Helen:
 
How about a select statement that will do it for you instead of all the looping? Does this work?
 
select   asset_num
       , decode(min(num), 1, 'DHI', 2, 'AHI', 3, 'CHI', '???') as company_id
    from (select asset_num
               , decode(company_id, 'DHI', 1, 'AHI',2, 'CHI', 3, 4) as num
            from assets)
group by asset_num;
 
Hope this helps.
 
Jon Walthour
----- Original Message -----
Sent: Wednesday, August 01, 2001 5:12 PM
Subject: Your ideas will be helpful

 Hello, all:

I'm import data from schema1.table1 (ASSET_NUM,COMPANY_ID,CONSOLI_ID……) to schema2.table2 (ASSET_NUM,COMPANY_ID,CONSOLI_ID……)using SQL script on Oracle 8i. The situation is like this:

In schema1.table1, it does not has a Primary Key and Table1.ASSET_NUM has duplicated records while values in COMPANY_ID are different. For example:

ASSET_NUM    COMPANY_ID

----------   ----------

AA237334       AHI

AA237334       DHI

On the other hand, in schema2.table2, the corresponding column Table2.ASSET_NUM is defined as a NOT NULL, Primary Key. These tables' definition can't be modified.

So, I need to drop the duplicated ASSET_NUM/records from schema1.table1 and then import into schema2.table2.

The rule of drop will depend on the priorities of COMPANY_ID (ranking as < Dxx, Axx, Cxx >, from < Keep to Drop >. which means between Axx & Dxx, "Dxx" will be imported. For instances, like the above example, record contains 'AHI' will be dropped. This rule is only depended on the first Character -- substr(company_id, 1,1). )

I thought about:

1/ loop a cursor on whole table1?

2/ just duplicate records and then the rest records will be "insert into table2" directly. Then how do I decide which record to drop? Use a second cursor?

3/ screen all the duplicate record into a temp table and create a PK on that table and then decide… (this sounds very redundant)

Any ideas how can I do this?

Thanks in advance

Helen



Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/

Reply via email to