我已经按顺序阅读了一些关于分组的主题，这几乎是我需要的，但我找不到解决问题的方法。
我有一张这样的表：
PlanificatorPozitieID JalonID DataStart DataFinal
--------------------- ----------- ----------------------- -----------------------
26 46 2012-05-21 00:00:00.000 2012-05-31 00:00:00.000
28 48 2012-06-01 00:00:00.000 2012-06-01 00:00:00.000
27 60 2012-06-02 00:00:00.000 2012-06-02 00:00:00.000
29 60 2012-06-07 00:00:00.000 2012-06-08 00:00:00.000
37 60 2012-06-08 00:00:00.000 2012-06-10 00:00:00.000
30 65 2012-06-10 00:00:00.000 2012-06-13 00:00:00.000
31 65 2012-06-18 00:00:00.000 2012-06-24 00:00:00.000
32 65 2012-06-23 00:00:00.000 2012-07-01 00:00:00.000
33 66 2012-07-02 00:00:00.000 2012-07-02 00:00:00.000
34 66 2012-07-02 00:00:00.000 2012-07-05 00:00:00.000
36 66 2012-07-06 00:00:00.000 2012-07-10 00:00:00.000
Desired output:
PlanificatorPozitieID JalonID DataStart DataFinal
--------------------- ----------- ----------------------- -----------------------
26 46 2012-05-21 00:00:00.000 2012-05-31 00:00:00.000
28 48 2012-06-01 00:00:00.000 2012-06-01 00:00:00.000
27 60 2012-06-02 00:00:00.000 2012-06-02 00:00:00.000
29 60 2012-06-07 00:00:00.000 2012-06-10 00:00:00.000
30 65 2012-06-10 00:00:00.000 2012-06-13 00:00:00.000
31 65 2012-06-18 00:00:00.000 2012-07-01 00:00:00.000
33 66 2012-07-02 00:00:00.000 2012-07-05 00:00:00.000
36 66 2012-07-06 00:00:00.000 2012-07-10 00:00:00.000
所以我必须按JalonID进行分组，但只有在DataFinal＆gt; = DataStart的情况下才能进行分组。我想得到每个JalonID的时间段，但我想只得到没有暂停时间的句点。
家我清楚了。
select MIN(pp.DataStart) as DataStart, MAX(pp.DataFinal) as DataFinal, pp.JalonID FROM #PlanPozitii pp
GROUP BY pp.JalonID
但是这个查询并不能满足我的条件，可以按连续的句点分组。
至于澄清。采取以下示例
30 65 2012-06-10 00:00:00.000 2012-06-13 00:00:00.000
31 65 2012-06-18 00:00:00.000 2012-06-24 00:00:00.000
32 65 2012-06-23 00:00:00.000 2012-07-01 00:00:00.000
2012-06-13 00:00:00.000＆lt; 2012-06-18 00:00:00.000因此，PlanificatorPozitieID 30和31之间不会发生任何群组。
但是2012-06-24 00:00:00.000＆gt; 2012-06-23 00:00:00.000所以现在PlanificatorPozitieID 31和32之间会有一个小组。
所以从这3行我们将获得两行。
30 65 2012-06-10 00:00:00.000 2012-06-13 00:00:00.000
31 65 2012-06-18 00:00:00.000 2012-07-01 00:00:00.000
DECLARE @YourTable TABLE(PlanificatorPozitieID INT, JalonID INT,DataStart DATETIME, DataFinal DATETIME)
INSERT INTO @YourTable VALUES
(39,1223,'2015-02-16 00:00:00.000','2015-02-20 00:00:00.000'),
(43,1223,'2015-02-19 00:00:00.000','2015-02-24 00:00:00.000'),
(40,1223,'2015-02-23 00:00:00.000','2015-02-27 00:00:00.000'),
(42,1223,'2015-03-09 00:00:00.000','2015-03-13 00:00:00.000')
;WITH cte AS
(
SELECT a.PlanificatorPozitieID,
a.JalonID,
a.DataStart,
COALESCE(b.DataFinal,a.datafinal) AS [DataFinal],
ROW_NUMBER() OVER (PARTITION BY a.JalonID ORDER BY DATEDIFF(dd,a.datastart, COALESCE(b.DataFinal,a.datafinal))) [rn],
COUNT(*) OVER (PARTITION BY a.JalonID) [cnt]
FROM @YourTable a
LEFT JOIN @YourTable b
ON a.JalonID = b.JalonID AND
b.DataStart BETWEEN a.DataStart AND a.DataFinal AND
a.PlanificatorPozitieID <> b.PlanificatorPozitieID AND
