-06
20002 3 2014-10-07 NULL
30003 3 2014-07-25 2014-11-13
30003 3 2014-11-14 2014-11-24
30003 2 2014-11-25 NULL
我希望“折叠”具有相同Code和相邻EndDate和StartDate的所有记录。结果应该是：
ID Code StartDate EndDate
10001 3 2014-07-25 2014-10-10
10001 1 2014-10-11 2014-11-15
10001 3 2014-11-16 NULL
20002 3 2014-07-25 NULL
30003 3 2014-07-25 2014-11-24
30003 2 2014-11-25 NULL
我一直在尝试使用各种子查询和ROW_NUMBER()函数，但却无法使其工作。我怀疑这可以通过CTE轻松完成，但我无法围绕这些工作如何在这里尝试。有什么想法吗？
答案 0 :(得分：0)
试试这个，
CREATE TABLE #TEMP
(
ID INT,
CODE INT,
STARTDATE DATE,
ENDDATE DATE
)
INSERT INTO #TEMP VALUES
(10001,3,'2014-07-25','2014-07-28'),
(10001,3,'2014-07-29','2014-10-06'),
(10001,3,'2014-10-07','2014-10-10'),
(10001,1,'2014-10-11','2014-10-31'),
(10001,1,'2014-11-01','2014-11-15'),
(10001,3,'2014-11-16','2014-11-25'),
(10001,3,'2014-11-26',NULL),
(20002,3,'2014-07-25','2014-07-28'),
(20002,3,'2014-07-29','2014-10-06'),
(20002,3,'2014-10-07',NULL),
(30003,3,'2014-07-25','2014-11-13'),
(30003,3,'2014-11-14','2014-11-24'),
(30003,2,'2014-11-25',NULL)
SELECT T1.ID,T1.CODE,T1.STARTDATE,A.ENDDATE FROM (SELECT L.ID,L.CODE,MIN(STARTDATE) AS STARTDATE,DIFF FROM (SELECT ID,
CODE,
STARTDATE,
ENDDATE,
IsNull(Lag(CODE, 2)
OVER (
ORDER BY ID, STARTDATE, ENDDATE), CODE) AS T_LAG,
CODE - IsNull(Lag(CODE, 2)
OVER (
ORDER BY ID, STARTDATE, ENDDATE), CODE) AS DIFF
FROM #TEMP ) L
GROUP BY L.ID,L.CODE,DIFF) T1
CROSS APPLY(
SELECT ID,CODE,ENDDATE,DIFF FROM (SELECT ID,CODE,ENDDATE,DIFF,ROW_NUMBER() OVER (PARTITION BY ID,CODE,DIFF ORDER BY ID,CODE,STARTDATE DESC,ENDDATE DESC) AS T_R FROM (SELECT ID,
CODE,
STARTDATE,
ENDDATE,
IsNull(Lag(CODE, 2)
OVER (
ORDER BY ID, STARTDATE, ENDDATE), CODE) AS T_LAG,
CODE - IsNull(Lag(CODE, 2)
OVER (
ORDER BY ID, STARTDATE, ENDDATE), CODE) AS DIFF
FROM #TEMP ) A) A
WHERE T_R=1 AND ID=T1.ID AND CODE=T1.CODE AND DIFF=T1.DIFF)A
ORDER BY T1.ID,T1.STARTDATE
答案 1 :(得分：0)
试试这个（我在2008年使用Row_Number（）复制SQL 2012 Lead And Lag功能）：
MS SQL Server 2008架构设置：
CREATE TABLE EventLog
(
ID Int,
Code tinyint,
StartDate Date,
EndDate Date Null
)
INSERT INTO EventLog
Values
(10001, 3, '2014-07-25', '2014-07-28'),
(10001, 3, '2014-07-29', '2014-10-06'),
(10001, 3, '2014-10-07', '2014-10-10'),
(10001, 1, '2014-10-11', '2014-10-31'),
(10001, 1, '2014-11-01', '2014-11-15'),
(10001, 3, '2014-11-16', '2014-11-25'),
(10001, 3, '2014-11-26', NULL),
(20002, 3, '2014-07-25', '2014-07-28'),
(20002, 3, '2014-07-29', '2014-10-06'),
(20002, 3, '2014-10-07', NULL),
(30003, 3, '2014-07-25', '2014-11-13'),
(30003, 3, '2014-11-14', '2014-11-24'),
(30003, 2, '2014-11-25', NULL)
查询1 ：
WITH CTE
AS
(
SELECT ID,
Code,
StartDate,
EndDate,
ROW_NUMBER() OVER (PARTITION BY Id, Code ORDER BY Id, Code, StartDate) As RN
FROM EventLog
),
CTE2
AS
(
SELECT CTE.Id, CTE.Code, CTE.StartDate, CTE.EndDate,
