Can I write merge SQL statement having count(*)?

Question

Hi this is a followup question of my previous post. I tried to use a merge SQL statement to solve my problem but now I bump into another problem.
Now I have a table where the field I need to update is a partial PK
when using a merge SQL statement, I have to put a where clause that check if count(*) = 1 because of the PK constraint
Where can I put the count(*) = 1 clause?
Here are the details:
I have two tables TA and TB, where TA contains the fields ID, FULLNAME, TYPE and TB contains the fields ID, FIRSTNAME
I want to update the firstnames in TB to be the firstnames from TA where TB.ID = TA.ID and TA.TYPE = 'ABC'
{ID, FIRSTNAME} are PKs but for the same ID, there can be more than 1 firstname.
e.g.
TA
ID | FULLNAME | TYPE
1 Caroline T ABC
2 Mary C DEF
3 Peter J ABC
TB
ID | FIRSTNAME
1 Caroline
1 Carol
1 C,
3 Peter
I need to update TB with the new firstnames from TA where type is 'ABC' but only for those fields that have count(TB.ID) = 1
when I try to run this SQL statement
merge into TB B using TA A
on (A.ID = B.ID and A.TYPE = 'ABC')
when matched then update set B.FIRSTNAME = substr(A.FULLNAME, 1, instr(A.FULLNAME, ',') - 1)
I got this error SQL Error: ORA-00001: unique constraint (TEST.PK_TB) violated
which I believe is because I updated those fields say ID = 1, all with 'Caroline'
that means I will have to add a clause having count(TB.ID) = 1
How would you do it?
Server is Oracle 11g
Thank you!

Answer 1

Hi,
One way is to join ta and tb in the USING clause, and eliminate the duplicates there.
MERGE INTO     tb     dst
USING   (
         SELECT    ta.id
         ,           REGEXP_SUBSTR ( MIN (ta.fullname)
                                , '^[^,]*'
                        )     AS firstname
         FROM      ta
         JOIN      tb  ON  ta.id     = tb.id
         WHERE     ta.type      = 'ABC'
         GROUP BY  ta.id
         HAVING    COUNT (*)     = 1
     )                   src
ON     (scr.id      = dst.id)
WHEN MATCHED THEN UPDATE
SET     dst.firstname     = src.firstname
;If you'd care to post CREATE TABLE and INSERT statements for your sample data, then I could test this.
I used REGEXP_SUBST instead of SUBSTR and INSTR to find the firstname, because I find it a little cleaner, and becuase it returns something even when there is no ',' in fullname, which I'm guessing is what you really want. (None of the fullnames in your sample data have ','s.) You could use SUBSTR and INSTR instead.