July 10, 2020 at 8:58 pm
I have this function, but I wanted to pass a table so as to use the same function to get the job done for multiple tables. For example, I want this function work for table1, and table2. But it is just for table1 currently. I was trying to use a dynamic sql in vain; it doesn't pass the parameter selected. Can someone help? Give me guide on how to pass table as a parameter.
Sample data, table1
CREATE TABLE table1 (id int identity (1,1), name varchar(60))
INSERT INTO table1
VALUES ('a1, a2, a9, a8')
Sample data, table2
CREATE TABLE table2 (id int identity (1,1), name varchar(60))
INSERT INTO table2
VALUES ('a1, a2, a9, a8')
The function:
CREATE FUNCTION f_split
(@id INT)
RETURNS @ab
TABLE (name VARCHAR(20),
ab1 VARCHAR(5)
)
AS
BEGIN
DECLARE @temp TABLE (rn INT, name VARCHAR(5))
INSERT INTO @temp(rn, name)
SELECT ROW_NUMBER() OVER(ORDER BY LTRIM(RTRIM(Split.a.value('.', 'NVARCHAR(MAX)'))) ASC) rn, LTRIM(RTRIM(Split.a.value('.', 'NVARCHAR(MAX)'))) Result
FROM
(
SELECT CAST('<X>'+REPLACE([name], ',', '</X><X>')+'</X>' AS XML) AS String
FROM table1 where id = @id
) AS A
CROSS APPLY String.nodes('/X') AS Split(a)
ORDER BY 1
INSERT INTO @ab
SELECT * FROM @temp
RETURN
END
This gives the result from table1.
SELECT * FROM F_SPLIT(1)
But I want the same function to work for table2 as well.
Any help is appreciated.
July 10, 2020 at 9:09 pm
You can't use dynamic SQL in functions. But I think that dynamic SQL is the only way that you'll achieve this, so you probably need to rethink. Can you use a stored proc instead?
The absence of evidence is not evidence of absence
- Martin Rees
The absence of consumable DDL, sample data and desired results is, however, evidence of the absence of my response
- Phil Parkin
July 11, 2020 at 2:27 pm
If I am reading your function correctly - all it is doing is splitting the delimited string and returning the values as rows. There is no reason to do this since we now have STRING_SPLIT or other string split utilities available.
And - this XML method of splitting strings is quite slow.
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”
― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs
July 11, 2020 at 3:52 pm
If I am reading your function correctly - all it is doing is splitting the delimited string and returning the values as rows. There is no reason to do this since we now have STRING_SPLIT or other string split utilities available.
Not so fast, Jeff... if you look at the code the OP wrote, it's returning an ordinal position for the split-out elements. STRING_SPLIT doesn't return such a thing and, since MS does NOT guarantee the order of the returned split-out elements, you can't rely on tricks like ROWNUMBER() to create such ordinals.
Of course, it looks like the OP is also generating the ordinals incorrectly, as well. 😀
And - this XML method of splitting strings is quite slow.
Totally agreed on that!
--Jeff Moden
Change is inevitable... Change for the better is not.
July 11, 2020 at 4:30 pm
Jeffrey Williams wrote:If I am reading your function correctly - all it is doing is splitting the delimited string and returning the values as rows. There is no reason to do this since we now have STRING_SPLIT or other string split utilities available.
Not so fast, Jeff... if you look at the code the OP wrote, it's returning an ordinal position for the split-out elements. STRING_SPLIT doesn't return such a thing and, since MS does NOT guarantee the order of the returned split-out elements, you can't rely on tricks like ROWNUMBER() to create such ordinals.
Of course, it looks like the OP is also generating the ordinals incorrectly, as well. 😀
Jeffrey Williams wrote:And - this XML method of splitting strings is quite slow.
Totally agreed on that!
Or other utilities...like DelimitedSplit8K that someone wrote a fantastic article about, if I could just remember who 🙂
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”
― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs
July 13, 2020 at 7:16 am
Using a function to parse string arrays stored in 2 or more different tables containing homogeneous data - how many more wrong approaches can you fit into such a simple task?
_____________
Code for TallyGenerator
July 13, 2020 at 1:21 pm
😀
--Jeff Moden
Change is inevitable... Change for the better is not.
July 13, 2020 at 2:28 pm
Whoever removed my post - are you sure it was really a spam?
In the very first reply Phil suggested “you probably need to rethink”. And I just pointed out that there is more than one point to rethink here.
