Assuming I have this query ( pseudo) :
Select T.a, T.b, (select top 1 element from fn_split(c,',') where element=T.element) From largeTable T
Where fn_split runs for each row , I would like to use inline table valued udf so , that performance will be better.
NB : fn_split just create a table via splitting via , :
But looking at inline table valued udf structure :
create FUNCTION [dbo].[fn_...] ( ... ) RETURNS table AS RETURN SELECT ...(!!!)
It should return the select right away as the first statement !
But what if my UDF looks like :
CREATE FUNCTION [dbo].[FN_Split] ( @InDelimitedString varchar(max), @InDelimiter varchar(10) ) RETURNS @tblArray TABLE ( ElementID smallint IDENTITY(1,1), Element varchar(1000) ) AS BEGIN DECLARE @StrPos smallint, @StrStart smallint, @DelimiterLength smallint SET @DelimiterLength = LEN(@InDelimiter) WHILE LEN(@InDelimitedString) > 0 BEGIN --Removed for clarity . do some CHARINDEX manipulation ETc. END RETURN END
Question :
I can't return select right away , but still , I want to change the fn_split to inline table valued udf.
How can I do it ?
The problem is with your split function. It is doing the split in an RBAR fashion. You should use a set-based splitter. Here is the DelimitedSplit8k by Jeff Moden, which is one of the fastest splitter there is:
CREATE FUNCTION [dbo].[DelimitedSplit8K]( @pString VARCHAR(8000), @pDelimiter CHAR(1) ) RETURNS TABLE WITH SCHEMABINDING AS RETURN WITH E1(N) AS ( SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 UNION ALL SELECT 1 ) ,E2(N) AS (SELECT 1 FROM E1 a, E1 b) ,E4(N) AS (SELECT 1 FROM E2 a, E2 b) ,cteTally(N) AS( SELECT TOP (ISNULL(DATALENGTH(@pString), 0)) ROW_NUMBER() OVER (ORDER BY (SELECT NULL)) FROM E4 ) ,cteStart(N1) AS( SELECT 1 UNION ALL SELECT t.N+1 FROM cteTally t WHERE SUBSTRING(@pString, t.N, 1) = @pDelimiter ), cteLen(N1, L1) AS( SELECT s.N1, ISNULL(NULLIF(CHARINDEX(@pDelimiter, @pString, s.N1),0) - s.N1, 8000) FROM cteStart s ) SELECT ItemNumber = ROW_NUMBER() OVER(ORDER BY l.N1), Item = SUBSTRING(@pString, l.N1, l.L1) FROM cteLen l
Note: Be sure to look into the article for the updated function
For more split functions, read these articles by Sir Aaron Bertrand:
- http://sqlperformance.com/2012/07/t-sql-queries/split-strings
- http://sqlblog.com/blogs/aaron_bertrand/archive/2009/08/06/more-on-splitting-lists-custom-delimiter-preventing-duplicates-and-maintaining-order.aspx
