Unique ID base on other column - sql

Question

DROP TABLE #ABC
CREATE TABLE #ABC (ID INT NOT NULL, Name VARCHAR (2) NOT NULL, name2 VARCHAR(2))
INSERT INTO #ABC (ID, NAME)
VALUES (1,'01'),(1,'F5'),(1,'05'),(1,'08'),(1,'02'), (1,'03'), (1,'04'), (1,'06'),(1,'07'),(1,'09'),(1,'10'),(1,'11'),(1,'12'),(1,'13'),(1,'14'),
(1,'15'),(1,'2D'),(1,'2E'),(1,'4B'),(1,'5F'),(1,'64'),(1,'73'),(1,'83'),(1,'88'),(1,'A9'),(1,'AC'),(1,'D0'),(1,'D7'),(1,'15'),(2,'76'),(2,'J5')
Script I am using to populate name2 (not working)
UPDATE A
SET name2 = SUBSTRING(REPLACE(CONVERT(VARCHAR(36), NEWID()) , '-', ''), 0, 3)
FROM #ABC AS A
select * from #ABC
So, I have table #ABC already populated with ID and NAME, I want to populate 'Name2' so that 'name2' should not have the same value as 'name' for the same ID. Example, for ID = 1 , all name2 values should be different than name values.
Thanks

Answer 1

It looks like you are using SQL Server. If so, you can make use of the row_number() function to append a unique identifier to the end of name:
with toupdate as (
select t.*, row_number() over (partition by id order by name desc) as num
from #abc t
)
update toupdate
set name2 = cast(num as varchar(2))
This will work for up to 99 duplicates for a given id.
Unfortunately, I don't have SQL Server available right now. But here is an idea. You can take the maximum value of name and then do arithmetic, base 36 (26 alpha and 10 numeric). The result is something like this:
with toupdate as (
select t.*, row_number() over (partition by id order by name desc) as seqnum,
max(name) over (partition by id) as maxname,
((case when left(name, 1) between '0' and '9'
then ascii(left(name, 1))
else ascii(upper(left(name, 1))) - ascii('A')+10
end) * 36 +
(case when right(name, 1) between '0' and '9'
then ascii(right(name, 1))
else ascii(upper(right(name, 1))) - ascii('A')+10
end)
) as namenum
from #abc t
)
update toupdate
set name2 = (case when (namenum+seqnum)/36 < 10
then char(ascii('0')+((namenum+seqnum)/36))
else char(ascii('A')+((namenum+seqnum)/36) - 10)
end)+
(case when ((namenum+seqnum)%36) < 10
then char(ascii('0')+((namenum+seqnum)%36))
else char(ascii('A')+((namenum+seqnum)%36) - 10)
end)
This finds the maximum name and then generates names bigger than that. This assumes that you are using only alphabetic characters (upper case) and numbers. It can fail if name takes on a value near the maximum possible value ('ZZ' in this case).