join comma delimited data column
my table1 is :
T1
col1 col2
C1 john
C2 alex
C3 piers
C4 sara
and so table 2:
T2
col1 col2
R1 C1,C2,C4
R2 C3,C4
R3 C1,C4
how to result this?:
query result
col1 col2
R1 john,alex,sara
R2 piers,sara
R3 john,sara
please help me?
Solution 1:
Ideally, your best solution would be to normalize Table2 so you are not storing a comma separated list.
Once you have this data normalized then you can easily query the data. The new table structure could be similar to this:
CREATE TABLE T1
(
[col1] varchar(2),
[col2] varchar(5),
constraint pk1_t1 primary key (col1)
);
INSERT INTO T1
([col1], [col2])
VALUES
('C1', 'john'),
('C2', 'alex'),
('C3', 'piers'),
('C4', 'sara')
;
CREATE TABLE T2
(
[col1] varchar(2),
[col2] varchar(2),
constraint pk1_t2 primary key (col1, col2),
constraint fk1_col2 foreign key (col2) references t1 (col1)
);
INSERT INTO T2
([col1], [col2])
VALUES
('R1', 'C1'),
('R1', 'C2'),
('R1', 'C4'),
('R2', 'C3'),
('R2', 'C4'),
('R3', 'C1'),
('R3', 'C4')
;
Normalizing the tables would make it much easier for you to query the data by joining the tables:
select t2.col1, t1.col2
from t2
inner join t1
on t2.col2 = t1.col1
See Demo
Then if you wanted to display the data as a comma-separated list, you could use FOR XML PATH and STUFF:
select distinct t2.col1,
STUFF(
(SELECT distinct ', ' + t1.col2
FROM t1
inner join t2 t
on t1.col1 = t.col2
where t2.col1 = t.col1
FOR XML PATH ('')), 1, 1, '') col2
from t2;
See Demo.
If you are not able to normalize the data, then there are several things that you can do.
First, you could create a split function that will convert the data stored in the list into rows that can be joined on. The split function would be similar to this:
CREATE FUNCTION [dbo].[Split](@String varchar(MAX), @Delimiter char(1))
returns @temptable TABLE (items varchar(MAX))
as
begin
declare @idx int
declare @slice varchar(8000)
select @idx = 1
if len(@String)<1 or @String is null return
while @idx!= 0
begin
set @idx = charindex(@Delimiter,@String)
if @idx!=0
set @slice = left(@String,@idx - 1)
else
set @slice = @String
if(len(@slice)>0)
insert into @temptable(Items) values(@slice)
set @String = right(@String,len(@String) - @idx)
if len(@String) = 0 break
end
return
end;
When you use the split, function you can either leave the data in the multiple rows or you can concatenate the values back into a comma separated list:
;with cte as
(
select c.col1, t1.col2
from t1
inner join
(
select t2.col1, i.items col2
from t2
cross apply dbo.split(t2.col2, ',') i
) c
on t1.col1 = c.col2
)
select distinct c.col1,
STUFF(
(SELECT distinct ', ' + c1.col2
FROM cte c1
where c.col1 = c1.col1
FOR XML PATH ('')), 1, 1, '') col2
from cte c
See Demo.
A final way that you could get the result is by applying FOR XML PATH directly.
select col1,
(
select ', '+t1.col2
from t1
where ','+t2.col2+',' like '%,'+cast(t1.col1 as varchar(10))+',%'
for xml path(''), type
).value('substring(text()[1], 3)', 'varchar(max)') as col2
from t2;
See SQL Fiddle with Demo
Solution 2:
Here's a way of splitting the data without a function, then using the standard XML PATH method for getting the CSV list:
with CTE as
(
select T2.col1
, T1.col2
from T2
inner join T1 on charindex(',' + T1.col1 + ',', ',' + T2.col2 + ',') > 0
)
select T2.col1
, col2 = stuff(
(
select ',' + CTE.col2
from CTE
where T2.col1 = CTE.col1
for xml path('')
)
, 1
, 1
, ''
)
from T2
SQL Fiddle with demo.
As has been mentioned elsewhere in this question it is hard to query this sort of denormalised data in any sort of efficient manner, so your first priority should be to investigate updating the table structure, but this will at least allow to get the results you require.
Solution 3:
If you wanted to do this task in oracle we can use listagg and can accomplish this easily.
A possible equivalent available in SQL Server for listagg is Stuff
So using stuff you can try with following query:
SELECT T2.Col1,
Stuff((SELECT ',' + CAST(T1.Col2 AS VARCHAR(100))
FROM T1
WHERE T2.Col2 LIKE T1.Col1
FOR Xml Path('')),
1,
1,
'')
FROM T2
Solution 4:
First write a table value function for split col2 on tbl2.
