Final version (I hope):
Since sql server 2008 doesn't support order by in the over clause of aggregate functions, I've added another cte to add the row index instead of the sum I've used in the previous version:
;WITH cteAllRows as
(
SELECT Item,
ItemIndex,
CASE WHEN ISNUMERIC(Item) = 0 THEN 'String'
WHEN ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) > 0 THEN 'Double'
WHEN ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) = 0 THEN 'Integer'
END As DataType
FROM dbo.SplitStrings_Numbers(@string, ',')
), cteAll as
(
SELECT Item,
DataType,
ItemIndex,
(
SELECT COUNT(*)
FROM cteAllRows tInner
WHERE tInner.DataType = 'String'
AND tInner.ItemIndex <= tOuter.ItemIndex
) As RowIndex
FROM cteAllRows tOuter
)
All the rest is the same as the previous version.
Update
The first thing I've done is to change the string split function to a function based on a tally table, so that I can easily add the row number to it. So, if you don't already have a tally table, create one.
If you are asking your self what is a tally table and why do you need it, read this article by Jeff Moden:
SELECT TOP 10000 IDENTITY(int,1,1) AS Number
INTO Tally
FROM sys.objects s1
CROSS JOIN sys.objects s2
ALTER TABLE Tally ADD CONSTRAINT PK_NumbersTest PRIMARY KEY CLUSTERED (Number)
GO
Then, create string split function based on the tally table (taken from Aaron's article but added the row index column):
CREATE FUNCTION dbo.SplitStrings_Numbers
(
@List NVARCHAR(MAX),
@Delimiter NVARCHAR(255)
)
RETURNS TABLE
WITH SCHEMABINDING
AS
RETURN
(
SELECT Item = SUBSTRING(@List, Number, CHARINDEX(@Delimiter, @List + @Delimiter, Number) - Number),
ROW_NUMBER() OVER (ORDER BY Number) As ItemIndex
FROM dbo.Tally
WHERE Number <= CONVERT(INT, LEN(@List))
AND SUBSTRING(@Delimiter + @List, Number, LEN(@Delimiter)) = @Delimiter
);
GO
Now, The trick I've used is very much like the previous one, only now I've added to the first cte a new column I've called RowIndex, that's basically a running total of the count of strings, based on the row index of all rows:
SELECT Item,
CASE WHEN ISNUMERIC(Item) = 0 THEN 'String'
WHEN ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) > 0 THEN 'Double'
WHEN ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) = 0 THEN 'Integer'
END As DataType,
SUM(CASE WHEN ISNUMERIC(Item) = 0 THEN 1 END) OVER(ORDER BY ItemIndex) As RowIndex
FROM dbo.SplitStrings_Numbers(@string, ',')
It gave me this result:
Item DataType RowIndex
---------- -------- -----------
ddd String 1
1.5 Double 1
1 Integer 1
eee String 2
2.3 Double 2
0 Integer 2
fff String 3
1.2 Double 3
ggg String 4
6.123 Double 4
1 Integer 4
As you can see, I now have a number for each row, so from now on it's simple:
;WITH cteAll as
(
SELECT Item,
CASE WHEN ISNUMERIC(Item) = 0 THEN 'String'
WHEN ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) > 0 THEN 'Double'
WHEN ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) = 0 THEN 'Integer'
END As DataType,
SUM(CASE WHEN ISNUMERIC(Item) = 0 THEN 1 END) OVER(ORDER BY ItemIndex) As RowIndex
FROM dbo.SplitStrings_Numbers(@string, ',')
), cteString AS
(
SELECT Item, RowIndex
FROM cteAll
WHERE DataType = 'String'
), cteDouble AS
(
SELECT Item, RowIndex
FROM cteAll
WHERE DataType = 'Double'
), cteInteger AS
(
SELECT Item, RowIndex
FROM cteAll
WHERE DataType = 'Integer'
)
SELECT T1.Item As [String],
T2.Item As [Double],
T3.Item As [Integer]
FROM dbo.Tally
LEFT JOIN cteString T1 ON T1.RowIndex = Number
LEFT JOIN cteDouble T2 ON t2.RowIndex = Number
LEFT JOIN cteInteger T3 ON t3.RowIndex = Number
WHERE COALESCE(T1.Item, T2.Item, T3.Item) IS NOT NULL
That gave me this result:
String Double Integer
---------- ---------- ----------
ddd 1.5 1
eee 2.3 0
fff 1.2 NULL
ggg 6.123 1
As you can see, the items are now sorted by the original order in the string.
Thanks for the challenge, It's been a while since I've had a decent one :-)
First attempt
Well, first you have to split that string into a table. To do that you should use a user defined function. You can pick the one best suited for you from Aaron Bertrand's Split strings the right way – or the next best way article.
For this demonstration I've chosen to use the SplitStrings_XML.
So first, create the function:
CREATE FUNCTION dbo.SplitStrings_XML
(
@List NVARCHAR(MAX),
@Delimiter NVARCHAR(255)
)
RETURNS TABLE
WITH SCHEMABINDING
AS
RETURN
(
SELECT Item = y.i.value('(./text())[1]', 'nvarchar(4000)')
FROM
(
SELECT x = CONVERT(XML, '<i>'
+ REPLACE(@List, @Delimiter, '</i><i>')
+ '</i>').query('.')
) AS a CROSS APPLY x.nodes('i') AS y(i)
);
GO
Now, declare and initialize the variable:
declare @string nvarchar(max) = 'ddd,1.5,1,eee,2.3,0,fff,1.2,ggg,6.123,1'
Then, Create 4 common table expressions - one for all items, one for strings, one for doubles and one for integers. Note the use of the row_number() function - it will be used later to join all the results together:
;WITH AllItems as
(
SELECT Item, ROW_NUMBER() OVER(ORDER BY (select null)) as rn
FROM dbo.SplitStrings_XML(@string, ',')
)
, Strings as
(
SELECT Item as StringItem, ROW_NUMBER() OVER(ORDER BY (select null)) as rn
FROM dbo.SplitStrings_XML(@string, ',')
WHERE ISNUMERIC(Item) = 0
), Doubles as
(
SELECT Item as DoubleItem, ROW_NUMBER() OVER(ORDER BY (select null)) as rn
FROM dbo.SplitStrings_XML(@string, ',')
WHERE ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) > 0
), Integers as
(
SELECT Item as IntegerItem, ROW_NUMBER() OVER(ORDER BY (select null)) as rn
FROM dbo.SplitStrings_XML(@string, ',')
WHERE ISNUMERIC(Item) = 1 AND CHARINDEX('.', Item) = 0
)
Then, select from joining all these common table expressions. Note the use of the COALESCE built in function to only return rows where at least one value is present:
SELECT StringItem, DoubleItem, IntegerItem
FROM AllItems A
LEFT JOIN Strings S ON A.rn = S.rn
LEFT JOIN Doubles D ON A.rn = D.rn
LEFT JOIN Integers I ON A.rn = I.rn
WHERE COALESCE(StringItem, DoubleItem, IntegerItem) IS NOT NULL
Results:
StringItem DoubleItem IntegerItem
---------- ---------- -----------
ddd 1.5 1
eee 2.3 0
fff 1.2 1
ggg 6.123 NULL
