How to SORT in order as entered in SQL Server? - sql

Question

I'm using SQL Server and I'm trying to find results but I would like to get the results in the same order as I had input the conditions.
My code:
SELECT
AccountNumber, EndDate
FROM
Accounts
WHERE
AccountNumber IN (212345, 312345, 145687, 658975, 256987, 365874, 568974, 124578, 125689) -- I would like the results to be in the same order as these numbers.

Answer 1

Here is an in-line approach
Example
Declare #List varchar(max)='212345, 312345, 145687, 658975, 256987, 365874, 568974, 124578, 125689'
Select A.AccountNumber
,A.EndDate
From Accounts A
Join (
Select RetSeq = Row_Number() over (Order By (Select null))
,RetVal = v.value('(./text())[1]', 'int')
From (values (convert(xml,'<x>' + replace(#List,',','</x><x>')+'</x>'))) x(n)
Cross Apply n.nodes('x') node(v)
) B on A.AccountNumber = B.RetVal
Order By B.RetSeq
EDIT - the subquery Returns
RetSeq RetVal
1 212345
2 312345
3 145687
4 658975
5 256987
6 365874
7 568974
8 124578
9 125689

Answer 2

You can replace IN with a JOIN, and set a field for ordering, like this:
SELECT AccountNumber , EndDate
FROM Accounts a
JOIN (
SELECT 212345 AS Number, 1 AS SeqOrder
UNION ALL
SELECT 312345 AS Number, 2 AS SeqOrder
UNION ALL
SELECT 145687 AS Number, 3 AS SeqOrder
UNION ALL
... -- and so on
) AS inlist ON inlist.Number = a.AccountNumber
ORDER BY inlist.SeqOrder

Answer 3

I will offer one more approach I just found out, but this needs v2016. Regrettfully the developers forgot to include the index into the resultset of STRING_SPLIT(), but this would work and is documented:
A solution via FROM OPENJSON():
DECLARE #str VARCHAR(100) = 'val1,val2,val3';
SELECT *
FROM OPENJSON('["' + REPLACE(#str,',','","') + '"]');
The result
key value type
0 val1 1
1 val2 1
2 val3 1
The documentation tells clearly:
When OPENJSON parses a JSON array, the function returns the indexes of the elements in the JSON text as keys.

Answer 4

This is not an answer, just some test-code to check John Cappelletti's approach.
DECLARE #tbl TABLE(ID INT IDENTITY,SomeGuid UNIQUEIDENTIFIER);
--Create more than 6 mio rows with an running number and a changing Guid
WITH tally AS (SELECT ROW_NUMBER()OVER(ORDER BY (SELECT NULL)) AS Nmbr
FROM master..spt_values v1
CROSS JOIN master..spt_values v2)
INSERT INTO #tbl
SELECT NEWID() from tally;
SELECT COUNT(*) FROM #tbl; --6.325.225 on my machine
--Create an XML with nothing more than a list of GUIDs in the order of the table's ID
DECLARE #xml XML=
(SELECT SomeGuid FRom #tbl ORDER BY ID FOR XML PATH(''),ROOT('root'),TYPE);
--Create one invalid entry
UPDATE #tbl SET SomeGuid = NEWID() WHERE ID=10000;
--Read all GUIDs out of the XML and number them
DECLARE #tbl2 TABLE(Position INT,TheGuid UNIQUEIDENTIFIER);
INSERT INTO #tbl2(Position,TheGuid)
SELECT ROW_NUMBER() OVER(ORDER BY (SELECT NULL))
,g.value(N'text()[1]',N'uniqueidentifier')
FROM #xml.nodes(N'/root/SomeGuid') AS A(g);
--then JOIN them via "Position" and check,
--if there are rows, where not the same values get into the same row.
SELECT *
FROM #tbl t
INNER JOIN #tbl2 t2 ON t2.Position=t.ID
WHERE t.SomeGuid<>t2.TheGuid;
At least in this simple case I always get exactly only the one record back which was invalidated...

Answer 5

Okay, after some re-thinking I'll offer the ultimative XML based type-safe and sort-safe splitter:
Declare #List varchar(max)='212345, 312345, 145687, 658975, 256987, 365874, 568974, 124578, 125689';
DECLARE #delimiter VARCHAR(10)=', ';
WITH Casted AS
(
SELECT (LEN(#List)-LEN(REPLACE(#List,#delimiter,'')))/LEN(REPLACE(#delimiter,' ','.')) + 1 AS ElementCount
,CAST('<x>' + REPLACE((SELECT #List AS [*] FOR XML PATH('')),#delimiter,'</x><x>')+'</x>' AS XML) AS ListXml
)
,Tally(Nmbr) As
(
SELECT TOP((SELECT ElementCount FROM Casted)) ROW_NUMBER() OVER(ORDER BY (SELECT NULL)) FROM master..spt_values v1 CROSS JOIN master..spt_values v2
)
SELECT Tally.Nmbr AS Position
,(SELECT ListXml.value('(/x[sql:column("Tally.Nmbr")])[1]','int') FROM Casted) AS Item
FROM Tally;
The trick is to create a list of running numbers with the fitting number of element (a number's table was even better) and to pick the elements according to their position.
Hint: This is rather slow...
UPDATE: even better:
WITH Casted AS
(
SELECT (LEN(#List)-LEN(REPLACE(#List,#delimiter,'')))/LEN(REPLACE(#delimiter,' ','.')) + 1 AS ElementCount
,CAST('<x>' + REPLACE((SELECT #List AS [*] FOR XML PATH('')),#delimiter,'</x><x>')+'</x>' AS XML)
.query('
for $x in /x
return <x p="{count(/x[. << $x])}">{$x/text()[1]}</x>
') AS ListXml
)
SELECT x.value('#p','int') AS Position
,x.value('text()[1]','int') AS Item
FROM Casted
CROSS APPLY Casted.ListXml.nodes('/x') AS A(x);
Elements are create as
<x p="99">TheValue</x>
Regrettfully the XQuery function position() is not available to retrieve the value. But you can use the trick to count all elements before a given node. this is scaling badly, as this count must be performed over and over. The more elements the worse it goes...
UPDATE2: With a known count of elements one might use this (much better performance)
Use XQuery to iterate a literally given list:
WITH Casted AS
(
SELECT (LEN(#List)-LEN(REPLACE(#List,#delimiter,'')))/LEN(REPLACE(#delimiter,' ','.')) + 1 AS ElementCount
,CAST('<x>' + REPLACE((SELECT #List AS [*] FOR XML PATH('')),#delimiter,'</x><x>')+'</x>' AS XML)
.query('
for $i in (1,2,3,4,5,6,7,8,9)
return <x p="{$i}">{/x[$i]/text()[1]}</x>
') AS ListXml
)
SELECT x.value('#p','int') AS Position
,x.value('text()[1]','int') AS Item
FROM Casted
CROSS APPLY Casted.ListXml.nodes('/x') AS A(x);

Answer 6

In Azure SQL, there is now extended version of STRING_SPLIT which also can return the order of items if the third optional argument enable_ordinal is set to 1.
Then this simple task is finally easy:
DECLARE #string AS varchar(200) = 'a/b/c/d/e'
DECLARE #position AS int = 3
SELECT value FROM STRING_SPLIT(#string, '/', 1) WHERE ordinal = #position
Unfortunately not available in SQL Server 2019, only in Azure for now, lets hope it will be in SQL Server 2022.