Is there a SQL Server function equivalent to AutoNumber() of QlikView? - sql-server

Question

First of all: this is not a kind of a IDENTITY() field.
In QlikView, it is used to generate a number based on parameters send to function.
See its documentation here: https://help.qlik.com/en-US/qlikview/November2017/Subsystems/Client/Content/Scripting/CounterFunctions/autonumber.htm
In short, you send a parameter to it and it returns an integer that will identify the same arguments for the rest of script. If you send...
AutoNumber('Name 900') -> returns 1
AutoNumber('Name 300') -> returns 2
AutoNumber('Name 001') -> returns 3
AutoNumber('Name 900') -> returns 1 ... again
and because the parameter is already in the intern list of AutoNumber
I tried to build some like that in SQL Server, but is not possible use SELECTs inside scalar functions.
My need is to get something like...
INSERT INTO FacSales (SumaryID, InvoiceID, InvoiceDate
, ProductID, SaleValue, CustomerID, VendorID)
SELECT AutoNumber(sale.VendorID, sale.CustomerID, sale.ProductID)
, sale.InvoiceID
, sale.SaleDate
, details.ProductID
, etc, etc, etc.
Is there, inside SQL Server, a "native" function that perform this?
Or, is there a way to build this using a procedure/function?
Thanks.

Answer 1

You could use DENSE_RANK (Transact-SQL)
Returns the rank of rows within the partition of a result set, without
any gaps in the ranking. The rank of a row is one plus the number of
distinct ranks that come before the row in question.
declare #T table
(
ID int identity,
VendorID int,
CustomerID int,
ProductID int
);
insert into #T values
(1, 2, 3),
(1, 2, 3),
(1, 2, 4),
(1, 2, 3);
select sale.ID,
sale.VendorID,
sale.CustomerID,
sale.ProductID,
dense_rank() over(order by sale.VendorID,
sale.CustomerID,
sale.ProductID) as AutoNumber
from #T as sale
order by sale.ID;
Result:
ID VendorID CustomerID ProductID AutoNumber
----------- ----------- ----------- ----------- --------------------
1 1 2 3 1
2 1 2 3 1
3 1 2 4 2
4 1 2 3 1

Answer 2

You basically want a key value store. There are lots of ways to make one.
Here is a possible solution. It uses a stored procedure.
However, you did not say if the values are retained indefinitely or if they are just for a single call. This example shows how to do it indefinitely.
It could be modified to be for a single call or connection via careful use of temporary tables. If it is other than a call or connection then the autoNumber.AutoNumber table and the autoNumber.NextAutoNumber will need to be cleaned up on what ever that schedule is.
-- Create the table, sequence and sproc
-- Create a schema to hold our autonumber table and sequence
CREATE SCHEMA autoNumber
GO
-- Create a sequence. This just gives us a new number when ever we want.
-- This could be replaced with an identity column.
CREATE SEQUENCE autoNumber.NextAutoNumber AS [bigint]
START WITH 1
INCREMENT BY 1
NO CACHE
GO
-- Create a table to hold the auto number key value pairs.
CREATE TABLE autoNumber.AutoNumber(KeyValue varchar(255), Number bigint)
go
-- This is the stored procedure that actually does the work of getting the autonumber
CREATE PROCEDURE autoNumber.GetAutoNumber #KeyValue varchar(255), #AutoNumber bigint = -1 output AS
BEGIN
DECLARE #Number bigint = null
-- See if we already have an autonumber created for this keyvalue
-- If we do, then set #Number to that value
SELECT #Number = autoNum.Number
FROM autoNumber.AutoNumber autoNum
WHERE autoNum.KeyValue = #KeyValue
IF (#Number is null)
BEGIN
-- If #Number was not changed, then we did not find one
-- in the table for this #KeyValue. Make a new one
-- and insert it.
SET #Number = NEXT VALUE FOR autonumber.NextAutoNumber
INSERT INTO autoNumber.AutoNumber ( KeyValue, Number)
VALUES (#KeyValue, #Number)
END
-- Return our number to the caller.
-- This uses either an output parameter or a select.
IF (#AutoNumber = -1)
BEGIN
select #Number
END ELSE
BEGIN
set #AutoNumber = #Number
END
END
GO
-- End Create
-- Testing with "select"
EXEC autoNumber.GetAutoNumber 'Name 900'
EXEC autoNumber.GetAutoNumber 'Name 300'
EXEC autoNumber.GetAutoNumber 'Name 001'
EXEC autoNumber.GetAutoNumber 'Name 900'
-- Testing with output parameter
DECLARE #AutoNumber bigint
EXEC autoNumber.GetAutoNumber 'Name 900', #AutoNumber OUTPUT
SELECT #AutoNumber
EXEC autoNumber.GetAutoNumber 'Name 300', #AutoNumber OUTPUT
SELECT #AutoNumber
EXEC autoNumber.GetAutoNumber 'Name 001', #AutoNumber OUTPUT
SELECT #AutoNumber
EXEC autoNumber.GetAutoNumber 'Name 900', #AutoNumber OUTPUT
SELECT #AutoNumber
-- End Testing
-- Clean up
DROP PROCEDURE autoNumber.GetAutoNumber
GO
DROP TABLE autoNumber.AutoNumber
GO
drop SEQUENCE autoNumber.NextAutoNumber
DROP SCHEMA autoNumber
GO
-- End Cleanup

Answer 3

The closest SQL Server has is CHECKSUM function.
You can use it to calculate a hash value of any number of columns e.g.
SELECT CHECKSUM( 'abc', 123, 'zxc' )
UNION ALL
SELECT CHECKSUM( 'abc', 124, 'zxc' )
UNION ALL
SELECT CHECKSUM( 'abc', 123, 'zxc' )
Output:
-----------
53066784
53066832
53066784

Answer 4

I think you are looking for ROW_NUMBER().
With this sql function you can partition and order by all the field you need.
SELECT ROW_NUMBER() OVER(PARTITION BY sale.VendorID, sale.CustomerID, sale.ProductID ORDER BY sale.VendorID, sale.CustomerID, sale.ProductID)
, sale.InvoiceID
, sale.SaleDate
, details.ProductID FROM table