DATEDIFF(dd,a.DataStart,a.DataFinal) < DATEDIFF(dd,a.DataStart,b.DataFinal)
)
SELECT *
FROM cte
WHERE rn= 1 OR rn=cnt
结果：
PlanificatorPozitieID JalonID DataStart DataFinal rn cnt
--------------------- ----------- ----------------------- ----------------------- -------------------- -----------
40 1223 2015-02-23 00:00:00.000 2015-02-27 00:00:00.000 1 4
43 1223 2015-02-19 00:00:00.000 2015-02-27 00:00:00.000 4 4
预期结果：
PlanificatorPozitieID JalonID DataStart DataFinal
--------------------- ----------- ----------------------- -----------------------
39 1223 2015-02-16 00:00:00.000 2015-02-27 00:00:00.000
42 1223 2015-03-09 00:00:00.000 2015-03-13 00:00:00.000
答案 0 :(得分：2)
我不知道它是否可以正常使用实际数据，因为我没有严格测试，但这是一个解决方案：
DECLARE @YourTable TABLE(PlanificatorPozitieID INT, JalonID INT,DataStart DATETIME, DataFinal DATETIME)
INSERT INTO @YourTable VALUES
(26,46,'2012-05-21 00:00:00.000','2012-05-31 00:00:00.000'),
(28,48,'2012-06-01 00:00:00.000','2012-06-01 00:00:00.000'),
(27,60,'2012-06-02 00:00:00.000','2012-06-02 00:00:00.000'),
(29,60,'2012-06-07 00:00:00.000','2012-06-08 00:00:00.000'),
(37,60,'2012-06-08 00:00:00.000','2012-06-10 00:00:00.000'),
(30,65,'2012-06-10 00:00:00.000','2012-06-13 00:00:00.000'),
(31,65,'2012-06-18 00:00:00.000','2012-06-24 00:00:00.000'),
(32,65,'2012-06-23 00:00:00.000','2012-07-01 00:00:00.000'),
(33,66,'2012-07-02 00:00:00.000','2012-07-02 00:00:00.000'),
(34,66,'2012-07-02 00:00:00.000','2012-07-05 00:00:00.000'),
(36,66,'2012-07-06 00:00:00.000','2012-07-10 00:00:00.000')
;WITH cte AS
(
SELECT a.PlanificatorPozitieID,
a.JalonID,
a.DataStart,
COALESCE(b.DataFinal,a.datafinal) AS [DataFinal],
ROW_NUMBER() OVER (PARTITION BY a.JalonID ORDER BY DATEDIFF(dd,a.datastart, COALESCE(b.DataFinal,a.datafinal))) [rn],
COUNT(*) OVER (PARTITION BY a.JalonID) [cnt]
FROM @YourTable a
LEFT JOIN @YourTable b
ON a.JalonID = b.JalonID AND
b.DataStart BETWEEN a.DataStart AND a.DataFinal AND
a.PlanificatorPozitieID <> b.PlanificatorPozitieID AND
DATEDIFF(dd,a.DataStart,a.DataFinal) < DATEDIFF(dd,a.DataStart,b.DataFinal)
)
SELECT *
FROM cte
WHERE rn= 1 OR rn=cnt
答案 1 :(得分：0)
我找到了一个解决方案，因为它使用了2个游标，效率不高。但它适用于任何人需要一个例子的情况
DROP TABLE #DateTEst
DROP TABLE #pozitii
drop table #PozitiiJaloaneStandard
CREATE TABLE #DateTest (JalonStandardID int,DataStart datetime,DataFinal datetime)
INSERT INTO #DateTest VALUES (1,'2015-05-05','2015-05-08')
INSERT INTO #DateTest VALUES (1,'2015-05-09','2015-05-13')
INSERT INTO #DateTest VALUES (1,'2015-05-12','2015-05-15')
INSERT INTO #DateTest VALUES (1,'2015-05-16','2015-05-18')
INSERT INTO #DateTest VALUES (1,'2015-05-14','2015-05-19')
INSERT INTO #DateTest VALUES (2,'2015-05-05','2015-05-06')
INSERT INTO #DateTest VALUES (2,'2015-05-06','2015-05-07')
INSERT INTO #DateTest VALUES (2,'2015-05-06','2015-05-09')
INSERT INTO #DateTest VALUES (3,'2015-05-05','2015-05-07')
INSERT INTO #DateTest VALUES (3,'2015-05-08','2015-05-10')
INSERT INTO #DateTest VALUES (4,'2015-05-05','2015-05-08')
