SQL Self-JOIN To Update rows

Question

I have a table in SQL Server that contains rows, some of which I need to update, depending upon the absence/presence of other rows in the same table. the table and sample data look like this:

SQL Server table:

    TABLE tblConnector(
      ID1 [int] NOT NULL,
      ID2 [int] NOT NULL,
      ID3 [int] NOT NULL,
      ...other cols...)

It has a PRIMARY KEY CLUSTERED on (ID1, ID2, ID3) - meaning that the compound key must be unique.

Sample data:

ID1  ID2  ID3
-------------
111  222  1
111  222  9999
333  444  1    <--- update only this row
555  666  1        
555  666  9999
777  888  2
777  888  9999 
123  456  3    <--- don't update this row

I need to update any row in this table, setting ID3 = 9999, where ID3 currently is 1 and there are no other rows in the table with the same values for ID1 and ID2, with ID3 = 9999.

In my sample data, I only want to update the 3rd row, setting ID3 = 9999 - because it has ID3 = 1 and there are no other rows in the table with the same ID1 and ID2 values (where ID3 <> 1). I don't want to update the last row - so using count() to isolate rows isn't a valid approach.

I want to update the ID3 column to 9999 ONLY if the current value in ID3 is 1 ... AND there are no other rows that have the same ID1 and ID2 values where ID3 = 9999. ID1, ID2 and ID3 comprise a unique key and setting 9999 in ID3 can't duplicate a key value already in existence.

I've tried numerous attempts to join the table to itself, but I can't seem to get an UPDATE statement that will only affect the rows I want.

Any advice on how to code this SQL?

Answer 1

Here's one option using a common table expression and count() over partition:

with cte as (
  select *, count(*) over (partition by id1, id2) rn
  from tblConnector
  )
update cte 
set id3 = 9999
where id3 = 1
  and rn = 1;

• SQL Fiddle Demo