_____________
Code for TallyGenerator
July 13, 2020 at 5:32 pm
Unspammed post. It might have been a report from a user or some glitch in the automated process. Either way, it's fixed.
January 12, 2021 at 9:02 pm
You can preserve order by creating a table with an identity column and then inserting values from string_split() into that table, like this:
Create Function [dbo].[fn_PreserveOrder]
(
@InColumn varchar (max)
,@Delimiter varchar (1)
)
Returns @Columns TABLE (RowName varchar (20), Value varchar (max))
As
Begin
Declare @OrderedTable Table (RowNum bigint Identity, Value varchar (max))
Insert into @OrderedTable select value from string_split(@InColumn, @Delimiter)
Insert Into @Columns select Concat('Value', RowNum), Value from @OrderedTable
Return
End
Test it:
select * from dbo.fn_PreserveOrder('1,9,7,8,3', ',')
Results:
RowNameValue
Value11
Value29
Value37
Value48
Value53
Hope that helps.
January 13, 2021 at 5:31 am
Again, just to say it out loud... use such methods at your own risk because MS has stated that there is to guaranteed output order from STRING_SPLIT(). I've not run into anyone where it has become a problem but, if it does, MS is simply going to repeat that the order is not guaranteed.
--Jeff Moden
Change is inevitable... Change for the better is not.
January 13, 2021 at 12:54 pm
It works until it’s below the parallel execution threshold. You never tested it on really long strings, like 10k elements, did you?
once the parallelism kicks in the order of the elements inserted into a table - permanent or temporary - won’t be preserved anymore.
_____________
Code for TallyGenerator
January 13, 2021 at 3:15 pm
I was thinking the same thing but I've not yet seen it go parallel. That's most likely because I don't use it for much of anything because it does require special handling to produce the ordered ordinals for the returned elements and it's just easier to use what I built previously, which is still very fast especially after adding in the "Lead" optimization the Eirikur Eiriksson wrote about for systems using SQL Server 2012 and up.
--Jeff Moden
Change is inevitable... Change for the better is not.
January 13, 2021 at 4:52 pm
Jeff Moden, you mention what you built previously. Did I overlook a code sample or link?
Okay, how about this? It doesn't use string_split(). I've tested it to 40,000 values several times and it took about 7 seconds to return the desired result. (When I tried to build a string of 50,000 values, I waited 20 seconds and killed it. That's building the string, not splitting it.)
Anyway,
Create Function [dbo].[fn_SplStrPrsvOrder](@str nvarchar (max), @dlm varchar (1))
Returns @Values TABLE (RowNum int identity, Value nvarchar (max))
as
Begin
Declare @i bigint = 1
Declare @out varchar(max) = ''
Set @i = 1
While @i < len(@str) + 1
Begin
Begin
Set @out = Concat(@out, Substring(@str, @i, 1))
If Substring(@str, @i, 1 ) = @dlm
Begin
Insert into @Values Select Left(@out, Len(@out) - 1)
Set @out = ''
End
If @i = Len(@str)
Begin
Insert into @Values Select Left(@out, Len(@out))
End
Set @i = @i + 1
Continue
End
Set @i = @i + 1
End
Return
End
Test it:
-- build test string
Declare @str nvarchar(max) = ''
Declare @x bigint = 1
While @x < 40001
Begin
Set @str = Concat(@str, convert(varchar (10), @x), ',')
Set @x = @x + 1
End
Set @str = Left(@str, Len(@str) - 1)
Declare @dlm varchar (1) = ','
Select Value from dbo.fn_SplStrPrsvOrder(@str, @dlm) order by RowNum
I have read a little about parallelism but I don't claim to grok it in fullness. Maybe I just got lucky in my limited number (maybe 12 or 15) of trials.
Rick
January 13, 2021 at 5:10 pm
Jeff Moden, you mention what you built previously. Did I overlook a code sample or link?
Jeff would be referring to their very well known DelimitedSplit8k function, which also lead to the DelimitedSplit8k_LEAD function.
Okay, how about this?
Hmm, that's a multi-line table-value function; they're known for their poor performance. Add a WHILE
in there, and it's going to be a very slow contender I'm afraid. Don't forget, SQL is a set based language, so using looping processes are something you want to avoid unless you have very good reason to.
Thom~
Excuse my typos and sometimes awful grammar. My fingers work faster than my brain does.
Larnu.uk
Viewing 15 posts - 1 through 15 (of 20 total)
You must be logged in to reply to this topic. Login to reply