INSERT INTO #DateTest VALUES (5,'2015-05-07','2015-05-07')
INSERT INTO #DateTest VALUES (5,'2015-05-08','2015-05-08')
INSERT INTO #DateTest VALUES (5,'2015-05-09','2015-05-12')
INSERT INTO #DateTest VALUES (5,'2015-05-11','2015-05-12')
INSERT INTO #DateTest VALUES (6,'2015-05-05','2015-05-20')
INSERT INTO #DateTest VALUES (6,'2015-05-15','2015-05-18')
CREATE TABLE #Pozitii (DataStart datetime, DataFinal datetime)
CREATE TABLE #PozitiiJaloaneStandard ( JalonStandardID int, DataStart datetime, DataFinal datetime)
Declare @JalonStandarID int
DEclare @PlanificatorPozitieID int
Declare @DataStartPozitie datetime
Declare @DataFinalPozitie datetime
DEclare @DataStartMin datetime
Declare @PozitieMinima int
Declare @DataFinalMin datetime
Declare @DataStartJalonStandard datetime
Declare @DataFinalJalonStandard datetime
Declare Crs_JaloaneStandard Cursor For
Select JalonStandardID
From #DateTest
ORDER BY JalonStandardID
Open Crs_JaloaneStandard
Fetch Next From Crs_JaloaneStandard Into
@JalonStandarID
While @@Fetch_Status = 0
Begin
INSERT INTO #Pozitii
SELECT pp.DataStart,pp.DataFinal
FROM #DateTest pp
WHERE pp.JalonStandardID = @JalonStandarID
GROUP BY pp.DataStart,pp.DataFinal
SELECT @DataStartMin = MIN(DataStart) FROM #Pozitii
SELECT @DataFinalMin = DataFinal FROM
#Pozitii WHERE DataStart = @DataStartMin
Declare Crs_Pozitii Cursor For
SELECT
p.DataStart,p.DataFinal
FROM #Pozitii p
ORDER by p.DataStart ASC
Open Crs_Pozitii
Fetch Next From Crs_Pozitii Into
@DataStartPozitie,@DataFinalPozitie
while @@FETCH_STATUS = 0
begin
if (@DataFinalMin > @DataStartPozitie) and (@DataFinalMin <= @DataFinalPozitie )
begin
set @DataFinalMin = @DataFinalPozitie
end
if (@DataFinalMin <= @DataStartPozitie) begin
INSERT INTO #PozitiiJaloaneStandard VALUES (@JalonStandarID,@DataStartMin,@DataFinalMin)
set @DataFinalMin = @DataFinalPozitie
set @DataStartMin = @DataStartPozitie
print @DataStartPozitie
print @DataFinalPozitie
end
Fetch Next From Crs_Pozitii Into
@DataStartPozitie,@DataFinalPozitie
End
INSERT INTO #PozitiiJaloaneStandard VALUES (@JalonStandarID,@DataStartMin,@DataFinalMin)
DELETE FROM #Pozitii
Close Crs_Pozitii
Deallocate Crs_Pozitii
Fetch Next From Crs_JaloaneStandard Into
@JalonStandarID
End
Close Crs_JaloaneStandard
Deallocate Crs_JaloaneStandard
SELECT * FROM #PozitiiJaloaneStandard
GROUP BY JalonStandardID,DataStart,DataFinal
答案 2 :(得分：-1)
https://msdn.microsoft.com/en-us/library/hh231256.aspx
我正在寻找LAG（）函数，它可以访问查询中的前一行。我的想法是计算当前行和上一行之间的差异，以使列分组;
with a as (
select 1 as ID, 1 as A, 2 as B union all
select 1 as ID, 2 as A, 3 as B union all
select 1 as ID, 3 as A, 4 as B union all
select 1 as ID, 4 as A, 5 as B union all
select 1 as ID, 6 as A, 7 as B
)
select ID, A, B, A-LAG(B,1,0) OVER (order by ID) as koe from a where B > A
如果您运行该查询，则会得到结果为;
ID A B koe
----------- ----------- ----------- -----------
1 1 2 1
1 2 3 0
1 3 4 0
1 4 5 0
1 6 7 1
想象一下A是DataStart而B是DataFinal，并且计算出的koe是差异的，因为你可以看到它可以在除第一行之外的所有行上运行...第一行在不存在的行之间获得差异（因此它的0）。但这是我开始尝试的方向。