CREATE FUNCTION [dbo].[Split](@String varchar(100), @Delimiter char(1))
returns @temptable TABLE (items VARCHAR(5))
as
begin
declare @idx int
declare @slice VARCHAR(5)
select @idx = 1
if len(@String)<1 or @String is null return
while @idx!= 0
begin
set @idx = charindex(@Delimiter,@String)
if @idx!=0
set @slice = left(@String,@idx - 1)
else
set @slice = @String
if(len(@slice)>0)
insert into @temptable(Items) values(@slice)
set @String = right(@String,len(@String) - @idx)
if len(@String) = 0 break
end
return
end
Go
;WITH SplitList
AS ( SELECT T2.Col1 ,
T1.Col2
FROM T2
CROSS APPLY dbo.Split(T2.Col2, ',') S
INNER JOIN T1 ON T1.Col1 = S.Items
)
SELECT T2.Col1 ,
STUFF(( SELECT ', ' + SplitList.Col2
FROM SplitList
WHERE SplitList.Col1 = T2.Col1
FOR
XML PATH('')
), 1, 2, '')
FROM T2
Solution 5:
If you are like me and you are a stickler for CTE's especially recursive CTE's as supposed to STUFF and XML Path:
DECLARE @T1 TABLE (
col1 CHAR(2),
col2 VARCHAR(10)
)
INSERT INTO @T1
VALUES ('C1', 'john'),
('C2', 'alex'),
('C3', 'piers'),
('C4', 'sara');
DECLARE @T2 TABLE (
col1 CHAR(2),
col2 CHAR(100)
)
INSERT INTO @T2
VALUES ('R1', 'C1,C2,C4'),
('R2', 'C3,C4'),
('R3', 'C1,C4');
WITH T2Sorted AS (
SELECT col1, col2, RN = ROW_NUMBER() OVER (ORDER BY col1) FROM @T2
), CTERecursionOnT2 AS (
SELECT RN, col1, col2, 0 AS PrevCharIndex, CHARINDEX(',', col2, 1) AS NextCharIndex FROM T2Sorted
UNION ALL
SELECT a.RN, a.col1, a.col2, b.NextCharIndex, CHARINDEX(',', a.col2, b.NextCharIndex + 1)
FROM T2Sorted a
JOIN CTERecursionOnT2 b ON a.RN = b.RN
WHERE b.NextCharIndex > 0
), CTEIndividualCol2Items AS (
SELECT *, SUBSTRING(col2, PrevCharIndex + 1, CASE WHEN NextCharIndex = 0 THEN LEN(col2) ELSE NextCharIndex - 1 END - PrevCharIndex) AS itemCol2
FROM CTERecursionOnT2
), CTELookupT1 AS (
SELECT a.col1, b.col2, RN = ROW_NUMBER() OVER (PARTITION BY a.col1 ORDER BY a.PrevCharIndex)
FROM CTEIndividualCol2Items a
JOIN @T1 b ON a.itemCol2 = b.col1
), CTERecursionOnLookupT1 AS (
SELECT col1, CAST(col2 AS VARCHAR(MAX)) AS col2, RN
FROM CTELookupT1
WHERE RN = 1
UNION ALL
SELECT a.col1, b.col2 + ',' + a.col2, a.RN
FROM CTELookupT1 a
JOIN CTERecursionOnLookupT1 b ON a.col1 = b.col1 AND a.RN = b.RN + 1
), CTEFinal AS (
SELECT *, RNDesc = ROW_NUMBER() OVER (PARTITION BY col1 ORDER BY RN DESC)
FROM CTERecursionOnLookupT1
)
SELECT col1, col2
FROM CTEFinal
WHERE RNDesc = 1
ORDER BY col1
Obviously you could break up the first recursion part into separate functions as the already agreed solution has suggested i.e. CTERecursionOnT2 and thus CTEIndividualCol2Items can be your alternative Split function (I would include the order id as well), and thus:
;WITH CTEIndividualCol2Items AS (
SELECT a.col1, b.value as itemCol2, b.id AS PrevCharIndex
FROM @T2 a
CROSS APPLY (
SELECT id, items FROM dbo.Split(a.col2, ',')
) b
) ...
and you split function:
CREATE FUNCTION dbo.Split(@String varchar(100), @Delimiter char(1))
RETURNS TABLE
AS
RETURN
(
WITH CTERecursion AS (
SELECT id = 1, PrevCharIndex = 0, NextCharIndex = CHARINDEX(@Delimiter, @String, 1)
UNION ALL
SELECT id + 1, NextCharIndex, CHARINDEX(@Delimiter, @String, NextCharIndex + 1) FROM CTERecursion WHERE NextCharIndex > 0
)
SELECT Id, items = SUBSTRING(@String, PrevCharindex + 1, (CASE WHEN NextCharIndex = 0 THEN LEN(@String) ELSE NextCharIndex - 1 END) - PrevCharIndex)
FROM CTERecursion
WHERE @String > ''
)