Sql table comma separated values contain any of variable values checking - sql

Question

I have a variable #a='1,2,3,4' and a table that contain a column B that contain comma separated values.
How can I check that column B values contain any of the #a variable values?

Answer 1

You need to implement a function for splitting the values. There are a lot of variations, you can use this:
CREATE FUNCTION [dbo].[fn_Analysis_ConvertCsvListToNVarCharTableWithOrder](#List nvarchar(max), #Delimiter nvarchar(10) = ',')
RETURNS #result TABLE
(
[Value] nvarchar(max),
[SortOrder] bigint NOT NULL
)
AS
BEGIN
IF #Delimiter is null
BEGIN
SET #Delimiter = ','
END
DECLARE #XML xml = N'<r><![CDATA[' + REPLACE(#List, #Delimiter, ']]></r><r><![CDATA[') + ']]></r>'
DECLARE #BufTable TABLE (Value nvarchar(max), SortOrder bigint NOT NULL IDENTITY(1, 1) PRIMARY KEY)
INSERT INTO #BufTable (Value)
SELECT Tbl.Col.value('.', 'nvarchar(max)')
FROM #xml.nodes('//r') Tbl(Col)
OPTION (OPTIMIZE FOR (#xml = NULL))
INSERT INTO #result (Value, SortOrder)
SELECT Value, SortOrder
FROM #BufTable
RETURN
END
Having such function, its pretty easy:
DECLARE #DataSource TABLE
(
[column] VARCHAR(1024)
);
DECLARE #column VARCHAR(1024) = '1,2,3,4';
INSERT INTO #DataSource ([column])
VALUES ('100,200,300')
,('100,1,500')
,('1,2,3,500')
,('200')
,('33,32,31,4,30');
SELECT DISTINCT [column]
FROM #DataSource
CROSS APPLY [dbo].[fn_Analysis_ConvertCsvListToNVarCharTableWithOrder] ([column], ',') DSV
INNER JOIN [dbo].[fn_Analysis_ConvertCsvListToNVarCharTableWithOrder] (#column, ',') FV
ON DSV.[Value] = FV.[Value];
Using CROSS APPLY we are splitting the values for each column. Then we are splitting the filtering values and performing INNER JOIN in order to match only the rows having a value contained in the filter value. After that, we need a DISTINCT because column value may contains many values from the filter.

Answer 2

A t-sql string "splitter" is what you need but I would NOT use the mTVF recommended above as it is extremely inefficient and will kill parallelism. An inline table valued function (iTVF) is what you want for splitting strings.
I would suggest using delimitedSplit8k or delimitedSplit8k_lead which will perform ~30-90 times faster; or STRING_SPLIT if you're on SQL 2016+ and only need the value which will be several hundred times faster. Note this performance test:
-- sample data
declare #rows int = 10000;
if object_id('tempdb..#strings') is not null drop table #strings;
select top (#rows)
someid = identity(int,1,1),
somestring = replace(right(left(cast(newid() as varchar(36)), 27),21),'-',',')
into #strings
from sys.all_columns a, sys.all_columns b;
-- Performance test
set nocount on;
print 'fn_Analysis_ConvertCsvListToNVarCharTableWithOrder'+char(10)+replicate('-',50);
go
declare #st datetime = getdate(), #item varchar(10);
select #item = [value]
from #strings t
cross apply dbo.fn_Analysis_ConvertCsvListToNVarCharTableWithOrder(t.somestring,',');
print datediff(ms,#st,getdate());
go 5
print 'delimitedSplit8K (serial)'+char(10)+replicate('-',50);
go
declare #st datetime = getdate(), #item varchar(10);
select #item = item
from #strings t
cross apply dbo.DelimitedSplit8K(t.somestring,',')
option (maxdop 1);
print datediff(ms,#st,getdate());
go 5
print 'delimitedSplit8K (parallel)'+char(10)+replicate('-',50);
go
declare #st datetime = getdate(), #item varchar(10);
select #item = item
from #strings t
cross apply dbo.DelimitedSplit8K(t.somestring,',')
option (recompile, querytraceon 8649);
print datediff(ms,#st,getdate());
go 5
Results
fn_Analysis_ConvertCsvListToNVarCharTableWithOrder
--------------------------------------------------
Beginning execution loop
4183
4274
4536
4294
4406
Batch execution completed 5 times.
delimitedSplit8K (serial)
--------------------------------------------------
Beginning execution loop
50
50
50
54
53
Batch execution completed 5 times.
delimitedSplit8K (parallel)
--------------------------------------------------
Beginning execution loop
133
134
133
140
136
Batch execution completed 5 times.
How you could use to solve your problem
declare #sometable table(someid int identity, someNbr tinyint);
insert #sometable values (1),(3),(6),(12),(7),(15),(19);
declare #searchstring varchar(1000) = '1,2,3,4,19';
select someid, someNbr
from #sometable t
cross apply dbo.DelimitedSplit8K(#searchstring,',') s
where t.someNbr = s.Item;
Results
someid someNbr
----------- -------
1 1
2 3
7 19