CASE WHEN DATEDIFF(d, LAG.EndDate, CTE.StartDate) = 1
THEN Lag.EndDate
ELSE NULL
END AS PrevEndDate,
CASE WHEN DateDiff(d, LEAD.StartDate, CTE.EndDate) = -1
THEN Lead.StartDate
ELSE NULL
END As NextStartDate
FROM CTE
LEFT OUTER JOIN CTE AS Lag
ON CTE.ID = Lag.ID AND CTE.Code = Lag.Code AND Lag.Rn = CTE.RN - 1
LEFT OUTER JOIN CTE AS Lead
ON CTE.ID = Lead.ID AND CTE.Code = Lead.Code AND Lead.Rn = CTE.RN + 1
),
StartAndEnd
As
(
SELECT ID,
Code,
StartDate,
EndDate,
PrevEndDate,
NextStartDate,
ROW_NUMBER() OVER (PARTITION BY Id, Code ORDER BY ID, Code, StartDate) As RN
FROM CTE2
WHERE (PrevEndDate IS NULL Or NextStartDate IS NULL)
)
SELECT S.ID, s.Code, S.StartDate, E.EndDate
FROM StartAndEnd as S
LEFT JOIN StartAndEnd E
ON S.ID = E.ID AND S.Code = E.Code AND E.RN = S.Rn + 1
WHERE S.PrevEndDate Is Null
ORDER By s.Id, S.StartDate
<强> Results ：
| ID | CODE | STARTDATE | ENDDATE |
|-------|------|------------|------------|
| 10001 | 3 | 2014-07-25 | 2014-10-10 |
| 10001 | 1 | 2014-10-11 | 2014-11-15 |
| 10001 | 3 | 2014-11-16 | (null) |
| 20002 | 3 | 2014-07-25 | (null) |
| 30003 | 3 | 2014-07-25 | 2014-11-24 |
| 30003 | 2 | 2014-11-25 | (null) |
答案 2 :(得分：0)
由于您的范围是连续的，因此问题基本上变为gaps-and-islands。 © Andriy M
declare @EventLog table
(
Id int,
Code tinyint,
StartDate date,
EndDate date null
)
insert into @EventLog
values
(10001, 3, '2014-07-25', '2014-07-28'),
(10001, 3, '2014-07-29', '2014-10-06'),
(10001, 3, '2014-10-07', '2014-10-10'),
(10001, 1, '2014-10-11', '2014-10-31'),
(10001, 1, '2014-11-01', '2014-11-15'),
(10001, 3, '2014-11-16', '2014-11-25'),
(10001, 3, '2014-11-26', null),
(20002, 3, '2014-07-25', '2014-07-28'),
(20002, 3, '2014-07-29', '2014-10-06'),
(20002, 3, '2014-10-07', null),
(30003, 3, '2014-07-25', '2014-11-13'),
(30003, 3, '2014-11-14', '2014-11-24'),
(30003, 2, '2014-11-25', null);
declare @MaxDate date = '9999-12-31';
with cte as
(
select *,
g = row_number() over (partition by Id order by StartDate)
- row_number() over (partition by Id, Code order by StartDate)
from @EventLog
)
select
Id,
Code,
StartDate = min(StartDate),
EndDate = nullif(max(isnull(EndDate, @MaxDate)), @MaxDate)
from cte
group by
Id, Code, g;
答案 3 :(得分：0)
;WITH StartDates
AS(
Select e1.ID, e1.Code, e1.StartDate, ROW_NUMBER() OVER (Order By e1.ID asc) as RowNumber
From @EventLog e1
LEFT JOIN @EventLog e2 ON e1.Code = e2.Code and e2.EndDate = DATEADD(day,-1,e1.StartDate)
WHERE e2.Id is null
),
EndDates as(
Select e1.ID, e1.Code, e1.EndDate, ROW_NUMBER() OVER (Order by e1.ID asc) as RowNumber
FROM @EventLog e1
LEFT JOIN @EventLog e2 ON e1.Code = e2.Code and e2.StartDate = DATEADD(day,1,e1.EndDate)
WHERE e2.Id is null
)
Select s.ID, s.Code, s.StartDate, e.EndDate
FROM StartDates s
JOIN EndDates e ON s.Code = e.Code and s.RowNumber = e.RowNumber