Oracle: Normalizing data during migration

Question

I have a table with a lot of repeated data that I'd like to refactor as 3 tables.

Current structure looks like:

meeting_desc
meeting_date
topic_desc
...

And the data in the current_table looks like:

meeting1,2/3/2009,abc
meeting1,2/3/2009,efg
meeting1,2/3/2009,xyz
meeting2,4/5/2009,aaa
meeting2,4/5/2009,bbb

I would like to create a meeting table and a topic table, with PKs coming from a sequence:

MEETING: 
  id
  meeting_desc
  meeting_date

TOPIC:
  id
  meeting_id
  topic_desc

What I can't figure out is how to insert data into the new tables. I've tried:

insert into MEETING select distinct 
  seq.nextval, meeting_desc, meeting_date from current_table

but of course that doesn't work. Is there an easy way to normalize the data?

Answer 1

Hi Chris,

The DISTINCT placed in a subquery should work:

SQL> INSERT INTO meeting
  2     SELECT seq.nextval, meeting_desc, meeting_date
  3       FROM (SELECT DISTINCT meeting_desc, meeting_date
  4               FROM current_table);

2 rows inserted

Once this is done, you would join this newly created table with the old table to associate the generated ids to the children tables:

SQL>   INSERT INTO topic
  2       SELECT m.id, topic_seq.NEXTVAL, ct.topic_desc
  3         FROM current_table ct
  4         JOIN meeting m ON (ct.meeting_desc = m.meeting_desc 
  5                            AND ct.meeting_date = m.meeting_date);

5 rows inserted